State of the art review Chest electrical impedance ...

Chest electrical impedance tomography examination,data analysis, terminology, clinical use andrecommendations: consensus statement of theTRanslational EIT developmeNt stuDy groupInéz Frerichs,1 Marcelo B P Amato,2 Anton H van Kaam,3 David G Tingay,4

Zhanqi Zhao,5 Bartłomiej Grychtol,6 Marc Bodenstein,7 Hervé Gagnon,8

Stephan H Böhm,9 Eckhard Teschner,10 Ola Stenqvist,11 Tommaso Mauri,12

Vinicius Torsani,2 Luigi Camporota,13 Andreas Schibler,14 Gerhard K Wolf,15

Diederik Gommers,16 Steffen Leonhardt,17 Andy Adler,8 TREND study group

Additional material ispublished online only. To viewplease visit the journal online(http://dx.doi.org/10.1136/thoraxjnl-2016-208357).

For numbered affiliations seeend of article.

Correspondence toProfessor Dr Inéz Frerichs,Department of Anesthesiologyand Intensive Care Medicine,University Medical CenterSchleswig-Holstein, CampusKiel, Arnold-Heller-Str. 3,Kiel 24105, Germany;[email protected]

Received 19 January 2016Revised 12 July 2016Accepted 16 July 2016Published Online First5 September 2016

To cite: Frerichs I,Amato MBP, van Kaam AH,et al. Thorax 2017;72:83–93.

ABSTRACTElectrical impedance tomography (EIT) has undergone30 years of development. Functional chest examinationswith this technology are considered clinically relevant,especially for monitoring regional lung ventilation inmechanically ventilated patients and for regionalpulmonary function testing in patients with chronic lungdiseases. As EIT becomes an established medicaltechnology, it requires consensus examination,nomenclature, data analysis and interpretation schemes.Such consensus is needed to compare, understand andreproduce study findings from and among differentresearch groups, to enable large clinical trials and,ultimately, routine clinical use. Recommendations of howEIT findings can be applied to generate diagnoses andimpact clinical decision-making and therapy planning arerequired. This consensus paper was prepared by aninternational working group, collaborating on the clinicalpromotion of EIT called TRanslational EIT developmeNtstuDy group. It addresses the stated needs by providing(1) a new classification of core processes involved inchest EIT examinations and data analysis, (2) focus onclinical applications with structured reviews and outlooks(separately for adult and neonatal/paediatric patients),(3) a structured framework to categorise and understandthe relationships among analysis approaches and theirclinical roles, (4) consensus, unified terminology withclinical user-friendly definitions and explanations, (5) areview of all major work in thoracic EIT and (6)recommendations for future development (193 pages ofonline supplements systematically linked with the chiefsections of the main document). We expect thisinformation to be useful for clinicians and researchersworking with EIT, as well as for industry producers ofthis technology.

INTRODUCTIONElectrical impedance tomography (EIT) is aradiation-free functional imaging modality inventedover 30 years ago.1 Scientific and clinical interest inthis method is driven by the clinical need for moni-toring of lung ventilation and perfusion and forassessment of regional lung function at the bedside.The growing interest is testified by the continuouslyincreasing number of publications on EIT and by

the commercial availability of devices from severalcompanies.This paper was prepared by a working group,

collaborating on the clinical promotion of EITcalled TRanslational EIT developmeNt stuDy(TREND) group. The group was formed at themeeting ‘Chest EIT: Status, Vision and Priorities’(Manchester, UK, 16–17 November 2012), atwhich leaders of all clinically oriented EIT researchgroups from North and South America, Europe,Asia and Australia and a representative from eachcompany producing EIT technology were repre-sented. The TREND group consists of pre-eminentresearchers and clinical leaders spanning neonatal,paediatric and adult fields of chest EIT. Meetings ofthis group take place at the annual EIT conferencesand congresses of the American Thoracic Societyand the European Respiratory Society.One key concern of the TREND group that

resulted in the preparation of this paper was thelack of recommendations for chest EIT examina-tions, consistent terminology and generally acceptedapproaches to EIT image analysis and interpretation.The structure of this paper differs from former,rather descriptive reviews.2–11 It is based on fivecore processes that we identified during EITexamin-ation and data analysis (figure 1). These processesinvolve: (1) execution of EIT measurements, (2)generation of raw EIT images, (3) EIT waveformsand regions-of-interest (ROI), (4) functional EITimages and (5) EIT measures. They are con-secutively addressed in the first five sections of thepaper. Each section is accompanied by an electroniconline supplement (EOS) with detailed figures,definitions, recommendations and examples. Aseparate online supplement with the definitions ofEIT-related terms is provided.The last part of the article focuses on the clinical

use of EIT in adult, neonatal and paediatricpatients. The major target populations are mechan-ically ventilated patients, typically requiring inten-sive care therapy, and pulmonology patientssuffering from chronic lung diseases. Two supple-ments linked to these clinical sections provide sum-maries and lists of clinical issues benefiting fromEIT-guided clinical decisions. Finally, we propose

Frerichs I, et al. Thorax 2017;72:83–93. doi:10.1136/thoraxjnl-2016-208357 83

State of the art review on A

pril 11, 2022 by guest. Protected by copyright.

http://thorax.bmj.com

/T

horax: first published as 10.1136/thoraxjnl-2016-208357 on 5 Septem

ber 2016. Dow

nloaded from

on April 11, 2022 by guest. P

rotected by copyright.http://thorax.bm

j.com/

Thorax: first published as 10.1136/thoraxjnl-2016-208357 on 5 S

eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from



j.com/


eptember 2016. D

ownloaded from

on A



/T


ber 2016. Dow

nloaded from

http://dx.doi.org/10.1136/thoraxjnl-2016-208357



http://crossmark.crossref.org/dialog/?doi=10.1136/thoraxjnl-2016-208357&domain=pdf&date_stamp=2016-09-05

http://thorax.bmj.com/

https://www.brit-thoracic.org.uk/












































































































































































































recommendations for the future research and development ofEIT presented in detail in the last EOS. All nine online supple-ments contain comprehensive lists of references.

Our paper is the result of a joint effort of EIT expertsbased on extensive literature search and, their expertise andexperience with this medical technology. It is neither a healthtechnology assessment nor a conventional systematic review.Nonetheless, we believe that it will become a referencedocument that reviews main achievements and outlines recom-mendations for clinical use, describes how errors and misinter-pretations in EIT data acquisition and analysis can be avoidedand facilitates sharing and comparison of research and clinicalresults.

We expect our consensus document to be relevant to currentclinical users of EIT (adult and neonatal/paediatric intensivists),potential new users (adult and paediatric pneumologists) andproducers of EIT technology.

EXECUTION OF EIT CHEST MEASUREMENTSEIT examinations require the placement of electrodes on thechest circumference, positioned to be sensitive to the phenom-ena of interest. Electrodes are either placed individually withequal spacing or are integrated into electrode belts or stripes,which render the application user-friendly. Several EIT devicesuse 16 electrodes, although systems with a higher and lowernumber of electrodes have also been developed. The electrodesare typically placed in one transverse plane, although obliqueplacement has also been described. Large chest wounds, mul-tiple chest tubes, non-conductive bandages or conductive wiresutures may preclude or affect the measurements.

The location of the electrode plane impacts the findings;12–15

thus, comparability of examinations performed on separate occa-sions requires electrode locations to be the same. It is not

recommended to place the electrodes lower than the sixth inter-costal space because the diaphragm may periodically enter themeasurement plane.13 Posture14 16–19 and type of ventilation20–25

affect the findings. Specific ventilation manoeuvres, mechanicalventilation mode and settings, etc should be recorded to ease theinterpretation of EIT findings.

During an EIT examination, very small alternating electricalcurrents are applied through pairs of electrodes while the result-ing voltages are measured on the remaining electrodes. Themost widespread spatial pattern of current applications andvoltage measurements is through adjacent electrode pairs. Otherpatterns of current applications and voltage measurements areincreasingly used and can be expected to replace the adjacentpattern because they offer technical advantages. Further infor-mation on the execution of EIT measurements is provided inEOS 1.

RAW EIT IMAGESThe set of EIT data acquired during one cycle of current appli-cations and voltage measurements is typically called frame. OneEIT data frame contains the information necessary to generateone raw image. The number of frames (or raw images) acquiredper second corresponds with the EIT scan rate. Current EITdevices offer maximum scan rates of about 50 images/s allowingthe assessment of lung function under dynamic conditions.

EIT is sensitive to periodic and non-periodic changes in elec-trical tissue conductivity in a slice with a vertical thicknessroughly half the chest width.26 An increase in intrapulmonarygas volume decreases conductivity, while increased blood orfluid volume or disruption of cellular barriers raise it.

Image reconstruction is the process of generating raw EITimages from the measured voltages, typically of a two-dimensional slice through the electrode plane.27 Time-difference

Figure 1 Schematic presentation ofthe chest EIT examination and dataanalysis. The drawings, examples ofEIT images and EIT measures illustratethe different steps involved. Theimages were generated using theGREIT image reconstruction algorithmfrom data acquired in a healthy adultsubject with the Goe-MF II EIT device(CareFusion, Höchberg, Germany).(These images as well as the datashown in subsequent figures originatefrom examinations approved by theinstitutional ethics committees andacquired with written informedconsent.) ARDS, acute respiratorydistress syndrome; EIT, electricalimpedance tomography; rel. ΔZ,relative impedance change; GI, globalinhomogeneity index; CoV, centre ofventilation; PEEP, positive end-expiratory pressure; ROP, regionalopening pressure; RVD, regionalventilation delay.

84 Frerichs I, et al. Thorax 2017;72:83–93. doi:10.1136/thoraxjnl-2016-208357




/T


ber 2016. Dow

nloaded from



EIT image reconstruction calculates images of the change intissue properties between a baseline (reference) measurementframe and the current frame. Time-difference imaging is wellsuited to trace time-varying physiological phenomena like lungventilation and perfusion. Two other imaging modalities areactive areas of research, but currently insufficiently robust forchest EIT: (1) frequency-difference imaging, sensitive to the dif-ference in tissue properties between two stimulation frequenciesat a given time and (2) absolute imaging, which calculatesimages of the properties at a given time (ie, not just relativeimpedance changes).

An ideal reconstruction algorithm should guarantee uniformamplitude response, small and uniform position error, smallringing artefacts, uniform resolution, limited shape deformation,high resolution and exhibit small sensitivity to electrode andboundary movement.28 The images are either round or, innewer algorithms, their contour reflects the anatomical form ofthe chest. Development of new algorithms is an active field inEIT research with advances in image quality achieved by using,for example, a priori anatomical information.

The orientation of EIT images is identical to the images gen-erated by established imaging modalities like CT with the rightchest side on the left side of the image and with anterior at thetop. The colour coding of EIT images depends on the choice ofthe baseline frame and is not unified (for examples, see EOS 2).

EIT WAVEFORMS AND ROIEIT data analysis is based on the EITwaveforms that are gener-ated from a series of raw EIT images in individual image pixels.ROI can be defined in an image (minimum size being one image

pixel). In each ROI, the waveform over time displays the short-term and long-term changes in local electrical impedance result-ing from various physiological or pathological effects. Periodicsignal fluctuations are induced by ventilation or by heart actionand lung perfusion (figure 2). Digital frequency filtering is oftenused in EIT data analysis to isolate these periodic phenomena. Anon-periodic change may be caused, for example, by an overallincrease in gas volume induced by raising the positiveend-expiratory pressure (PEEP) during mechanical ventila-tion.29–31

The magnitude of the impedance changes associated withspontaneous breathing or mechanical ventilation is approxi-mately one order of magnitude larger than the changes inducedby heart action and lung perfusion.32 Besides physiological andpathological effects, EIT waveforms may also display artefactsoriginating from, for example, body movement33 or interferencewith some medical devices.34 Modern EIT devices have some-what improved robustness to interference. The artefacts canoften be eliminated by adapting the EIT data acquisition para-meters or by signal postprocessing.

EIT images show not only the lungs but the whole chestcross-section. To increase the sensitivity of EIT data analysis,ideally only waveforms originating from lung regions should beanalysed. The border between the pulmonary and non-pulmonary tissue is blurred in EIT images. Therefore, severalapproaches have been proposed to determine ROIs representingthe lungs.35 36

EIT lung ROIs are mostly functionally identified as regionswhere ventilation-related impedance changes occur. If thisapproach is applied, lung regions with absent or small

Figure 2 Electrical impedance tomography (EIT) waveforms simultaneously registered in one image pixel in the dorsal region of the dependent(left) and in another pixel in the dorsal region of the non-dependent lungs (right) in a healthy man aged 43 years lying on his right side. The rawdata were obtained using the Goe-MF II EIT device (CareFusion, Höchberg, Germany) and reconstructed with the GREIT algorithm. During the first30 s, the subject was breathing quietly, then he was instructed to hold his breath for 20 s after tidal inspiration, finally he took three deep breaths.The periodic ventilation-related changes of the EIT signal, given as relative impedance change (rel. ΔZ), are higher than those associated with thecardiac action. (The latter are best discernible during the apnoea phase.) The tidal changes in rel. ΔZ during the first and the third phases of thismeasurement are higher in the dependent than in the non-dependent lung reflecting the physiologically known higher ventilation of the dependentlung regions in adult subjects spontaneously breathing at the functional residual capacity level. During the apnoeic phase, the EIT signal fallscontinuously in the dependent lung due to local gas volume loss caused by the continuing gas exchange. This is not observed in the non-dependentlung. The changes in rel. ΔZ synchronous with the heartbeat are of comparable amplitude in both pixels. The breathing rate (BR) and heart rate (HR)in each of the three examination phases given in the lower part of the figure were derived from frequency filtering of the EIT signal.





/T


ber 2016. Dow

nloaded from



ventilation-related impedance changes (eg, atelectasis, pleuraleffusion or pneumothorax) may be missed. If unaccounted for,this may impact subsequent analyses, for example, by showingless ventilation heterogeneity, than present.

The other motivation for ROI-based analysis of EIT data is tocharacterise the spatial heterogeneity of lung ventilation. Forthis application, arbitrary ROIs, like image quadrants or layersare used.24 37–39 These are either applied to the whole image orare combined with a previously defined lung ROI. Furtherdetails can be found in EOS 3.

FUNCTIONAL EIT IMAGESFunctional EIT images are generated from series of raw imagesand the corresponding pixel EITwaveforms, by use of appropri-ate mathematical equations in each image pixel. All calculatedvalues are then plotted (or image colour-coded) in the respectivepixels. Various functional EIT measures have been proposed andused to quantify and characterise regional lung ventilation (andperfusion). These are briefly addressed here but described indetail in EOS 4.

Today’s high EIT scan rates allow precise identification of thetidal peak-to-trough changes in the pixel EIT waveforms andgeneration of functional EIT images proportional to the localtidal volume (VT).

25 37 40 41 Identification of pixelend-expiratory minima before and after a change in ventilatorsettings (eg, in PEEP) creates a functional EIT image of the localchanges in end-expiratory lung volume (EELV).29 If EIT data areacquired during the forced full expiration then the calculation ofthe electrical impedance change within the first second of theforced exhalation will image the distribution of local FEV1.

42

The analysis of the dynamic characteristics of pixel EITwave-forms enables the assessment of the non-linearity of local fillingand emptying of the lungs and creation of functional EITimages showing the spatial distribution of this non-linear behav-iour.43 If EIT examination is performed during a stepwise infla-tion or deflation of the lungs then fitting of mathematicalfunctions to pixel EIT waveforms may be used to calculate andimage local respiratory time constants.44

Combination of pixel EIT waveforms with other simultan-eously registered signals like the airway pressure allows the gen-eration of pixel pressure-volume curves45 46 and generation offunctional images showing, for example, regional lung openingand closing47 or regional respiratory system compliance (Crs).

48

Several types of functional images can be generated from asingle EIT measurement. They address different aspects ofregional lung function, thus, the combination of these findingsallows a more thorough interpretation of EIT data.

EIT MEASURESFunctional EIT images display the calculated functional measureswith the highest spatial resolution available. Based on theseimages, it is desirable to create quantifiable clinically relevantmeasures to assess the present state of ventilation distributionand its trends. Images and numeric values are complementary.

Quantitative EIT measures can be divided into three groups.The first group has the same functional measures as the onesused to generate functional EIT images, only the pixel values ofthese measures are averaged (or summed) across the wholeimage or its sections. For instance, the sum of tidal impedancechanges in all image pixels provides an estimate of regional VT

in the studied chest slice.The second group aims at characterising the spatial distribu-

tion of ventilation. One subgroup describes the overall degree ofspatial heterogeneity of ventilation. Examples are the global

inhomogeneity index36 40 49 or the coefficient of vari-ation,36 42 50 calculated from the maps of tidal impedance vari-ation. Another subgroup describes the orientation of the spatialdistribution of functional measures, usually in the anteroposter-ior direction. This direction is clinically relevant when patientsare examined in the supine position, where it corresponds tothe gravity vector. The simplest measure is the anteroposterior(upper-to-lower) ventilation ratio, calculated as the ratiobetween the sum of tidal impedance changes in the anterior andposterior halves of the functional image or the lung ROI withinit.51 Alternatively, the anterior and posterior fractions (percen-tages) of ventilation are calculated.24 Fractions of ventilation canalso be computed in larger number of smaller ROIs spanningthe whole image in the anteroposterior direction and used tocreate ventilation profiles.37 52 53 The centre of ventilation,frequently used to describe the ventilation distribution withrelation to the chest diameter, is derived from suchprofiles.20 37 53 54

The third group includes examination-specific measures.Examples are measures of perfusion via an infusion ofconductivity-contrasting bolus,55 or measures that are derivedfrom the EIT data registered in parallel with other signals,mostly airway pressure. Such measures aim at characterisingregional Crs under quasistatic45 56 but mostly dynamic condi-tions.57 58 The high scan rates allow the assessment of intratidalchanges in regional Crs.

38 Another EIT measure of respiratorymechanics is the regional respiratory time constant.44 59

Temporal heterogeneity of ventilation can be characterised bycalculating the phase shifts in regional ventilation60 or ventila-tion delay index.61 Regional expiration times needed to exhalespecified percentages of regional gas volumes can also bederived from EIT waveforms.42 62 63 Further details on EITmeasures are presented in EOS 5.

The taxonomy of EIT terms with clinical user-friendly defini-tions and explanations is provided in EOS 6.

CLINICAL USE OF EIT IN ADULT PATIENTSThree general uses of thoracic EIT have been promising in adultpatients: (a) monitoring of mechanical ventilation, (b) monitor-ing of heart activity and lung perfusion and (c) pulmonary func-tion testing. The validity and reproducibility of EIT findings isderived from several experimental and clinical studies compar-ing EITwith reference techniques like CT,61 64–66 single-photonemission CT,67 positron emission tomography,68 vibrationresponse imaging,69 inert-gas washout70 and spirometry.14 30 71

EIT examinations over time often provide unique clinicalinformation difficult to obtain by other technologies at thebedside. With almost no side effects, EIT allows the sensitiveand prompt assessment of lung characteristics over the course ofdisease and treatment.

Monitoring of mechanical ventilationThe need for monitoring regional effects of mechanical ventila-tion is based on our understanding of the direct damageinflicted by the ventilator on the fragile lung. Protective ventila-tion has been increasingly demanded inside and outside inten-sive care units to prevent/minimise ventilator-associated lunginjury. Global measures of oxygenation or respiratory systemmechanics, traditionally used as references to adjust mechanicalventilation, may produce misleading information by ‘averaging’opposite pathological phenomena (eg, tidal recruitment andoverdistension) in different lung units. This underlines the needfor regional functional lung monitoring.





/T


ber 2016. Dow

nloaded from






EIT can generate regional measures characterising this hetero-geneous behaviour of the lung tissue under dynamic conditionsthat might be used in guidance of mechanical ventilation.Experimental studies showed that regional responses of the lungto a ‘recruiting manoeuvre’,37 72 PEEP,29 41 57 fraction ofoxygen54 or VT adjustment52 can be continuously obtained byEIT. The more recent clinical data suggest the promising use offast-response EIT parameters to titrate protective PEEP/VT com-binations, or to assess the potential for lung recruitment inpatients with acute respiratory distress syndrome (ARDS). Thesemeasures have most often been derived from the decrementalPEEP-titration manoeuvre.73–75 Figure 3 shows an examplepatient examination during this manoeuvre. By sequentiallymeasuring the EIT-derived regional Crs, it is possible to quantifythe amounts of tissue that (a) recollapse during the trial, exclud-ing them from ventilation and (b) are brought back to adequateventilation, previously impaired by hyperdistension.

The side effects of mechanical ventilation are aggravated byheterogeneities in lung properties or in pressure transmissioncaused by disease, ageing or gravity forces. Major heterogene-ities in regional ventilation in diseases like ARDS,44 66

COPD,42 62 lung cancer76 or cystic fibrosis77 can be traced byEIT. The effects of ageing can also be detected.16 42 The cap-ability of EIT to determine gravity-dependent changes has beenconfirmed in studies where human subjects were exposed toweightlessness43 or posture changes.14 16 Posture-dependentchanges in ventilation can be traced by EIT in a clinical setting78

as well and used to monitor the patient’s response to rotation.These and other clinical studies, although often with relativelylow numbers of patients, corroborate the perspective of real-time monitoring of lung function heterogeneities by EIT at thebedside, potentially helping to optimise lung protection.

A few clinical studies imply the usefulness of EIT monitoringduring assisted mechanical ventilation where spontaneousbreathing activity is combined with gas delivery by the ventila-tor. EIT can determine regional Crs during this ventilation31 anduncover potentially harmful excessive regional VT, even in thecontext of a ‘protective’ VT delivered to the whole lung.22 Thedetection of this ‘pendelluft’ phenomenon by EIT is shown infigure 4.

A relevant feature of EIT is its ability to identify variousadverse events during mechanical ventilation. This would enableearly therapeutic intervention and/or reduce the need for otherdiagnostic methods. Experimental studies confirmed the detec-tion of pneumothorax39 79 or derecruitment, for example, aftersuctioning.61 80 Clinical data are rare: there exist studies inpatients with pleural effusion, which only confirm the ability todetect impedance changes related to fluid removal,81 but notprospectively identify pleural effusion development. The abilityof EIT to detect one-sided ventilation with selective intubationhas been validated in patients.82

Our perspective is that EITwill become a standard monitoringtechnique for personalised protective mechanical ventilation andoptimised ventilator adjustments, increasing patient safety

Figure 3 An adult mechanically ventilated patient examined by electrical impedance tomography (EIT) (Enlight, Timpel, Sao Paulo, Brazil; with afinite element method (FEM)-based Newton Raphson image reconstruction algorithm) during a decremental positive end-expiratory pressure (PEEP)trial. The patient suffered from severe acute respiratory distress syndrome as detected in the CT scan (top left) obtained at the same chest planewhere the EIT electrode belt was located during the EIT examination. The panels 1–10 show regional lung hyperdistension and collapse at eachPEEP step in white and blue colours, respectively. The percentages of collapse and overdistension are provided at the right side of each panel and inthe diagram (bottom left). With decreasing PEEP, hyperdistension fell (green curve) and collapse rose (blue curve). The crossover point between thecurves is highlighted by the red arrow. The corresponding PEEP step preceding the crossover shows the value of 17 cm H2O in red in panel 4.





/T


ber 2016. Dow

nloaded from


during mechanical ventilation and providing continuous surveil-lance. Automatic interpretation of large sequences of EITimages will be the key to improve its daily use.

Monitoring of heart activity and lung perfusionThe major focus of EIT in this field is the real-time measure-ment of regional lung perfusion, although smaller studiesaiming at cardiac-output measurement, heart-lung interactionsand PE also exist. Lung perfusion has mostly been assessed by theamplitude of regional EITsignal pulsatility. This heartbeat-relatedEIT signal variation reflects the periodic changes in blood vesseldimension. Using this approach, regional hypoxic pulmonaryvasoconstriction83 84 and prostacyclin-induced vasodilation85

have been identified in patients and hyperoxic vasodilation inhealthy volunteers.86 Another method, examined in mechanic-ally ventilated patients, is based on the measurement of regionaloxygen uptake by EIT.87 The only approach specifically tracking

pulmonary blood flow uses the first-pass kinetics of intravenouslyadministered hypertonic saline bolus (figure 5) and was describedand validated in experimental studies.55 88

Monitoring of lung perfusion might be important in thealready addressed use of EIT for guidance of ventilator therapybecause it would allow ventilation/perfusion distribution mis-match to be taken into account. However, clinical validation inthis field is lacking.

Pulmonary function testingAn increased demand is observed for EIT imaging in ambulatorypatients, as an adjunct to conventional pulmonary functiontests. In patients with COPD,42 62 asthma63 and cystic fibrosis,77

EIT generates cross-sectional maps of the pulmonary functionmeasures known from conventional spirometry, enabling theassessment of both spatial and temporal heterogeneity ofregional lung function over time or after intervention (eg,

Figure 4 Ventilation distribution in a supine adult patient in the course of one tidal inflation during controlled and assisted mechanical ventilationin the pressure-controlled mode. The data were acquired with the Enlight device (Timpel, Sao Paulo, Brazil) using a finite element method(FEM)-based Newton Raphson reconstruction algorithm. The electrical impedance tomography (EIT) images show the impedance variation comparedwith the beginning of the respiratory cycle at five time points highlighted by dashed lines (A–E). In controlled ventilation, anterior and posteriorregions inflated synchronously, as seen in the blue and red waveforms. In contrast, when the same patient was allowed to perform spontaneousefforts, the pendelluft phenomenon was detected by EIT with a volume shift from the anterior to the posterior regions attributed to differences inlocal driving forces and lung mechanics. The stronger posterior diaphragm excursion ‘sucked’ air from the anterior regions, which deflated at thebeginning of inspiration, causing a transient local overdistension in the posterior regions. This excessive volume shift can be seen in the bluewaveform as an overshoot (simultaneously with an undershoot in the red anterior lung waveform) before decreasing to the mechanical equilibriumwith the ventilator, followed by relatively synchronous expiration. Red and blue regions in the EIT images imply a decrease and an increase inregional impedance, respectively, when compared with the beginning of inspiration.





/T


ber 2016. Dow

nloaded from


bronchodilation) during spontaneous tidal breathing and forcedventilation manoeuvres (figure 6). Regional pulmonary functiontesting by EIT might improve the patient phenotyping, the mon-itoring of natural history of disease and therapy effects and theprediction of clinical outcomes. Clearly, more clinical trials areneeded in this field, especially those studying the relationshipbetween the disease severity and EIT findings.

Besides the described three major fields of EIT use inadult patients, there are additional clinical situations in whichEIT information has been helpful for clinical decisions (refer toEOS 7 for further information and the extensive references).

CLINICAL USE OF EIT IN NEONATAL AND PAEDIATRICPATIENTSBy virtue of being non-invasive and radiation-free, as well asallowing respiratory monitoring independent of an endotrachealtube, EIT offers particular promise in the vulnerable populationof neonatal and paediatric patients. To date, the focus of moststudies has been on monitoring regional changes in lung aer-ation and, to a lesser extent, lung perfusion, with the aim of

improving knowledge on lung physiology and better under-standing the interaction between the impact of lung disease andclinical interventions on lung function.

Safe and effective guidance of mechanical ventilation in neo-nates, infants and children requires considering the regionalvolumes of the diseased lung. Consequently, most EIT studiesinvestigating lung aeration have focused on regional changes inEELV and/or (tidal) ventilation. Studies in preterm infants withrespiratory distress syndrome (RDS) receiving high-frequencyventilation showed that stepwise lung recruitment resulted in arelatively homogenous increase in EELV, with patterns similar toglobal volume measures.89 EIT can map the regional pressure-volume relationships of the respiratory system, measuring theextent of lung hysteresis and allowing identification of anoptimal applied airway pressure (figure 7). EIT has helped showthat exogenous surfactant therapy in preterm infants results inan increase and stabilisation of regional EELV (figure 7)46 andalters the respiratory time constants.59 In conventionally venti-lated children with acute lung injury, EIT could monitor regionalEELV changes during stepwise recruitment.90 In this population,

Figure 5 An adult patient with a mediastinal mass admitted to the intensive care unit with severe respiratory failure and difficult mechanicalventilation (peak inspiratory pressures >40 cmH2O). The CT image suggested severe airway obstruction caused by tumour invasion, with probableobstruction of right pulmonary artery (top left). Black arrows in the top left panel show the severe obstruction of the main bronchi, morepronounced on the left. This supported the surgical plan of right-sided pneumonectomy along with tumour excision. As part of the detailed patientevaluation, an electrical impedance tomography (EIT) examination (Enlight, Timpel, Sao Paulo, Brazil; with a finite element method (FEM)-basedNewton Raphson algorithm) was performed to estimate regional ventilation and perfusion. Regional EIT waveforms showed smaller ventilationamplitude in the left lung (red) in comparison to the right lung (blue). During a prolonged expiratory pause, intrinsic positive end-expiratory pressure(PEEPi) was detected in the proximal airway pressure (Pprox) signal. At the same time, an evident redistribution of air occurred: the left lungexhibited a decrease and the right lung an increase in air content. When ventilation returned, the opposite occurred, with cumulative air trapping inthe left and concomitant deflation of the right lungs. A CT image, colour-coded to enhance the low lung density, demonstrated the marked airtrapping in the left lung (bottom left). The functional EIT images showed the right lung to exhibit higher ventilation (bottom middle) and perfusion(bottom right) than the left lung. These findings resulted in changed surgical plans and preservation of the right lung. After the removal of thecompressing mass (rhabdomyosarcoma), which was ultimately not invading the right pulmonary artery, the patient was successfully extubated2 days later.





/T


ber 2016. Dow

nloaded from



there was clear heterogeneity in the regional volume response,with the largest increase in EELV in the dependent lung regions.Significant breath-to-breath variability in regional tidal ventila-tion, unrecognised using current bedside tools, was reported ina population of preterm infants receiving volume-targeted con-ventional ventilation, suggesting another way EIT could guide

ventilator settings at the bedside.91 EIT has also been used tomonitor EELV during clinical interventions, such as induction ofanaesthesia, intubation and endotracheal suction.21 92

EIT studies on the effect of body positioning on regional ven-tilation in infants and children have challenged the physiologicalparadigm that ventilation is preferentially distributed towards

Figure 6 Electrical impedancetomography (EIT) examination of apatient aged 67 years with COPDperformed during forced full expiration(Goe-MF II EIT device (CareFusion,Höchberg, Germany) with the GREITimage reconstruction). The regional EITwaveforms (bottom left) originate fromfour image pixels highlighted in thefunctional EIT ventilation image of thispatient (top). The waveforms werenormalised to better visualise theregional dissimilarities in lungemptying. The histogram (bottomright) shows the heterogeneity ofEIT-derived pixel ratios of FEV1 andFVC. rel. ΔZ, relative impedancechange.

Figure 7 End-expiratory lung volume changes measured by electrical impedance tomography (EIT) (Goe-MF II EIT device (CareFusion, Höchberg,Germany) with the GREIT image reconstruction) in a preterm infant (weight 800 g) with respiratory distress syndrome subjected to anoxygenation-guided lung recruitment procedure during high-frequency ventilation before and after exogenous surfactant administration. Using theimpedance changes at a continuous distending pressure of 8 cmH2O as the reference value, both the inflation (solid red line) and the deflation limbs(solid blue line) before and the deflation limb (dashed blue line) after surfactant treatment (grey arrow) are mapped (A). In addition, the regionalimpedance changes in the cross-sectional slice of the chest during each incremental and decremental pressure step are presented as functional EITimages (B). Red colour indicates a large variation and blue a small variation in impedance. Note the presence of lung hysteresis before surfactanttreatment (A) and the dissimilar regional changes in lung aeration (B). Also note the increase in lung aeration and its distribution after surfactanttreatment and the stabilising effect on the deflation limb. AU, arbitrary unit; CDP, continuous distending pressure; R, right; L, left; V, ventral; D,dorsal.





/T


ber 2016. Dow

nloaded from


the non-dependent lung. They have shown highly variable venti-lation distribution.17–19 93 A positional effect can favour boththe dependent and non-dependent lung regions, probablydepending on factors such as the presence of underlying lungdisease and/or respiratory support, initial posture, use of sed-ation, breathing pattern, transpulmonary pressure and age.

Asymmetrical distribution of ventilation measured by EIT canassist the clinician in diagnosing unilateral lung disease such aspneumothorax, atelectasis and malpositioning of the endo-tracheal tube.94–96 In the short term, the use for these purposesmay be the easiest method of broadly integrating EIT into theclinical environment. The results of a few studies in childrenwith asthma and cystic fibrosis showed the potential ofEIT-based regional pulmonary function testing to detect andmonitor lung function heterogeneity. This is a new prospect ofEIT use in the diagnostics of paediatric lung diseases that needsto be further evaluated.

To date, studies in infants and children on EIT-derived lungperfusion have been limited. A study in preterm infants showedthat measuring changes in pulmonary blood volume by EITwasfeasible and that the changes were greater in the non-dependentlung.97 EIT has been used to demonstrate a decrease in globaland regional pulmonary blood flow following ventricularseptum defect repair in children.98

Validation studies in infants and children are limited, becausecomparative monitoring with CT and radionuclide scanning isoften not feasible or safe. To date, functional EIT images havemainly been compared with chest radiography, showing asimilar loss in aeration.23 94 95 One study in preterm infantswith RDS showed that the changes measured in the cross-sectional slice by EIT are representative of changes in the wholelung measured by respiratory inductance plethysmography.99

Further details on the use of EIT in the neonatal and paediat-ric populations with tables summarising the clinical issues moni-tored by EIT can be found in EOS 8.

FUTURE DIRECTION IN EIT DEVELOPMENT, EVALUATIONAND USEThe priority of EIT research is to establish this method in clinicalpractice.100 This process requires the identification of relevantunmet clinical needs and remaining deficits in EIT technologyand clinical validation. We provide this information in EOS 9together with the recommendations for future steps needed.

CONCLUSIONWe are convinced that EIT is at an important stage of its devel-opment, which bears both opportunities and risks. EIT has thepotential of bringing new possibilities to the management ofmechanically ventilated patients and patients with chronic lungdiseases. Common standardised techniques, terminology, coord-ination and consensus on EIT application with large multicenterclinical trials are needed to promote the clinical use. We hope tohave provided most of this essential information in this article.

Author affiliations1Department of Anesthesiology and Intensive Care Medicine, University MedicalCenter Schleswig-Holstein, Campus Kiel, Kiel, Germany2Pulmonary Division, Heart Institute (InCor), Hospital das Clínicas da Faculdade deMedicina da Universidade de São Paulo, São Paulo, Brazil3Department of Neonatology, Emma Children’s Hospital, Academic Medical Center,Amsterdam, The Netherlands4Neonatal Research, Murdoch Childrens Research Institute, Parkville, Victoria,Australia5Institute of Technical Medicine, Furtwangen University, Villingen-Schwenningen,Germany

6Fraunhofer Project Group for Automation in Medicine and Biotechnology PAMB,Mannheim, Germany7Department of Anesthesiology, University Medical Center of the JohannesGutenberg-University, Mainz, Germany8Department of Systems and Computer Engineering, Carleton University, Ottawa,Ontario, Canada9Swisstom AG, Landquart, Switzerland10Dräger Medical GmbH, Lübeck, Germany11Department of Anesthesiology and Intensive Care Medicine, Sahlgrenska UniversityHospital, Gothenburg, Sweden12Department of Anesthesia, Critical Care and Emergency, Fondazione IRCCS Ca’Granda Ospedale Maggiore Policlinico, Milan, Italy13Department of Adult Critical Care, Guy’s and St Thomas’ NHS Foundation Trust,London, UK14Paediatric Critical Care Research Group, Mater Research University of Queensland,South Brisbane, Australia15Children’s Hospital Traunstein, Ludwig Maximilian’s University, Munich, Germany16Department of Adult Intensive Care, Erasmus MC, Rotterdam, The Netherlands17Philips Chair for Medical Information Technology, Helmholtz-Institute forBiomedical Engineering, RWTH Aachen University, Aachen, Germany

Acknowledgements We acknowledge the inputs from Eddy Fan, William RBLionheart, Thomas Riedel, Peter C Rimensberger, Fernando Suarez Sipmann, NorbertWeiler and Hermann Wrigge.

Collaborators TRanslational EIT developmeNt study (TREND) group: Eddy Fan;William RB Lionheart; Thomas Riedel; Peter C Rimensberger; Fernando SuarezSipmann; Norbert Weiler and Hermann Wrigge.

Contributors Writing: IF, AA, MBPA, AHvK, DGT, ZZ, BG, MB, HG, SHB and ETwith input from OS, TM, VT, LC, AS, GKW, DG and SL. Revising and approval of themanuscript: all authors and collaborators of the TREND study group.

Competing interests IF: Grants from The European Union’s 7th FrameworkProgramme for Research and Technological Development (WELCOME, Grant No.611223) and from The European Union’s Framework Programme for Research andInnovation Horizon 2020 (CRADL, Grant No. 668259), personal fees from Dräger,outside the submitted work. MBPA: Grants from Dixtal/Philips and Timpel SA,outside the submitted work. AHvK: non-financial support from CareFusion, outsidethe submitted work. DGT: Goe-MF II EIT system provided by CareFusion forunrestricted research use, two Swisstom Pioneer EIT systems and consumables fullypurchased from the manufacturer, unrestricted assistance in customising researchsoftware for specific research needs by Swisstom. ZZ: Grant from the GermanFederal Ministry of Education and Research (MOSES, Grant No. 03FH038I3),personal fees from Dräger Medical, outside the submitted work. BG: Personal feesfrom Swisstom, outside the submitted work. SHB: Co-founder, employee and ChiefMedical Officer of Swisstom AG, inventor of several EIT-related patents and patentapplications owned by Swisstom AG and Timpel SA. ET: Employee of Dräger; severalEIT-related patents pending and issued to Dräger. OS: Personal fees andnon-financial support from Dräger Medical, outside the submitted work; owner ofan issued patent EP 228009A1. VT: Personal fees from Timpel, outside thesubmitted work. DG: Personal fees from Dräger Medical, outside the submittedwork. SL: Grants, personal fees and non-financial support from Dräger Medical,grants and non-financial support from Philips Research, grants and non-financialsupport from BMBF (private/public partnership project with Weinmann GmbH),grants and non-financial support from BMWi (public/private partnership ZIM projectwith Fritz Stephan GmbH), outside the submitted work; several patents pending,licensed and issued to Dräger Medical AG & Co KGaA.

Provenance and peer review Not commissioned; externally peer reviewed.

Open Access This is an Open Access article distributed in accordance with theCreative Commons Attribution Non Commercial (CC BY-NC 4.0) license, whichpermits others to distribute, remix, adapt, build upon this work non-commercially,and license their derivative works on different terms, provided the original work isproperly cited and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/

REFERENCES1 Barber D, Brown B. Applied potential tomography. J Phys E Sci Instrum

1984;17:723–33.2 Bodenstein M, David M, Markstaller K. Principles of electrical impedance

tomography and its clinical application. Crit Care Med 2009;37:713–24.3 Costa EL, Lima RG, Amato MB. Electrical impedance tomography. Curr Opin Crit

Care 2009;15:18–24.4 Frerichs I. Electrical impedance tomography (EIT) in applications related to lung

and ventilation: a review of experimental and clinical activities. Physiol Meas2000;21:R1–21.

5 Frerichs I, Becher T, Weiler N. Electrical impedance tomography imaging of thecardiopulmonary system. Curr Opin Crit Care 2014;20:323–32.





/T


ber 2016. Dow

nloaded from



http://creativecommons.org/licenses/by-nc/4.0/

http://creativecommons.org/licenses/by-nc/4.0/

http://dx.doi.org/10.1088/0022-3735/17/9/002

http://dx.doi.org/10.1097/CCM.0b013e3181958d2f

http://dx.doi.org/10.1097/MCC.0b013e3283220e8c

http://dx.doi.org/10.1097/MCC.0b013e3283220e8c

http://dx.doi.org/10.1088/0967-3334/21/2/201

http://dx.doi.org/10.1097/MCC.0000000000000088


6 Leonhardt S, Lachmann B. Electrical impedance tomography: the holy grail ofventilation and perfusion monitoring? Intensive Care Med 2012;38:1917–29.

7 Lundin S, Stenqvist O. Electrical impedance tomography: potentials and pitfalls.Curr Opin Crit Care 2012;18:35–41.

8 Moerer O, Hahn G, Quintel M. Lung impedance measurements to monitor alveolarventilation. Curr Opin Crit Care 2011;17:260–7.

9 Muders T, Luepschen H, Putensen C. Impedance tomography as a new monitoringtechnique. Curr Opin Crit Care 2010;16:269–75.

10 Pillow JJ, Frerichs I, Stocks J. Lung function tests in neonates and infants withchronic lung disease: global and regional ventilation inhomogeneity. PediatrPulmonol 2006;41:105–21.

11 Riedel T, Frerichs I. Electrical impedance tomography. Eur Respir Mon2010;47:195–205.

12 Frerichs I, Hahn G, Hellige G. Thoracic electrical impedance tomographicmeasurements during volume controlled ventilation-effects of tidal volume andpositive end-expiratory pressure. IEEE Trans Med Imaging 1999;18:764–73.

13 Karsten J, Stueber T, Voigt N, et al. Influence of different electrode belt positionson electrical impedance tomography imaging of regional ventilation: a prospectiveobservational study. Crit Care 2016;20:3.

14 Reifferscheid F, Elke G, Pulletz S, et al. Regional ventilation distribution determinedby electrical impedance tomography: reproducibility and effects of posture andchest plane. Respirology 2011;16:523–31.

15 Krueger-Ziolek S, Schullcke B, Kretschmer J, et al. Positioning of electrode planesystematically influences EIT imaging. Physiol Meas 2015;36:1109–18.

16 Frerichs I, Braun P, Dudykevych T, et al. Distribution of ventilation in young andelderly adults determined by electrical impedance tomography. Resp PhysiolNeurobiol 2004;143:63–75.

17 Frerichs I, Schiffmann H, Oehler R, et al. Distribution of lung ventilation inspontaneously breathing neonates lying in different body positions. Intensive CareMed 2003;29:787–94.

18 Heinrich S, Schiffmann H, Frerichs A, et al. Body and head position effects onregional lung ventilation in infants: an electrical impedance tomography study.Intensive Care Med 2006;32:1392–8.

19 Lupton-Smith AR, Argent AC, Rimensberger PC, et al. Challenging a paradigm:positional changes in ventilation distribution are highly variable in healthy infantsand children. Pediatr Pulmonol 2014;49:764–71.

20 Frerichs I, Hahn G, Golisch W, et al. Monitoring perioperative changes indistribution of pulmonary ventilation by functional electrical impedancetomography. Acta Anaesthesiol Scand 1998;42:721–6.

21 Humphreys S, Pham TM, Stocker C, et al. The effect of induction of anesthesiaand intubation on end-expiratory lung level and regional ventilation distribution incardiac children. Paediatr Anaesth 2011;21:887–93.

22 Yoshida T, Torsani V, Gomes S, et al. Spontaneous effort causes occult pendelluftduring mechanical ventilation. Am J Respir Crit Care Med 2013;188:1420–7.

23 Frerichs I, Schiffmann H, Hahn G, et al. Non-invasive radiation-free monitoring ofregional lung ventilation in critically ill infants. Intensive Care Med2001;27:1385–94.

24 Blankman P, Hasan D, van Mourik MS, et al. Ventilation distribution measuredwith EIT at varying levels of pressure support and neurally adjusted ventilatoryassist in patients with ALI. Intensive Care Med 2013;39:1057–62.

25 Mauri T, Bellani G, Confalonieri A, et al. Topographic distribution of tidalventilation in acute respiratory distress syndrome: effects of positive end-expiratorypressure and pressure support. Crit Care Med 2013;41:1664–73.

26 Rabbani KS, Kabir AM. Studies on the effect of the third dimension on atwo-dimensional electrical impedance tomography system. Clin Phys Physiol Meas1991;12:393–402.

27 Lionheart WR. EIT reconstruction algorithms: pitfalls, challenges and recentdevelopments. Physiol Meas 2004;25:125–42.

28 Yasin M, Böhm S, Gaggero PO, et al. Evaluation of EIT system performance.Physiol Meas 2011;32:851–65.

29 Frerichs I, Dargaville PA, Dudykevych T, et al. Electrical impedance tomography: amethod for monitoring regional lung aeration and tidal volume distribution?Intensive Care Med 2003;29:2312–16.

30 Grivans C, Lundin S, Stenqvist O, et al. Positive end-expiratory pressure-inducedchanges in end-expiratory lung volume measured by spirometry and electricimpedance tomography. Acta Anaesthesiol Scand 2011;55:1068–77.

31 Becher TH, Bui S, Zick G, et al. Assessment of respiratory system compliance withelectrical impedance tomography using a positive end-expiratory pressure wavemaneuver during pressure support ventilation: a pilot clinical study. Crit Care2014;18:679.

32 Geddes LA, Baker LE. The specific resistance of biological material – acompendium of data for the biomedical engineer and physiologist. Med Biol Eng1967;5:271–93.

33 Vogt B, Mendes L, Chouvarda I, et al. Influence of torso and arm positions on chestexaminations by electrical impedance tomography. Physiol Meas 2016;37:904–21.

34 Frerichs I, Pulletz S, Elke G, et al. Patient examinations using electrical impedancetomography-sources of interference in the intensive care unit. Physiol Meas2011;32:L1–L10.

35 Pulletz S, van Genderingen HR, Schmitz G, et al. Comparison of different methodsto define regions of interest for evaluation of regional lung ventilation by EIT.Physiol Meas 2006;27:S115–27.

36 Becher T, Vogt B, Kott M, et al. Functional regions of interest in electricalimpedance tomography: a secondary analysis of two clinical studies. PLoS One2016;11:e0152267.

37 Frerichs I, Dargaville PA, van Genderingen H, et al. Lung volume recruitment aftersurfactant administration modifies spatial distribution of ventilation. Am J RespirCrit Care Med 2006;174:772–9.

38 Lowhagen K, Lundin S, Stenqvist O. Regional intratidal gas distribution in acutelung injury and acute respiratory distress syndrome-assessed by electric impedancetomography. Minerva Anestesiol 2010;76:1024–35.

39 Bhatia R, Schmölzer GM, Davis PG, et al. Electrical impedance tomography canrapidly detect small pneumothoraces in surfactant-depleted piglets. Intensive CareMed 2012;38:308–15.

40 Becher T, Kott M, Schädler D, et al. Influence of tidal volume onventilation inhomogeneity assessed by electrical impedance tomographyduring controlled mechanical ventilation. Physiol Meas 2015;36:1137–46.

41 Meier T, Luepschen H, Karsten J, et al. Assessment of regional lung recruitmentand derecruitment during a PEEP trial based on electrical impedance tomography.Intensive Care Med 2008;34:543–50.

42 Vogt B, Pulletz S, Elke G, et al. Spatial and temporal heterogeneity of regionallung ventilation determined by electrical impedance tomography during pulmonaryfunction testing. J Appl Physiol 2012;113:1154–61.

43 Frerichs I, Dudykevych T, Hinz J, et al. Gravity effects on regional lung ventilationdetermined by functional EIT during parabolic flights. J Appl Physiol2001;91:39–50.

44 Pulletz S, Kott M, Elke G, et al. Dynamics of regional lung aeration determined byelectrical impedance tomography in patients with acute respiratory distresssyndrome. Multidiscip Respir Med 2012;7:44.

45 Frerichs I, Dargaville PA, Rimensberger PC. Regional respiratory inflation anddeflation pressure-volume curves determined by electrical impedance tomography.Physiol Meas 2013;34:567–77.

46 Miedema M, de Jongh FH, Frerichs I, et al. Changes in lung volume andventilation during surfactant treatment in ventilated preterm infants. Am J RespirCrit Care Med 2011;184:100–5.

47 Pulletz S, Adler A, Kott M, et al. Regional lung opening and closing pressures inpatients with acute lung injury. J Crit Care 2012;27:323.e11–8.

48 Gómez-Laberge C, Rettig JS, Smallwood CD, et al. Interaction of dependent andnon-dependent regions of the acutely injured lung during a stepwise recruitmentmanoeuvre. Physiol Meas 2013;34:163–77.

49 Zhao Z, Möller K, Steinmann D, et al. Evaluation of an electrical impedancetomography-based global inhomogeneity index for pulmonary ventilationdistribution. Intensive Care Med 2009;35:1900–6.

50 Frerichs I, Achtzehn U, Pechmann A, et al. High-frequency oscillatory ventilation inpatients with acute exacerbation of chronic obstructive pulmonary disease. J CritCare 2012;27:172–81.

51 Kunst PW, Vonk Noordegraaf A, Raaijmakers E, et al. Electrical impedancetomography in the assessment of extravascular lung water in noncardiogenic acuterespiratory failure. Chest 1999;116:1695–702.

52 Zick G, Elke G, Becher T, et al. Effect of PEEP and tidal volume on ventilationdistribution and end-expiratory lung volume: a prospective experimental animaland pilot clinical study. PLoS One 2013;8:e72675.

53 Tingay DG, Wallace MJ, Bhatia R, et al. Surfactant before the first inflation at birthimproves spatial distribution of ventilation and reduces lung injury in pretermlambs. J Appl Physiol 2014;116:251–8.

54 Grychtol B, Elke G, Meybohm P, et al. Functional validation and comparisonframework for EIT lung imaging. PLoS One 2014;9:e103045.

55 Borges JB, Suarez-Sipmann F, Bohm SH, et al. Regional lung perfusion estimatedby electrical impedance tomography in a piglet model of lung collapse. J ApplPhysiol 2012;112:225–36.

56 Grychtol B, Wolf GK, Arnold JH. Differences in regional pulmonarypressure-impedance curves before and after lung injury assessed with a novelalgorithm. Physiol Meas 2009;30:S137–48.

57 Dargaville PA, Rimensberger PC, Frerichs I. Regional tidal ventilation andcompliance during a stepwise vital capacity manoeuvre. Intensive Care Med2010;36:1953–61.

58 Wolf GK, Gómez-Laberge C, Rettig JS, et al. Mechanical ventilation guided byelectrical impedance tomography in experimental acute lung injury. Crit Care Med2013;41:1296–304.

59 Miedema M, de Jongh FH, Frerichs I, et al. Regional respiratory time constantsduring lung recruitment in high-frequency oscillatory ventilated preterm infants.Intensive Care Med 2012;38:294–9.

60 Riedel T, Kyburz M, Latzin P, et al. Regional and overall ventilation inhomogeneitiesin preterm and term-born infants. Intensive Care Med 2009;35:144–51.

61 Wrigge H, Zinserling J, Muders T, et al. Electrical impedance tomographycompared with thoracic computed tomography during a slow inflation maneuver inexperimental models of lung injury. Crit Care Med 2008;36:903–9.





/T


ber 2016. Dow

nloaded from

http://dx.doi.org/10.1007/s00134-012-2684-z

http://dx.doi.org/10.1097/MCC.0b013e32834eb462

http://dx.doi.org/10.1097/MCC.0b013e3283463c9c

http://dx.doi.org/10.1097/MCC.0b013e3283390cbf

http://dx.doi.org/10.1002/ppul.20319


http://dx.doi.org/10.1109/42.802754

http://dx.doi.org/10.1186/s13054-015-1161-9

http://dx.doi.org/10.1111/j.1440-1843.2011.01929.x

http://dx.doi.org/10.1088/0967-3334/36/6/1109

http://dx.doi.org/10.1016/j.resp.2004.07.014

http://dx.doi.org/10.1016/j.resp.2004.07.014

http://dx.doi.org/10.1007/s00134-003-1726-y

http://dx.doi.org/10.1007/s00134-003-1726-y

http://dx.doi.org/10.1007/s00134-006-0252-0


http://dx.doi.org/10.1111/j.1399-6576.1998.tb05308.x

http://dx.doi.org/10.1111/j.1460-9592.2011.03547.x

http://dx.doi.org/10.1164/rccm.201303-0539OC

http://dx.doi.org/10.1007/s001340101021

http://dx.doi.org/10.1007/s00134-013-2898-8

http://dx.doi.org/10.1097/CCM.0b013e318287f6e7

http://dx.doi.org/10.1088/0143-0815/12/4/009

http://dx.doi.org/10.1088/0967-3334/25/1/021

http://dx.doi.org/10.1088/0967-3334/32/7/S09

http://dx.doi.org/10.1007/s00134-003-2029-z

http://dx.doi.org/10.1111/j.1399-6576.2011.02511.x

http://dx.doi.org/10.1186/s13054-014-0679-6

http://dx.doi.org/10.1007/BF02474537

http://dx.doi.org/10.1088/0967-3334/37/6/904

http://dx.doi.org/10.1088/0967-3334/32/12/F01

http://dx.doi.org/10.1088/0967-3334/27/5/S10

http://dx.doi.org/10.1371/journal.pone.0152267



http://dx.doi.org/10.1007/s00134-011-2421-z

http://dx.doi.org/10.1007/s00134-011-2421-z

http://dx.doi.org/10.1088/0967-3334/36/6/1137

http://dx.doi.org/10.1007/s00134-007-0786-9

http://dx.doi.org/10.1152/japplphysiol.01630.2011

http://dx.doi.org/10.1186/2049-6958-7-44

http://dx.doi.org/10.1088/0967-3334/34/6/567



http://dx.doi.org/10.1016/j.jcrc.2011.09.002

http://dx.doi.org/10.1088/0967-3334/34/2/163

http://dx.doi.org/10.1007/s00134-009-1589-y



http://dx.doi.org/10.1378/chest.116.6.1695






http://dx.doi.org/10.1088/0967-3334/30/6/S09

http://dx.doi.org/10.1007/s00134-010-1995-1

http://dx.doi.org/10.1097/CCM.0b013e3182771516

http://dx.doi.org/10.1007/s00134-011-2410-2

http://dx.doi.org/10.1007/s00134-008-1299-x

http://dx.doi.org/10.1097/CCM.0B013E3181652EDD


62 Vogt B, Zhao Z, Zabel P, et al. Regional lung response to bronchodilatorreversibility testing determined by electrical impedance tomography in chronicobstructive pulmonary disease. Am J Physiol Lung Cell Mol Physiol 2016;311:L8–L19.

63 Frerichs I, Zhao Z, Becher T, et al. Regional lung function determined by electricalimpedance tomography during bronchodilator reversibility testing in patients withasthma. Physiol Meas 2016;37:698–712.

64 Frerichs I, Hinz J, Herrmann P, et al. Detection of local lung air content byelectrical impedance tomography compared with electron beam CT. J Appl Physiol2002;93:660–6.

65 Elke G, Fuld MK, Halaweish AF, et al. Quantification of ventilation distribution inregional lung injury by electrical impedance tomography and xenon computedtomography. Physiol Meas 2013;34:1303–18.

66 Victorino JA, Borges JB, Okamoto VN, et al. Imbalances in regional lungventilation: a validation study on electrical impedance tomography. Am J RespirCrit Care Med 2004;169:791–800.

67 Hinz J, Neumann P, Dudykevych T, et al. Regional ventilation by electricalimpedance tomography: a comparison with ventilation scintigraphy in pigs. Chest2003;124:314–22.

68 Richard JC, Pouzot C, Gros A, et al. Electrical impedance tomography compared topositron emission tomography for the measurement of regional lung ventilation:an experimental study. Crit Care 2009;13:R82.

69 Shi C, Boehme S, Bentley AH, et al. Assessment of regional ventilationdistribution: comparison of vibration response imaging (VRI) with electricalimpedance tomography (EIT). PLoS One 2014;9:e86638.

70 Hinz J, Moerer O, Neumann P, et al. Effect of positiveend-expiratory-pressure on regional ventilation in patients with acute lunginjury evaluated by electrical impedance tomography. Eur J Anaesthesiol2005;22:817–25.

71 Coulombe N, Gagnon H, Marquis F, et al. A parametric model of the relationshipbetween EIT and total lung volume. Physiol Meas 2005;26:401–11.

72 Odenstedt H, Lindgren S, Olegård C, et al. Slow moderate pressure recruitmentmaneuver minimizes negative circulatory and lung mechanic side effects:evaluation of recruitment maneuvers using electric impedance tomography.Intensive Care Med 2005;31:1706–14.

73 Costa EL, Borges JB, Melo A, et al. Bedside estimation of recruitable alveolarcollapse and hyperdistension by electrical impedance tomography. Intensive CareMed 2009;35:1132–7.

74 Karsten J, Grusnick C, Paarmann H, et al. Positive end-expiratory pressure titrationat bedside using electrical impedance tomography in post-operative cardiac surgerypatients. Acta Anaesthesiol Scand 2015;59:723–32.

75 Blankman P, Hasan D, Erik G, et al. Detection of ‘best’ positive end-expiratorypressure derived from electrical impedance tomography parameters during adecremental positive end-expiratory pressure trial. Crit Care 2014;18:R95.

76 Serrano RE, de LB, Casas O, et al. Use of electrical impedance tomography (EIT)for the assessment of unilateral pulmonary function. Physiol Meas2002;23:211–20.

77 Zhao Z, Müller-Lisse U, Frerichs I, et al. Regional airway obstruction in cysticfibrosis determined by electrical impedance tomography in comparison with highresolution CT. Physiol Meas 2013;34:N107–14.

78 Spaeth J, Daume K, Goebel U, et al. Increasing positive end-expiratory pressure(re-)improves intraoperative respiratory mechanics and lung ventilation after pronepositioning. Br J Anaesth 2016;116:838–46.

79 Costa EL, Chaves CN, Gomes S, et al. Real-time detection of pneumothorax usingelectrical impedance tomography. Crit Care Med 2008;36:1230–8.

80 Tingay DG, Copnell B, Grant CA, et al. The effect of endotracheal suction onregional tidal ventilation and end-expiratory lung volume. Intensive Care Med2010;36:888–96.

81 Alves SH, Amato MB, Terra RM, et al. Lung reaeration and reventilation afteraspiration of pleural effusions. A study using electrical impedance tomography.Ann Am Thorac Soc 2014;11:186–91.

82 Pulletz S, Elke G, Zick G, et al. Performance of electrical impedance tomography indetecting regional tidal volumes during one-lung ventilation. Acta AnaesthesiolScand 2008;52:1131–9.

83 Frerichs I, Pulletz S, Elke G, et al. Assessment of changes in distribution of lungperfusion by electrical impedance tomography. Respiration 2009;77:282–91.

84 Vonk Noordegraaf A, Kunst PW, Janse A, et al. Pulmonary perfusion measured bymeans of electrical impedance tomography. Physiol Meas 1998;19:263–73.

85 Smit HJ, Vonk Noordegraaf A, Roeleveld RJ, et al. Epoprostenol-inducedpulmonary vasodilatation in patients with pulmonary hypertension measured byelectrical impedance tomography. Physiol Meas 2002;23:237–43.

86 Li Y, Tesselaar E, Borges JB, et al. Hyperoxia affects the regional pulmonaryventilation/perfusion ratio: an electrical impedance tomography study. ActaAnaesthesiol Scand 2014;58:716–25.

87 Elke G, Pulletz S, Schädler D, et al. Measurement of regional pulmonary oxygenuptake – novel approach using electrical impedance tomography. Physiol Meas2011;32:877–86.

88 Frerichs I, Hinz J, Herrmann P, et al. Regional lung perfusion as determined byelectrical impedance tomography in comparison with electron beam CT imaging.IEEE Trans Med Imaging 2002;21:646–52.

89 Miedema M, de Jongh FH, Frerichs I, et al. Changes in lung volume andventilation during lung recruitment in high-frequency ventilated preterm infantswith respiratory distress syndrome. J Pediatr 2011;159:199–205.e2.

90 Wolf GK, Gómez-Laberge C, Kheir JN, et al. Reversal of dependent lung collapsepredicts response to lung recruitment in children with early acute lung injury.Pediatr Crit Care Med 2012;13:509–15.

91 Armstrong RK, Carlisle HR, Davis PG, et al. Distribution of tidal ventilation duringvolume-targeted ventilation is variable and influenced by age in the preterm lung.Intensive Care Med 2011;37:839–46.

92 van Veenendaal MB, Miedema M, de Jongh FH, et al. Effect of closedendotracheal suction in high-frequency ventilated premature infants measured withelectrical impedance tomography. Intensive Care Med 2009;35:2130–4.

93 Schibler A, Yuill M, Parsley C, et al. Regional ventilation distribution innon-sedated spontaneously breathing newborns and adults is not different. PediatrPulmonol 2009;44:851–8.

94 Miedema M, Frerichs I, de Jongh FH, et al. Pneumothorax in a preterm infantmonitored by electrical impedance tomography: a case report. Neonatology2011;99:10–13.

95 van der Burg PS, Miedema M, de Jongh FH, et al. Unilateral atelectasis in apreterm infant monitored with electrical impedance tomography: a case report. EurJ Pediatr 2014;173:1715–17.

96 Steinmann D, Engehausen M, Stiller B, et al. Electrical impedance tomography forverification of correct endotracheal tube placement in paediatric patients: afeasibility study. Acta Anaesthesiol Scand 2013;57:881–7.

97 Carlisle HR, Armstrong RK, Davis PG, et al. Regional distribution of blood volumewithin the preterm infant thorax during synchronised mechanical ventilation.Intensive Care Med 2010;36:2101–8.

98 Schibler A, Pham TM, Moray AA, et al. Ventilation and cardiac related impedancechanges in children undergoing corrective open heart surgery. Physiol Meas2013;34:1319–27.

99 van der Burg PS, Miedema M, de Jongh FH, et al. Cross-sectional changes in lungvolume measured by electrical impedance tomography are representative for thewhole lung in ventilated preterm infants. Crit Care Med 2014;42:1524–30.

100 Adler A, Amato MB, Arnold JH, et al. Whither lung EIT: where are we, where dowe want to go and what do we need to get there? Physiol Meas2012;33:679–94.





/T


ber 2016. Dow

nloaded from

http://dx.doi.org/10.1152/ajplung.00463.2015

http://dx.doi.org/10.1088/0967-3334/37/6/698


http://dx.doi.org/10.1088/0967-3334/34/10/1303



http://dx.doi.org/10.1378/chest.124.1.314

http://dx.doi.org/10.1186/cc7900


http://dx.doi.org/10.1017/S0265021505001377

http://dx.doi.org/10.1088/0967-3334/26/4/006

http://dx.doi.org/10.1007/s00134-005-2799-6

http://dx.doi.org/10.1007/s00134-009-1447-y

http://dx.doi.org/10.1007/s00134-009-1447-y

http://dx.doi.org/10.1111/aas.12518

http://dx.doi.org/10.1186/cc13866

http://dx.doi.org/10.1088/0967-3334/23/1/322

http://dx.doi.org/10.1088/0967-3334/34/11/N107

http://dx.doi.org/10.1093/bja/aew115

http://dx.doi.org/10.1097/CCM.0b013e31816a0380

http://dx.doi.org/10.1007/s00134-010-1849-x

http://dx.doi.org/10.1513/AnnalsATS.201306-142OC

http://dx.doi.org/10.1111/j.1399-6576.2008.01706.x

http://dx.doi.org/10.1111/j.1399-6576.2008.01706.x

http://dx.doi.org/10.1159/000193994

http://dx.doi.org/10.1088/0967-3334/19/2/013

http://dx.doi.org/10.1088/0967-3334/23/1/324



http://dx.doi.org/10.1088/0967-3334/32/7/S11

http://dx.doi.org/10.1109/TMI.2002.800585

http://dx.doi.org/10.1016/j.jpeds.2011.01.066

http://dx.doi.org/10.1097/PCC.0b013e318245579c

http://dx.doi.org/10.1007/s00134-011-2157-9

http://dx.doi.org/10.1007/s00134-009-1663-5



http://dx.doi.org/10.1159/000292626

http://dx.doi.org/10.1007/s00431-014-2399-y

http://dx.doi.org/10.1007/s00431-014-2399-y


http://dx.doi.org/10.1007/s00134-010-2049-4

http://dx.doi.org/10.1088/0967-3334/34/10/1319

http://dx.doi.org/10.1097/CCM.0000000000000230

http://dx.doi.org/10.1088/0967-3334/33/5/679


Chest electrical impedance tomography examination, data analysis,

terminology, clinical use and recommendations: consensus statement of

the TRanslational EIT developmeNt stuDy group

Inéz Frerichs, Marcelo B. P. Amato, Anton H. van Kaam, David G. Tingay, Zhanqi Zhao,

Bartłomiej Grychtol, Marc Bodenstein, Hervé Gagnon, Stephan H. Böhm, Eckhard Teschner,

Ola Stenqvist, Tommaso Mauri, Vinicius Torsani, Luigi Camporota, Andreas Schibler, Gerhard

K. Wolf, Diederik Gommers, Steffen Leonhardt, Andy Adler, TREND study group

ONLINE SUPPLEMENT 1

Execution of EIT chest measurements

Execut

Electrica

of the c

EIT me

part of

EIT dev

across s

The cur

body s

examina

EIT exa

highligh

Figure

EOS 1 (

EOS, ele

center o

right; L,

Genera

We est

inventio

attempt

last cen

1) have

clinical s

tion of EI

al impedanc

chest using

easurement

an electrod

vice. While a

sets of elec

rrents used

surface. Th

ations are c

aminations a

hted.

E1.1. Sequ

(gray backg

ectronic onl

of ventilatio

, left.

al commen

timate that

on of EIT, t

t to genera

ntury (1). In

e most frequ

studies), an

IT chest m

ce Tomogra

non-invasiv

requires th

e belt) onto

acquiring da

trodes, whi

are imperce

his electron

carried out.

and EIT dat

uence of pro

round) desc

line supplem

on; U/L, upp

nt

more tha

typically in

te the list o

n this supple

uently been

nd/or 2) are

measurem

aphy (EIT)

ve electrica

he placeme

o the area o

ata, an EIT

le the resu

eptible and

nic online

Figure E1.

ta analysis

ocesses invo

cribes the p

ment; tdEIT

per-to-lower

n 50 diffe

university r

of the exist

ement docu

n used unde

e currently c

ments

calculates t

al measurem

ent of elect

of interest a

T device per

ltant voltag

below elect

supplemen

.1 shows th

for orientat

olved in EIT

process of ra

T, time-diffe

r ventilation

rent EIT d

research lab

ting devices

ument, only

er in-vivo co

commercial

tomographi

ments at ele

trodes (eith

and connec

rforms a seq

ges are mea

trical safety

nt (EOS)

he sequence

tion. The pr

T chest exa

aw image g

erence EIT;

n ratio; RVD

devices hav

bs. Accordin

s originates

y a few EIT

onditions (i.

ly available

c (i.e. cros

ectrodes on

her individua

tion of thes

quence of c

asured at th

y norms for

1 describe

es of all pro

rocess desc

mination an

generation.

ROI, region

D, regional v

ve been de

ng to our k

s from the

devices are

.e. in anima

. (Table E1

ss-sectional)

n the body

al electrode

se electrode

current stim

he other ele

macroshoc

es how EI

ocesses inv

cribed in thi

nd data ana

Abbreviatio

n-of-interes

ventilation d

eveloped si

knowledge,

mid-ninetie

e mentioned

al experime

.1 provides

2

) images

surface.

es or as

es to the

mulations

ectrodes.

ck on the

T chest

volved in

s EOS is

alysis.

ons:

st; CoV,

delay; R,

ince the

the last

es of the

d which:

ental and

s a list of

3

these devices with some basic characteristics.) The choice to mention a device is thus based

on these criteria, and does not imply our recommendation. We describe the general features

of EIT data acquisition that are common in EIT examinations of humans and only briefly

refer to some device-specific characteristics. For other, detailed specific features of the EIT

devices the users are advised to follow the instructions of use of the individual producers of

EIT technology.

Table E1.1. List of commercial EIT devices frequently used in examinations of human

subjects.

Manufacturer EIT system Electrodes (number,

configuration)

Characteristics

CareFusion Goe-MF II 16, individual electrodes Pair drive (adjacent), serial measurement. Algorithm: Sheffield back-projection

Dräger Medical PulmoVista 500 * 16, electrode belt Pair drive (adjacent), serial measurement. Algorithm: FEM-based Newton-Raphson method

Maltron Inc Mark 1

Mark 3.5 *

16, individual electrodes

8, individual electrodes

Pair drive (adjacent), serial measurement. Algorithm: Sheffield back-projection

Swisstom AG BB2 * 32, electrode belt Pair drive (adjustable skip), serial measurement. Algorithm: GREIT

Timpel SA Enlight * 32, electrode stripes Pair drive (3-electrode skip), parallel measurement: Algorithm: FEM-based Newton-Raphson method

*, commercially available as of May 2016; FEM, finite element method.

EIT electrode array

EIT examinations require the placement of electrodes on the chest circumference usually in

the mid or caudal part of the chest. The plane is chosen to be sensitive to the phenomena of

interest. The electrodes are either single self-adhesive electrodes (e.g. ECG electrodes) that

have to be placed on the chest individually with possibly equal spacing or they are integrated

into one electrode belt or two self-adhesive electrode stripes. The latter solutions render the

application more user-friendly.

Electrode plane

The array of EIT electrodes is typically placed in one transverse plane, although oblique

placement has also been described (2) and is used in one commercially available EIT device.

4

More than one plane has been used in a few studies to document the regional ventilation

differences in two (3, 4) or three chest slices (5, 6). Acquisition of EIT data in more than one

plane may offer the possibility to assess the lung electrical properties in a large section of

the chest (7). However, true 3D EIT imaging has not achieved any relevant state of

development.

The location of the electrode plane impacts the findings because EIT is most sensitive

to nearby anatomical locations. The effective thickness of the studied slice of the chest is not

uniform (8). It is the largest in the center of the chest and decreases toward the chest

boundary (see EOS 2, Figure E2.2). Therefore, it is necessary to document the exact location

of the electrode array to secure comparability of examinations performed on separate

occasions. Typical anatomical landmarks should be used; when describing the EIT electrode

plane we recommend specifying the exact intercostal space at the parasternal line. The

distance relative to the internipple line has occasionally been used to describe the location of

the electrode plane. However, we do not recommend this as the distance is impacted by

body growth in children and not suitable in women.

Electrode placement in the 5th to 6th intercostal spaces at the parasternal line allows

good representation of impedance changes mainly in the lower lobes of the right and left

lungs as well as in the heart region. It is generally not recommended to place the electrodes

below the 6th intercostal space in supine subjects because the diaphragm may periodically

enter the measurement plane (4, 5), and the sensitivity to abdominal conductivity increases.

More cranial lung regions are assessed by EIT when the electrode plane is located in the 3rd

to 4th intercostal spaces. In certain applications it is considered an advantage that less heart

is represented in these parts of the chest (9, 10).

Placement of the electrode array

The placement of the electrode array is easy in awake, cooperative upright subjects.

However, patients are often sedated or anaesthetized and thus uncooperative. They usually

lie in the supine position. Placement of the electrode array on the chest under these

conditions usually requires two persons.

The placement of electrode belts on the chest is relatively straightforward. The

patient needs to be turned to one side to partially expose the back and the belt is then

moved below the body. Afterwards the patient is turned to the other side and the belt is

pulled out and closed. Alternatively, the patient is slightly leaned forward and the belt is

moved from above underneath the head towards the chest. When two electrode stripes are

used, the preferred way of placement is to turn the lying patient at first to one and then to

the other side and attach the stripes starting at the backbone.

5

With individual electrodes, we recommend placement of the electrode on the sternum,

i.e. on the front part of the chest at first and then one electrode each in the left and right

midaxilar lines. Afterwards, the electrodes in between these first three electrodes are

attached. Then the patient is slightly turned to one side and an electrode is placed directly

on the backbone. The remaining electrodes between the backbone and the midaxillar

electrodes are attached after that. Finally, the patient is turned to the other side and the last

remaining electrodes are attached.

Especially with individual electrodes, care must be taken that no electrical contact exists

between the electrodes. This is seldom a problem in adults, but can be when preterm infants

are examined. In this specific group of patients, the individual electrodes need to be trimmed

to fit on the very small chests (Fig. E1.2). Nonetheless, it has been shown in a few clinical

studies that even infants weighing less than 1000 g can be examined by EIT (11-14).

Unfortunately, there exist no EIT electrode belts for the use in neonates and small infants at

present.

Figure E1.2. Placement of individual EIT electrodes in a 10-day old spontaneously breathing

neonate. X-ray translucent electrodes were trimmed to fit on the small chest and placed on

the chest circumference. The examination (Goe-MF II EIT device, CareFusion, Höchberg,

Germany) was performed to assess the effect of prone position on ventilation distribution in

the transverse plane with written informed consent of the parents (15).

6

Number of EIT electrodes

The first commercially available EIT system, the Sheffield Mk1 device but also some of the

more recently developed commercial (Goe-MF II, CareFusion, Höchberg, Germany or the

PulmoVista 500, Dräger, Lübeck, Germany) or non-commercial devices operate with 16

electrodes. EIT systems with 32 electrodes (Enlight, Timpel, Sao Paulo, Brazil and Swisstom

BB2, Swisstom, Landquart, Switzerland) and 8 electrodes (Maltron Sheffield Mark 3.5,

Maltron, Raleigh, UK) have also been developed.

The higher number of electrodes gives two advantages: an improved spatial

resolution and a redundancy of data, from which the reconstruction algorithm can reject

measurements at poorly connected electrodes. Because of the diffuse pattern of electric

current flow, image resolution in the interior of the thorax does not improve in proportion to

the number of electrodes (16).

The lower number of electrodes may simplify the placement of electrodes in preterm

infants with very small chest dimensions (17), however, the spatial resolution deteriorates

(16). Also, clearly, an additional advantage is the reduced cost of the EIT hardware

production.

Numbering of EIT electrodes

The image reconstruction algorithm needs to “know” the position of the electrodes on the

chest circumference in order to generate proper chest EIT images. If electrode belts or the

individual electrodes are placed on the chest in an arbitrary way, image orientation will be

correspondingly arbitrary. Therefore, the instructions how the electrode belt is applied and

where it is closed need to be exactly followed. Individual electrodes need to be connected

with the leads in a specified order.

As mentioned above, in systems using individual electrodes like the Sheffield Mk1 or

Mk3.5 or the Goe-MF II devices, one of the electrodes is located directly in the middle of

sternum and one on the backbone. This is not the case with the electrode belts and stripes

of other EIT devices, where the electrodes are to each side of the sternum and backbone

and not directly on them.

In some cases, it is necessary to adjust placement of electrodes, in order to work

around body surface wounds, bandages or catheters. When individual electrodes are used,

the electrodes can be placed above or below the correct site. Such off-plane placement

should be recorded.

7

Electrode contact

Good electrode contact (with low electrical impedance) is essential to the recording of high

quality data. To improve the skin contact of the electrodes some producers of EIT

technology recommend the use of electrode gel or spray. In the specific group of preterm

and term neonates, where very small individual EIT electrodes close to each other need to

be applied on the chest, the use of skin lotions immediately before the measurement may

affect the EIT measured data. This has been observed with EIT systems using the adjacent

pattern of current applications and voltage measurements (see also EOS 2), therefore, we do

not recommended excessive use of body lotion prior to examination in this patient group.

Generally, EIT measurements can be performed even when the impedance at the

body-electrode interface is high (e.g. when the skin is dry) or non-uniform, however, it

should be aspired to keep the body-electrode impedance low, stable and similar among the

electrodes. It is recommended that no electrical contact is formed by conductive fluid (i.e.

saline or excessive sweat) between electrodes.

With modern EIT devices, an EIT examination can be initiated almost immediately

after the electrode belt or the individual electrodes have been attached on the chest and the

patient has been connected with the device. However, it is known that the electrode contacts

improve with time (18, 19); therefore, whenever possible, a delayed onset of the

measurement by approx. 5-10 minutes is recommended. This time period may be beneficial

for some devices to stabilize after the power-on. Passive or active movement of patients or

touching of electrodes or leads during EIT data acquisition should be avoided. Large chest

wounds or multiple chest tubes may preclude the measurement. The best quality of EIT

images is achieved when the EIT signal quality is good at all electrodes. However, even with

missing or faulty electrodes EIT images can be generated (20, 21), albeit of slightly lower

quality. With this respect, it is important that the signal quality at individual electrodes is

monitored and that low quality and/or missing data are taken into account during EIT image

reconstruction. Modern commercial EIT devices automatically warn the investigator when the

signal quality is low and a schematic drawing highlighting the faulty electrodes is displayed.

An automatic compensation is offered by some devices.

EIT measurement and data acquisition

During an EIT measurement, very small alternating electrical currents are applied through

pairs of electrodes, and the resulting voltages measured on the remaining passive

electrodes. The electrical currents used by EIT are imperceptible and safe for body surface

application. (The use of EIT in patients with cardiac pacemakers and with electrically active

implants like cardioverter-defibrillators is not recommended). The frequency and the

8

amplitude of the current should be specified. The most widespread spatial pattern of current

applications and voltage measurements is through adjacent electrode pairs. Because of its

prevalence, this pattern will continue to be broadly used. However, other patterns of current

applications and voltage measurements can be expected to replace the adjacent electrode

pattern in the future because they offer technical advantages (22). Most commercial EIT

devices do not offer the user the choice of measurement patterns, but select an appropriate

pattern compatible with the vendor-specific image reconstruction algorithm. It is important

to mention the measurement pattern used, if it is not determined by the choice of EIT device

used.

Each set of EIT data from which an image can be reconstructed is called a frame,

and the rate at which data are acquired is the scan (or frame) rate. Modern EIT devices are

capable of high scan rates with respect to the physiological processes of interest (for more

information see EOS 2). The scanning period of the first EIT devices was limited by the

performance of at that time existing hardware and computers. Nowadays, continuous

scanning is possible enabling long-term examinations.

Factors affecting or interfering with EIT measurements

Patient-related factors

Since posture (3, 23-25) and the type of ventilation (26-31) affect the findings they need to

be taken into account and stated in the measurement description. Specific ventilation

maneuvers performed during either spontaneous or mechanical ventilation also need to be

documented. The ventilation mode (assisted or controlled) and the settings should also be

provided during mechanical ventilation.

Body movement (either active or passive) and speech may affect the acquired EIT data.

Restricted thoracic movement has also been shown to exert an effect on the EIT findings

(32). Other patient-related factors typically encountered in the medical environment either

impact the electrode contact (presence of fresh blood or other body fluids, electrolyte

solutions, wet compresses) or the propagation of the electrical current in the chest (metal

sutures, breast implants, chest tubes, electrically active implants). Rapid changes in body

fluid volumes induced, e.g. by massive bleeding of quick administration of crystalloid or

colloid fluids modify the electrical properties of the chest. This may affect the interpretation

of EIT results, especially regarding the end-expiratory lung volume, ventilation distribution

findings seem not to be affected (33).

9

Other factors

Some medical devices may interfere with EIT examinations (34). Typical potential sources of

interference are devices using the same measuring principle as EIT like impedance

pneumography or impedance cardiography, especially when operating with drive currents of

a similar frequency as the EIT excitation currents. Modern EIT devices allow automatic or

manual change of the EIT drive frequency when this type of electrical interference is

detected, thus, this effect may be eliminated when present.

Continuous cardiac output monitoring has also been identified as a possible source of

interference with EIT measurements (34). Online or offline frequency filtering of the EIT

signals may enable meaningful interpretation of the acquired data (see also EOS 3).

However, it is generally recommended to eliminate the sources of interference before EIT

data acquisition whenever possible.

Pulsating air suspension mattresses, typically used in patients treated in intensive

care units, may induce periodic changes in electrode contact. In supine subjects, these are

mainly located on both sides of the dependent part of the chest (34). At low rates of

pulsation, the interference may not be easily identified, nevertheless, it may impact the

interpretation of EIT findings, especially when end-expiratory lung volume is assessed in

mechanically ventilated patients.

Document preparation

The first draft of this online document was prepared by I. Frerichs with collaboration of

A. Adler and V. Torsani. It was reviewed and approved by all other authors and

collaborators.

10

References

1. Rigaud B, Morucci JP. Bioelectrical impedance techniques in medicine. Part iii:

Impedance imaging. First section: General concepts and hardware. Crit Rev Biomed

Eng 1996;24:467-597.

2. Vonk Noordegraaf A, Faes TJ, Marcus JT, Janse A, Heethaar RM, Postmus PE, de Vries

PM. Improvement of cardiac imaging in electrical impedance tomography by means of

a new electrode configuration. Physiol Meas 1996;17:179-188.

3. Reifferscheid F, Elke G, Pulletz S, Gawelczyk B, Lautenschlager I, Steinfath M, Weiler N,

Frerichs I. Regional ventilation distribution determined by electrical impedance

tomography: Reproducibility and effects of posture and chest plane. Respirology

2011;16:523-531.

4. Bikker IG, Preis C, Egal M, Bakker J, Gommers D. Electrical impedance tomography

measured at two thoracic levels can visualize the ventilation distribution changes at the

bedside during a decremental positive end-expiratory lung pressure trial. Critical Care

2011;15:R193.

5. Frerichs I, Hahn G, Hellige G. Thoracic electrical impedance tomographic measurements

during volume controlled ventilation-effects of tidal volume and positive end-expiratory

pressure. IEEE Trans Med Imaging 1999;18:764-773.

6. Eyuboglu BM, Oner AF, Baysal U, Biber C, Keyf AI, Yilmaz U, Erdogan Y. Application of

electrical impedance tomography in diagnosis of emphysema--a clinical study. Physiol

Meas 1995;16:A191-211.

7. Metherall P, Barber DC, Smallwood RH, Brown BH. Three-dimensional electrical

impedance tomography. Nature 1996;380:509-512.

8. Barber DC. Quantification in impedance imaging. Clin Phys Physiol Meas 1990;11 Suppl

A:45-56.

9. Smit HJ, Vonk Noordegraaf A, Marcus JT, Boonstra A, de Vries PM, Postmus PE.

Determinants of pulmonary perfusion measured by electrical impedance tomography.

Eur J Appl Physiol 2004;92:45-49.

10. Vonk Noordegraaf A, Kunst PW, Janse A, Marcus JT, Postmus PE, Faes TJ, de Vries PM.

Pulmonary perfusion measured by means of electrical impedance tomography. Physiol

Meas 1998;19:263-273.

11. Miedema M, de Jongh FH, Frerichs I, van Veenendaal MB, van Kaam AH. Changes in

lung volume and ventilation during lung recruitment in high-frequency ventilated

preterm infants with respiratory distress syndrome. J Pediatr 2011;159:199-205 e192.

11


lung volume and ventilation during surfactant treatment in ventilated preterm infants.

Am J Respir Crit Care Med 2011;184:100-105.

13. Hough JL, Johnston L, Brauer S, Woodgate P, Schibler A. Effect of body position on

ventilation distribution in ventilated preterm infants. Pediatr Crit Care Med

2013;14:171-177.

14. Armstrong RK, Carlisle HR, Davis PG, Schibler A, Tingay DG. Distribution of tidal

ventilation during volume-targeted ventilation is variable and influenced by age in the

preterm lung. Intensive Care Med 2011;37:839-846.

15. Heinrich S, Schiffmann H, Frerichs A, Klockgether-Radke A, Frerichs I. Body and head

position effects on regional lung ventilation in infants: An electrical impedance

tomography study. Intensive Care Med 2006;32:1392-1398.

16. Seagar AD, Barber DC, Brown BH. Theoretical limits to sensitivity and resolution in

impedance imaging. Clin Phys Physiol Meas 1987;8 Suppl A:13-31.

17. Chatziioannidis I, Samaras T, Mitsiakos G, Karagianni P, Nikolaidis N. Assessment of lung

ventilation in infants with respiratory distress syndrome using electrical impedance

tomography. Hippokratia 2013;17:115-119.

18. Baisch FJ, Hahn G, Šipinková I, Beer M, Hellige G. Comparison of electrode belts with

"spot" electrodes for electrical impedance tomography. Innov Tech Biol Med

1995;16:119-125.

19. McAdams ET, Jossinet J, Lackermeier A, Risacher F. Factors affecting electrode-gel-skin

interface impedance in electrical impedance tomography. Med Biol Eng Comput

1996;34:397-408.

20. Hartinger AE, Guardo R, Adler A, Gagnon H. Real-time management of faulty electrodes

in electrical impedance tomography. IEEE Trans Biomed Eng 2009;56:369-377.

21. Adler A. Accounting for erroneous electrode data in electrical impedance tomography.

Physiol Meas 2004;25:227-238.

22. Adler A, Gaggero PO, Maimaitijiang Y. Adjacent stimulation and measurement patterns

considered harmful. Physiol Meas 2011;32:731-744.

23. Coulombe N, Gagnon H, Marquis F, Skrobik Y, Guardo R. A parametric model of the

relationship between eit and total lung volume. Physiol Meas 2005;26:401-411.

24. Frerichs I, Braun P, Dudykevych T, Hahn G, Genee D, Hellige G. Distribution of

ventilation in young and elderly adults determined by electrical impedance

tomography. Respir Physiol Neurobiol 2004;143:63-75.

12

25. Frerichs I, Dudykevych T, Hinz J, Bodenstein M, Hahn G, Hellige G. Gravity effects on

regional lung ventilation determined by functional EIT during parabolic flights. J Appl

Physiol 2001;91:39-50.

26. Frerichs I, Hahn G, Golisch W, Kurpitz M, Burchardi H, Hellige G. Monitoring

perioperative changes in distribution of pulmonary ventilation by functional electrical

impedance tomography. Acta Anaesthesiol Scand 1998;42:721-726.

27. Frerichs I, Hahn G, Schiffmann H, Berger C, Hellige G. Monitoring regional lung

ventilation by functional electrical impedance tomography during assisted ventilation.

Ann NY Acad Sci 1999;873:493-505.

28. Humphreys S, Pham TM, Stocker C, Schibler A. The effect of induction of anesthesia and

intubation on end-expiratory lung level and regional ventilation distribution in cardiac

children. Paediatr Anaesth 2011;21:887-893.

29. Blankman P, Hasan D, van Mourik MS, Gommers D. Ventilation distribution measured

with EIT at varying levels of pressure support and neurally adjusted ventilatory assist

in patients with ALI. Intensive Care Med 2013;39:1057-1062.

30. Mauri T, Bellani G, Confalonieri A, Tagliabue P, Turella M, Coppadoro A, Citerio G,

Patroniti N, Pesenti A. Topographic distribution of tidal ventilation in acute respiratory

distress syndrome: Effects of positive end-expiratory pressure and pressure support.

Crit Care Med 2013;41:1664-1673.

31. Yoshida T, Torsani V, Gomes S, De Santis RR, Beraldo MA, Costa EL, Tucci MR, Zin WA,

Kavanagh BP, Amato MB. Spontaneous effort causes occult pendelluft during

mechanical ventilation. Am J Respir Crit Care Med 2013;188:1420-1427.

32. Pulletz S, Elke G, Zick G, Schadler D, Reifferscheid F, Weiler N, Frerichs I. Effects of

restricted thoracic movement on the regional distribution of ventilation. Acta

Anaesthesiol Scand 2010;54:751-760.

33. Bodenstein M, Wang H, Boehme S, Vogt A, Kwiecien R, David M, Markstaller K.

Influence of crystalloid and colloid fluid infusion and blood withdrawal on pulmonary

bioimpedance in an animal model of mechanical ventilation. Physiol Meas

2012;33:1225-1236.

34. Frerichs I, Pulletz S, Elke G, Gawelczyk B, Frerichs A, Weiler N. Patient examinations

using electrical impedance tomography-sources of interference in the intensive care

unit. Physiol Meas 2011;32:L1-L10.








ONLINE SUPPLEMENT 2

EIT raw images

EIT raw

This ele

raw EIT

process

Figure

EOS 2 (

EIT raw

EIT dev

electrod

differen

one cyc

EIT data

EIT sca

The num

rate. Cu

scan ra

reported

changes

cycle. In

should

recomm

w images

ectronic onli

T data acqu

es involved

E2.1. Sequ

(gray backg

w data

vices intera

des. In a pr

ces are me

cle of curre

a frame con

an rate

mber of fra

urrent EIT d

ates of 100

d (1). A la

s. Figure E

n order to a

be larger t

mend scan r

s

ne supplem

uired during

in EIT exa

uence of pro

round) desc

ct with the

redefined se

easured at a

ent applicat

ntains the in

ames (or raw

devices offe

00 images/

arger scan

2.2 shows

avoid misre

han twice t

rates of at

ment (EOS)

g EIT chest

minations a

ocesses invo

cribes the p

e body usin

equence, cu

all other pai

ions and vo

nformation

w images)

er maximum

s for indus

rate perm

a time ser

epresentatio

the rate of

least 10 im

2 describes

t scanning.

and EIT data

olved in EIT

process of ra

ng an elect

urrent is ap

rs of electro

oltage mea

necessary t

acquired pe

m scan rate

strial proce

mits observ

ries of EIT

on of the u

f frequency

mages/s for

s how raw E

Figure E1.

a analysis fo

T chest exa

aw image g

trical curren

plied to pai

odes. The s

asurements

to generate

er second c

es of about

ess tomogra

vation of h

images ob

nderlying e

content of

r monitoring

EIT images

1 shows th

or orientatio

mination an

generation.

nt stimulati

rs of electro

set of EIT da

is typically

one raw im

corresponds

t 40 to 50

aphy EIT d

igher frequ

btained duri

vents (alias

f the most

g of breath

are generat

he sequenc

on.

nd data ana

on across

rodes, while

ata acquire

y called fram

mage.

s with the E

images/s, a

devices hav

uency phys

ing one res

sing), the s

rapid chan

hing and ve

2

ted from

es of all

alysis.

a set of

e voltage

d during

me. One

EIT scan

although

ve been

siological

spiratory

scan rate

ges. We

entilation

3

and 25 images/s for monitoring of cardiac related physiology. Even very high frequency

events such as high frequency oscillatory ventilation can be monitored by modern EIT

devices (2). Further discussion of scan rates and filtering is in EOS 3.

Figure E2.2. Example of EIT images acquired at a scan rate of 13 images/s (data from a

mechanically ventilated pig with experimental lung injury (3) acquired with the Goe-MF II

device (CareFusion, Höchberg, Germany)). EIT images frames (numbered) show subsequent

frames from left to right during a single respiratory cycle.

EIT tissue sensitivity

EIT is sensitive to changes in electrical conductivity properties of tissue at the stimulation

frequency. Conductivity is a bulk property of materials, and is measured in Siemens per

meter (S/m) with a range of 0.042 (inflated lung), 0.11 (deflated lung), 0.48 (heart) and

0.60 (blood) (4, 5). Resistivity (in Ω·m) is the inverse of conductivity. At low frequencies,

electrical current flows primarily through conductive pathways (in the body, regions of high

ionic conductivity). As the frequency increases, oscillating currents (also called displacement

currents) are able to travel across otherwise insulating barriers such as cell membranes. At a

specific current stimulation frequency the conduction properties of tissue are described by

the admittivity (S/m) and its inverse, the impedivity (Ω·m). To date EIT systems have

operated at relatively low frequencies (≤250 kHz), where current flow in tissue is mostly

conductive, and the resistivity and impedivity are roughly equal (which explains how these

terms are sometimes conflated). The measurement of tissue electrical properties at multiple

higher frequencies is called electrical impedance spectroscopy and serves a role in

characterizing properties of tissue samples (6-9). This method does not provide images, and

is typically not useful for real-time imaging, since its acquisition rates are relatively low

compared to EIT.

4

The conductivity properties of most tissues are relatively well known ex vivo (e.g.

(10)). The main factor in tissue is the fluid content, such that body fluids are conductive and

gases resistive. Since time-difference EIT (tdEIT) is a dynamic modality, it is sensitive to

changes in conductivity rather than the absolute conductivity level. Physiological changes

occur primarily due to the movements of body fluids (e.g. blood in blood vessels) and gases

(e.g. air in the lungs). For example, as the lungs inflate, electrical current must pass through

the parenchyma which is more widely spaced by the incoming air. Unfortunately, to our

knowledge, there are no good in vivo measurements of the lung conductivity as a function of

inflation; however, most models assume a linear relationship between resistivity and lung air

content (11). Changes in fluid content may be pulsatile (at the cardiac frequency) or more

slowly, due to intravascular or extravascular fluid accumulation.

The impedance values measured by EIT are very sensitive to the shape and

movement. Thus, the EIT signal in the heart region is a combination of the effect of blood

volume changes in the heart chambers, as well as the movement of the heart in the thoracic

cavity (12). EIT signals are similarly sensitive to changes in electrode position (13), body

posture (14-20), and electrode contact impedance (21, 22).

EIT position sensitivity

EIT is most sensitive to conductivity changes in the electrode plane, and has decreasing

sensitivity to off-plane effects. Note, however, that EIT does not provide a "slice" such as

from CT or MRI images. Instead, there is a gradual decrease in sensitivity with off-plane

distance. The decrease in sensitivity depends on the distance from the body surface. Close to

the skin, the off-plane sensitivity is low, while for conductivity changes near the thorax

centre, sensitivity is roughly half the chest width. Figure E2.3 explores this effect. Using a

finite element model (FEM), objects at different positions within the electrode plane are

simulated to move up- and downwards and the sensitivity calculated.

5

Figure E2.3. Frontal plane showing the relative vertical sensitivity of EIT measurements as

a function of the position above and below the electrode plane. Color intensity (black = 0)

corresponds to sensitivity. Contour lines indicate regions of equal sensitivity. From the

electrode plane lines indicate 95%, 90%, 75%, 50% and 25% sensitivity. Data were

calculated from a finite element mesh of the human thorax with an adjacent stimulation and

measurement pattern. In each vertical slice, sensitivity is normalized to its value on the

electrode plane.

EIT image reconstruction

Using the measured voltages, the process of generating an image is called "image

reconstruction". Typically, an image of the 2D slice through the electrode plane is

reconstructed. So far, the main image reconstruction framework used is tdEIT, used in all

clinical and the vast majority of experimental studies. tdEIT reconstructs an image of the

change in tissue properties between a baseline (or reference) measurement frame (Vref) and

the measurement frame at the current time (Vt). Vref is chosen to represent a physiologically

6

stable or relevant instant (see below). Since the meaning of an EIT image depends so

centrally on the Vref chosen, it is essential to describe its choice when reporting results.

Alternative image reconstruction frameworks

While all clinical results to date have been reconstructed using tdEIT, there are two other

frameworks for EIT image reconstruction which are active areas of research, and may see

experimental use in the near future, but are currently insufficiently robust for chest EIT:

frequency-difference EIT (fdEIT), which is sensitive to the difference in tissue properties

between stimulation frequencies. fdEIT uses special EIT hardware capable of stimulating

the body with electric current at two or more frequencies. Since the electrical

conductivity of tissues vary with frequencies (current may travel across cell membranes

rather than around it), the multifrequency measurements reflect these differences (23,

24). fdEIT uses the similar reconstruction algorithms to tdEIT, but images represent the

difference in tissue properties between frequencies rather than between Vref and the

current time. fdEIT has considerable promise, but has seen relatively less experimental

evaluation.

absolute imaging or absolute EIT (aEIT) calculates an image of the electrical tissue

properties at a given time. Thus it does not require a Vref, since each image represents

the tissue properties at the time of the measurement. aEIT is routine in the analogous

geophysical method; however, chest EIT presents some key challenges which have not

been satisfactorily overcome by current reconstruction algorithms. First, the regions of

interest (lungs) are relatively deep compared to the electrode spacing, and thus exhibit

relatively low sensitivity. Next, the electrode contact and placement in chest EIT are

subject to uncertainties due to sweat, contact quality, position inaccuracies and

movement due to breathing and posture change. Given these challenges, aEIT has not

yet seen significant clinical or experimental use.

Formulation of EIT image reconstruction

Time-difference EIT imaging

tdEIT imaging calculates an image of the impedance changes between two points in time,

and is the most widespread and robust mode of image reconstruction. Since many

physiological phenomena like lung ventilation and perfusion are periodic, their time-

dependent changes can thus be determined by tdEIT.

tdEIT reconstruction may be understood as calculating an inverse sensitivity function.

Given a tissue properties Pt at time t and properties Pref during the baseline measurement,

7

then difference EIT is concerned with the change in tissue properties ΔP. During these times,

corresponding measurements Mt and Mref are made, from which a change in measurements

ΔM is calculated (some reconstruction algorithms normalize this value). The sensitivity of the

EIT system, S (also called the Jacobian in the literature), relates

ΔM = S ΔP

where the matrix S describes how some EIT measurements are more sensitive to changes in

some image regions than others. In order to reconstruct an EIT image, we want to calculate

ΔP from ΔM, using a reconstruction matrix R such that

ΔP = R ΔM.

Here, R is a pseudo-inverse of S. Because of the low sensitivity of the diffusive propagation

of electrical current, EIT is an inverse problem and S is non-invertible. Instead, various

reconstruction algorithms calculate an approximate inverse; however, reconstruction

algorithms are required to choose a compromise between image features such as resolution

and the ability to suppress noise.

Thus, tdEIT calculates an estimate of the change in tissue properties

ΔP = Pt − Pref

between the current and the baseline frame. Correct interpretation of these images depends

on the definition of the baseline frame.

Development of new algorithms is an active field in EIT research with advances in

image quality achieved by using, e.g. a-priori anatomical information. For an introduction to

these algorithm developments, see (25-27).

Frequency- vs. time-difference imaging

fdEIT imaging calculates an image of the change in tissue properties between measurements

made at the same time, but at two different electrical stimulation frequencies.

Mathematically, fdEIT and tdEIT are formulated similarly, except that

ΔP = Pa − Pb,

ΔM = Ma − Mb

where a and b represent the two stimulation frequencies. Voltages are normally scaled and

normalized in practice. fdEIT systems can be constructed to make the electrical

measurements at exactly the same times, or in rapid succession. The frequencies may be

chosen such that an electrical dispersion occurs between them. For example, at the higher

frequency, current may pass through cell walls and through the cells, while this may not

occur at the lower frequency. fdEIT images can thus be interpreted as an image of the

distribution of tissue types whose electrical properties change most between the frequencies.

8

Absolute vs. time-difference imaging

aEIT seeks to calculate an image of the actual conductivity distribution within a body based

on a single set of measurements. aEIT thus does not require a baseline measurement frame.

Several groups are actively working on aEIT reconstruction algorithms, but it is not

sufficiently robust for experimental or clinical use. aEIT approaches are challenging in

medical applications because: first, EIT is highly sensitive to exact knowledge of the

boundary shape and electrode position, which is not easily available on the chest (which

moves with breathing and posture change) (28). Also, the higher electrical simulation

frequencies used in medical EIT are more susceptible to various types of electronic

inaccuracies. Additionally, aEIT is inherently non-linear, and its solutions require far more

computation than those of linear tdEIT.

Image reconstruction requirements

As indicated the mathematics of developing an EIT reconstruction requires performing

compromises in the selection of a pseudo-inverse. In order to select between the many

proposed algorithms, work has been done to identify figures of merit to evaluate algorithm

performance (25, 29, 30). Such work allows selection of appropriate image reconstruction to

best match an application requirement.

Based on the consensus reported in (25), Figure E2.4 illustrates the developed figures of

merit; an ideal reconstruction algorithm should exhibit (in order of importance):

uniform amplitude response (i.e. a small contrast of given size should result in the same

average image amplitude at any position in the imaging plane)

small and uniform position error (i.e. a small contrast should be reconstructed in the

correct position)

high and uniform resolution (i.e. a small contrast should be reconstructed as a small

region, in order to allow distinguishing of nearby changes; however, resolution should

be uniform to avoid misinterpretation of contrasts in different regions)

small ringing artifacts (where "ringing" means image regions with a change inverse to

the underlying tissue properties)

low noise amplification (i.e. the effect of electronic noise in measurements should be

minimized in the reconstructed images)

low shape deformation (i.e. the shape of regions should be preserved − except for

blurring)

small sensitivity to electrode and boundary movement.

Figure

position

correspo

1) the

distance

Shape d

and 5) t

Impact

The ma

inspired

(31). N

enginee

reconstr

I

shown.

circular

for the

describe

due to

E2.4. Figu

n (simulated

onding to im

amplitude

e from the s

deformation

the Ringing

t of image

athematical

d and challe

ovel and in

ering and p

ruction algo

In Figure E

In order t

shape assu

First (Figure

first EIT d

ed by (32,

its formulat

res of merit

d or measu

mage regio

is the sum

source cent

n is the ROI

is the relat

e reconstru

problems

enged math

nteresting

physics litera

orithms rele

E2.5, a sam

to compare

umption for

e E2.5, top

device, the

33). SBP h

tion − drive

t for EIT rec

red), an im

ns above 2

m of all im

tre to ROI c

I area outsid

tive amplitu

uction on i

underlying

ematicians

reconstruct

ature. We w

evant also fo

mple of tdE

e algorithm

all models.

left) is the

e Sheffield

has been ve

en by the li

construction

mage is reco

5% maximu

age pixel a

centre, 3) th

de of region

ude of the in

mage app

g EIT imag

and compu

tion approa

would like

or medical u

IT common

ms fairly, th

Sheffield b

Mark I in

ery widely u

mitations o

ns (25). Fro

onstructed (

um amplitu

amplitudes,

he resolutio

n in a circle

nverted area

pearance

ge reconst

utational sci

ches are e

to address

use in chest

n reconstru

his figure w

backprojectio

the mid 8

used; howe

of computer

om a source

(left) and t

de (center)

2) the po

n is the size

e of the sam

a of the ima

ruction are

entists for a

explored in

here a few

t EIT imagin

ction algori

was calculat

on algorithm

0´s of the

ever, it has

r hardware

e contrast o

the ROI det

). Using the

osition erro

e of the RO

me area as t

age.

e subtle an

a third of a

the mathe

w aspects o

ng.

ithm appro

ted by imp

m (SBP), de

e last centu

s some inac

in the early

9

of known

termined

ese data:

or is the

OI, 4) the

the ROI,

nd have

century

ematical,

of image

oaches is

posing a

eveloped

ury, and

ccuracies

y 1980s.

10

Specifically, it is limited by the use of an inadequate, 2D sensitivity model. From the point of

view of interpretation, SBP can display streak-like artefacts, which point toward the

boundary. Another limitation is the assumption of a circular geometry for the electrodes and

the body. Nonetheless, in spite of these limitations, SBP works surprisingly well (30) which

allows the findings of a great number of EIT studies to be considered valid.

Next, (Figure E2.5, top right) is the GREIT algorithm (25) developed by a consensus

group of EIT experts. GREIT is based on a mathematical optimization of reconstructed

images against a set of defined figure of merit parameters.

Next, (Figure E2.5, bottom left) is a regularized linear Gauss-Newton (GN)

reconstruction. The image is calculated to match the measured data and a "regularization"

(or smoothness) constraint. The choice of regularization constraint is a rich research area;

GN algorithms allow a trade-off between noise immunity, image resolution, and fidelity to

various image shape constraints (34, 35).

Last, (Figure E2.5, bottom right) is an example of a non-linear tdEIT algorithm. This

specific image uses a Total Variation (TV) constraint (36) which enhances edges between

regions in the image. TV approaches do not show the blur inherent with linear algorithms,

but have been criticised as showing the appearance of greater resolution than is actually

available from the data (36).

Comparing image reconstruction algorithms

Each EIT image reconstruction algorithms is best understood as a framework with many

"ingredients". Each ingredient can be varied, and will have an effect on the accuracy and

reliability of the calculated images. In the following subsections, we consider the effects of

four algorithm parameters: shape, orientation, background, and noise performance.

11

Figure E2.5. Examples of image reconstruction algorithms applied to an identical EIT data

set (3). BP, backprojection; GREIT, Graz consensus; GN, Gauss-Newton; TV, Total Variation.

Image reconstruction shape

The earliest EIT images assumed that electrodes were placed on a circular plane on a

cylindrical body. These round images were obtained using the so-far most widespread image

reconstruction procedure, SBP. The computational advances in the 1990s made it relatively

straightforward to model the contour of the anatomical form of the chest. Figure E2.6

illustrates the effect of the choice of anatomical shape; an image of tidal breathing is

reconstructed on four different models. Images illustrate that the ability of EIT to separate

the lung regions is compromised when the model is clearly wrong (ellipse and adult human).

Undoubtedly, the best selection is of a model shape which exactly matches the

subject. However, in many cases, this is not available, because a scan of the subject has not

been performed, and also because thorax shape changes with posture. In these cases the

best shape match should be chosen. A study of the required accuracy of shape matching

12

(37) suggested that an area mismatch of less than 4% corresponded to accurate

reconstructed images.

Figure E2.6. Algorithm shapes. EIT data obtained from a lung-healthy mechanically

ventilated piglet (3) representing tidal ventilation (end-inspiration − end-expiration) were

reconstructed using the GREIT algorithm with various assumed thorax shapes. 1: Cylindrical

thorax (top left) 2: Elliptical thorax (top right) 3: Porcine shape thorax (Generic) (bottom

left) 4: Adult human shape thorax (bottom right). The porcine shape was not specifically

adapted to the thorax shape of the specific animal. Note the relative deformations of the

reconstructed form.

Image reconstruction noise performance

EIT algorithms implicitly make assumptions about the level of random noise (interference) in

the raw data. When the noise level is assumed to be high, an algorithm will smooth (or blur)

the image to supress such noise. However, if data have low levels of noise, it is appropriate

to let the algorithm reduce blurring which allows better spatial separation of image regions.

One common way to measure the assumed noise level of the algorithm is using the Noise

13

Figure parameter (34). In several papers, the value of a parameter or "hyperparameter" is

adjusted to achieve a similar effect. Using the porcine shape Figure E2.7 shows the same

data as in Figure E2.6 reconstructed with the GREIT algorithm with different levels of

assumed noise.

Figure E2.7. Smoothing and noise suppression in image reconstruction. EIT data obtained

from a lung-healthy mechanically ventilated piglet (3) representing tidal ventilation (end-

inspiration − end-expiration) were reconstructed using the GREIT algorithm with various

assumed noise levels. Images are shown corresponding to an algorithm NF (Noise Figure) of

0.25, 0.5, 1.0 and 2.0. At higher NF values, there is reduced spatial blurring and regions

become more separate. The top right image is the same as the one shown in Figure E2.6.

Image reconstruction background

As indicated, tdEIT algorithms reconstruct a change in impedance distribution, between a

measurement, Mt and a baseline measurement, Mref. Reconstruction then uses a sensitivity

matrix, S, calculated at the assumed conditions during Mref. The assumed impedance

distribution at Mref, is typically called the "background" distribution. Almost all tdEIT

algorithms have made the simplest assumption, that the background is homogeneous;

however, for the thorax, this assumption is clearly false. Some recent algorithm work

14

considers non-uniform background conductivities. In these images, the contrast in the lung

region is enhanced because it accounts for the inherent lower sensitivity in low-conductivity

lungs. Figure E2.8 shows images of tidal breathing with different values of the assumed

background conductivity of the lungs with respect to the average of other thoracic tissues.

Figure E2.8. Effect of image reconstruction background distribution. EIT data obtained from

a lung-healthy mechanically ventilated piglet (3) representing tidal ventilation (end-

inspiration − end-expiration) were reconstructed using the GREIT algorithm with models of

the lung region conductivity with respect to the average of other thoracic tissues.

Pixel vs. finite element grid

EIT images have typically been represented in two ways, as a pixel grid or on a triangular

representation (Figure E2.9). A representation on a pixel grid means that all image elements

are the same size, and thus the calculation of functional parameters do not need to scale

image elements by size. A representation on a triangular grid is common when the

underlying algorithm uses a finite element mesh. In this case, image elements are typically

of different sizes, but can often approximate the boundary shape more accurately. We

recommend the use of pixel grid; since EIT is a functional modality, the ease of representing

functional parameters is more useful than the ability to represent the boundary shape.

15

Figure E2.9. Representation of EIT images. Left: EIT reconstructed image parameters

represent pixels in a matrix (green); Right: EIT reconstructed image parameters represent

triangular regions based on a finite element mesh (green).

Colour mapping

The color coding of EIT images is not unified. Different color representations have been used

by various research groups and vendors of EIT technology. Perhaps the most common are

the grey and "jet" color maps but many other representations have been used. Figure E2.10

shows some commonly used color mappings.

We identify specific issues with the choice of EIT color coding. First, for difference

images, there is a zero, which represents no change in the images. This zero has a clear

representation in some image color codings. Positive and negative changes from this zero

are then defined. However, some other color schemes do not define the zero color and only

the color limits representing the maximum and minimum image values are provided.

Two choices have been made to modify the visual strength of these conductivity

changes. First, the choice of colors may modulate the appearance of positive and negative

changes, and, second, the use of a uniform region which compresses all changes less than a

certain level to be represented as the zero color. In the absence of consensus, we

recommend that color maps be defined and clarified for each use.

16

Figure E2.10. Common color coding schemes used in EIT lung images, based on academic

publications (top row) and EIT system vendors (bottom row). The representation of

impedance increase (Z↑ or +Z) and impedance decrease (Z↓ or –Z) are described.

Top left: Blue-white-red, showing Z↑ as blue, Z↓ as red, and zero as white.

Top middle: Gray, representing Z↑ as white, and Z↓ as black. This color scheme does not

define a fixed color value for no change (zero).

Top right: Jet, defined from the definition in Matlab (Mathworks, Nantick, MA, USA).

representing Z↑ as red, and Z↓ as blue. This color scheme does not define a fixed color value

for no change (zero).

Bottom left: Draeger, representing Z↑ blue-white, Z↓ as purple, and zero as black with a

uniform color region around zero.

Bottom middle: Swisstom, representing Z↑ as blue-white, and zero as blue-grey with a

uniform color region at and above zero. Colors for Z↓ are not defined.

Bottom right: Timpel, representing Z↑ as blue-white, and zero as grey black with a uniform

color region at and above zero. Colors for Z↓ are not defined.

(The raw EIT data used to generate the images with different color coding in this figure were

acquired in a healthy human subject during tidal breathing and originate from (38).)

17

Image reconstruction orientation

The orientation of EIT images is identical with the images generated by established medical

imaging modalities like computed tomography with the subject right side on the left side of

the image and with anterior at the top of the image. In much of the older EIT literature

(before 2000), it was common to represent the EIT image horizontally reversed, that is with

the subject anterior shown at the bottom of the image (Figure E2.11).

Figure E2.11. Image orientation. Modern EIT systems use a medical image orientation

(left) while some older systems were vertically flipped (right). (Data is from (3)).

Selection of tdEIT baseline measurement

As mentioned, tdEIT reconstructs a change in volume distribution between a baseline

measurement and a current measurement. Therefore, interpretation of the image requires

the knowledge of what is "happening" in terms of physiology at both measurement points.

While the "current" measurement normally represents a point in time, the "baseline"

measurement represents either another single point in time or the mean value of a period of

time. It is thus essential to specify how the baseline measurement is selected. Various

strategies have been developed to select a baseline measurement; however, a small number

of approaches are most commonly used, and are discussed here.

Baseline as a mean of EIT data

The most common strategy is to simply use the mean of all measurements as the baseline

image. This approach has two benefits. It is simple and robust. It can be applied to all types

of breathing patterns, and is less affected by sighs, coughs, other "unusual" breathing

patterns, or signal disturbances. However, the main disadvantage is that images will be

represented as both positive and negative. It is generally more difficult to visually interpret

these images, as the "zero" image does not correspond to any particularly physiologically

18

significant time (typically it will represent lung midcapacity above end-expiratory lung

volume). Another disadvantage is that the global mean is only available after the complete

recording. This strategy is thus only suitable for post-processing of data. Figure E2.12 shows

how the appearance of EIT images is influenced when mean EIT measurement data are

used as baseline.

Figure E2.12. EIT images and the global EIT waveform for the baseline based on the mean

value of acquired EIT data. Top row, images (using a blue-black-red color coding)

corresponding to the indicated times (dashed lines). Bottom row: average image value vs.

time for the shown three breaths (data from (3)).

Baseline at end-expiration or end-inspiration

Another common strategy is to select a physiologically meaningful event as the baseline

scan. This is typically the end of expiration or inspiration. The advantage is that images then

have a more natural physiological interpretation (Figures E2.13 and E2.14). The chosen

baseline scan may be either a single instant, or an average of all end-expiration in the

recording. The baseline may be chosen manually or with automated software.

One disadvantage with this strategy is that "unusual" breathing patterns, such

coughs or sighs or other signal disturbances, can make identification of a true physiological

event difficult. Additionally, this strategy is most suitable for post-processing of data.

19

Figure E2.13. EIT images and the global EIT waveform for the "end-expiratory" baseline

measurement. Top row, images (using a blue-black-red color coding) corresponding to the

indicated times (dashed lines). Bottom row: average image value vs. time for the shown

three breaths (data from (3)).

Figure E2.14. EIT images and the global EIT waveform for the "end-inspiratory" baseline

measurement. Top row, images (using a blue-black-red color coding) corresponding to the

indicated times (dashed lines). Bottom row: average image value vs. time for the shown

three breaths (data from (3)).

20

Baseline from moving averages

In cases where it is essential to analyse EIT data on-line, a dynamic selection of the baseline

scan is required. Clearly, such a strategy must use only previous data, and must thus be

updated dynamically. In this case, the most natural extension of the mean scan baseline

strategy is a dynamic mean. As an example, in Figure E2.15, the mean of the previous 2 s of

data is used. However, in practice a longer interval is used. The dynamic mean has many of

the advantages and disadvantages of the mean scan. Additionally, it can show temporary

differences when the patient state changes. For example, when PEEP or patient posture

changes, the dynamic mean will slowly adjust to the new level at which point it will "forget"

the previous state.

Figure E2.15. EIT images and the global EIT waveform for the "dynamic mean" baseline

measurement, in which the mean is based on the last 2s of the signal. Top row, images

(using a blue-black-red color coding) corresponding to the indicated times (dashed lines).

Bottom row: average image value vs. time for the shown three breaths (data from (3)).

The natural extension of the end-expiratory strategy for on-line analysis is the

dynamic end-expiratory baseline. In this case, the average of the previous N end-expirations

is chosen (Figure E2.16). This strategy has the advantages of the dynamic mean scan, but

also has a more natural physiological interpretation since the baseline scan corresponds to a

natural physiological interpretation.

21

Figure E2.16. EIT images and the global EIT waveform for the "dynamic end-expiratory"

baseline measurement, in which the baseline is based on the last three end-expiratory

events. Top row, images (using a blue-black-red color coding) corresponding to the indicated

times (dashed lines). Bottom row: average image value vs. time for the shown three breaths

(data from (3)).


This online document was prepared by A. Adler. It was reviewed and approved by all other

authors and collaborators.

22

References

1. Wilkinson AJ, Randall EW, Cilliers JJ, Durrett DR, Naidoo T, Long T. A 1000-

measurement frames/second ert data capture system with real-time visualization. IEEE Sens

J 2005;5:300-307.

2. Frerichs I, Achtzehn U, Pechmann A, Pulletz S, Schmidt EW, Quintel M, Weiler N.

High-frequency oscillatory ventilation in patients with acute exacerbation of chronic

obstructive pulmonary disease. J Crit Care 2012;27:172-181.

3. Dargaville PA, Rimensberger PC, Frerichs I. Regional tidal ventilation and compliance

during a stepwise vital capacity manoeuvre. Intensive Care Med 2010;36:1953-1961.

4. Geddes LA, Baker LE. The specific resistance of biological material-a compendium of

data for the biomedical engineer and physiologist. Med Biol Eng 1967;5:271-293.

5. Faes TJ, van der Meij HA, de Munck JC, Heethaar RM. The electric resistivity of

human tissues (100 Hz-10 mHz): A meta-analysis of review studies. Physiol Meas

1999;20:R1-10.

6. Halter RJ, Hartov A, Heaney JA, Paulsen KD, Schned AR. Electrical impedance

spectroscopy of the human prostate. IEEE Trans Biomed Eng 2007;54:1321-1327.

7. Laufer S, Solomon SB, Rubinsky B. Tissue characterization using electrical

impedance spectroscopy data: A linear algebra approach. Physiol Meas 2012;33:997-1013.

8. Mohr P, Birgersson U, Berking C, Henderson C, Trefzer U, Kemeny L, Sunderkotter

C, Dirschka T, Motley R, Frohm-Nilsson M, Reinhold U, Loquai C, Braun R, Nyberg F, Paoli

J. Electrical impedance spectroscopy as a potential adjunct diagnostic tool for cutaneous

melanoma. Skin Res Technol 2013;19:75-83.

9. Murdoch C, Brown BH, Hearnden V, Speight PM, D'Apice K, Hegarty AM, Tidy JA,

Healey TJ, Highfield PE, Thornhill MH. Use of electrical impedance spectroscopy to detect

malignant and potentially malignant oral lesions. Int J Nanomedicine 2014;9:4521-4532.

10. Gabriel C, Peyman A, Grant EH. Electrical conductivity of tissue at frequencies below

1 mHz. Phys Med Biol 2009;54:4863-4878.

11. Nopp P, Harris ND, Zhao TX, Brown BH. Model for the dielectric properties of human

lung tissue against frequency and air content. Med Biol Eng Comput 1997;35:695-702.

12. Proenca M, Braun F, Rapin M, Sola J, Adler A, Grychtol B, Bohm SH, Lemay M,

Thiran JP. Influence of heart motion on cardiac output estimation by means of electrical

impedance tomography: A case study. Physiol Meas 2015;36:1075-1091.

13. Adler A, Guardo R, Berthiaume Y. Impedance imaging of lung ventilation: Do we

need to account for chest expansion? IEEE Trans Biomed Eng 1996;43:414-420.

14. Coulombe N, Gagnon H, Marquis F, Skrobik Y, Guardo R. A parametric model of the

relationship between EIT and total lung volume. Physiol Meas 2005;26:401-411.

23

15. Frerichs I, Braun P, Dudykevych T, Hahn G, Genee D, Hellige G. Distribution of

ventilation in young and elderly adults determined by electrical impedance tomography.

Respir Physiol Neurobiol 2004;143:63-75.

16. Frerichs I, Dudykevych T, Hinz J, Bodenstein M, Hahn G, Hellige G. Gravity effects

on regional lung ventilation determined by functional EIT during parabolic flights. J Appl

Physiol 2001;91:39-50.

17. Heinrich S, Schiffmann H, Frerichs A, Klockgether-Radke A, Frerichs I. Body and

head position effects on regional lung ventilation in infants: An electrical impedance


18. Lupton-Smith AR, Argent AC, Rimensberger PC, Morrow BM. Challenging a

paradigm: Positional changes in ventilation distribution are highly variable in healthy infants

and children. Pediatr Pulmonol 2014;49:764-771.

19. Reifferscheid F, Elke G, Pulletz S, Gawelczyk B, Lautenschlager I, Steinfath M,

Weiler N, Frerichs I. Regional ventilation distribution determined by electrical impedance

tomography: Reproducibility and effects of posture and chest plane. Respirology

2011;16:523-531.

20. Riedel T, Richards T, Schibler A. The value of electrical impedance tomography in

assessing the effect of body position and positive airway pressures on regional lung

ventilation in spontaneously breathing subjects. Intensive Care Med 2005;31:1522-1528.


using electrical impedance tomography-sources of interference in the intensive care unit.

Physiol Meas 2011;32:L1-L10.

22. Boyle A, Adler A. The impact of electrode area, contact impedance and boundary

shape on eit images. Physiol Meas 2011;32:745-754.

23. Brown BH, Barber DC, Wang W, Lu L, Leathard AD, Smallwood RH, Hampshire AR,

Mackay R, Hatzigalanis K. Multi-frequency imaging and modelling of respiratory related

electrical impedance changes. Physiol Meas 1994;15 Suppl 2a:A1-12.

24. Smallwood RH, Hampshire AR, Brown BH, Primhak RA, Marven S, Nopp P. A

comparison of neonatal and adult lung impedances derived from EIT images. Physiol Meas

1999;20:401-413.

25. Adler A, Arnold JH, Bayford R, Borsic A, Brown B, Dixon P, Faes TJ, Frerichs I,

Gagnon H, Garber Y, Grychtol B, Hahn G, Lionheart WR, Malik A, Patterson RP, Stocks J,

Tizzard A, Weiler N, Wolf GK. Greit: A unified approach to 2D linear EIT reconstruction of

lung images. Physiol Meas 2009;30:S35-55.

26. Lionheart WR. Eit reconstruction algorithms: Pitfalls, challenges and recent

developments. Physiol Meas 2004;25:125-142.

24

27. Bayford RH. Bioimpedance tomography (electrical impedance tomography). Annu

Rev Biomed Eng 2006;8:63-91.

28. Denai MA, Mahfouf M, Mohamad-Samuri S, Panoutsos G, Brown BH, Mills GH.

Absolute electrical impedance tomography (aEIT) guided ventilation therapy in critical care

patients: Simulations and future trends. IEEE Trans Inf Technol Biomed 2010;14:641-649.

29. Gagnon H, Grychtol B, Adler A. A comparison framework for temporal image

reconstructions in electrical impedance tomography. Physiol Meas 2015;36:1093-1107.

30. Grychtol B, Elke G, Meybohm P, Weiler N, Frerichs I, Adler A. Functional validation

and comparison framework for EIT lung imaging. PLoS One 2014;9:e103045.

31. Calderon AP. On an inverse boundary value problem. Comput Appl Math

2006;25:133-138.

32. Barber DC, Seagar AD. Fast reconstruction of resistance images. Clin Phys Physiol

Meas 1987;8 Suppl A:47-54.

33. Santosa F, Vogelius M. A backprojection algorithm for electrical-impedance imaging.

Siam J Appl Math 1990;50:216-243.

34. Adler A, Guardo R. Electrical impedance tomography: Regularized imaging and

contrast detection. IEEE Trans Med Imaging 1996;15:170-179.

35. Borsic A, Lionheart WRB, McLeod CN. Generation of anisotropic-smoothness

regularization filters for eit. IEEE Trans Med Imaging 2002;21:579-587.

36. Borsic A, Graham BM, Adler A, Lionheart WRB. In vivo impedance imaging with total

variation regularization. IEEE Trans Med Imaging 2010;29:44-54.

37. Grychtol B, Lionheart W, Wolf G, Bodenstein M, Adler A. Impact of model shape

mismatch on reconstruction quality in electricalimpedance tomography. IEEE Trans Med

Imaging 2012; 3:1754-1760.

38. Guardo R, Boulay C, Murray B, Bertrand M. An experimental study in electrical

impedance tomography using backprojection reconstruction. IEEE Trans Biomed Eng

1991;38:617-627.








ONLINE SUPPLEMENT 3

EIT waveforms and regions-of-interest

EIT wa

This ele

raw EIT

The mo

phenom

wavefor

wavefor

and EIT

Figure

EOS 3

Abbrevi

of-inter

ventilat

EIT wa

An EIT

generat

global o

impeda

or aver

from th

origins

spontan

aveforms

ectronic onli

T images an

ost commo

mena are ex

rms. Finally

rms. Figure

T data analy

E3.1. Seq

(gray back

iations: EOS

est; CoV, c

tion delay; R

aveforms

T waveform

ted online o

or regional

nce change

rages of rel

e smallest “

EIT wavefo

(Figure E3

neous venti

s and regi

ine supplem

nd how they

n regions o

xplained, fo

y, we addre

e E3.1 show

ysis and the

uence of p

kground) d

S, electroni

center of ve

R, right; L, l

m is a sequ

or offline, f

. The glob

es of all ima

ative imped

“ROI”, a sin

orms consi

3.2). Period

ilation and

ons-of-in

ment (EOS 3

y are affect

of interest

llowed by t

ess the issu

ws the sequ

e place of th

rocesses in

describes th

c online su

entilation; U

left.

uence of i

from a time

al impedan

age pixels,

dance chan

ngle image p

st of differ

dic (or qua

cardiac ac

nterest

3) describes

ted by the m

(ROI) used

the filtering

ues of nois

uences of al

his EOS in t

nvolved in E

he EIT wav

upplement;

U/L, upper-

mpedance

e series of

nce wavefo

while regio

ges within

pixel.

rent period

asiperiodic)

ction, as w

s how EIT w

main physio

d to quant

procedures

e and othe

ll processes

he sequenc

EIT chest ex

veforms an

tdEIT, time

-to-lower ve

change va

raw EIT im

rm is base

onal impeda

defined RO

dic and non

signals ar

well as inter

waveforms a

ological proc

titatively an

s useful for

er interferen

s involved in

ce.

xamination

nd regions

e-difference

entilation ra

alues as a

mages. EIT

d on the s

ance wavefo

OIs. Pixel w

n-periodic s

e related t

rferences fr

are generat

cesses in th

nalyze EIT

the analysi

nces impact

n EIT exam

and data a

of interest

e EIT; ROI,

atio; RVD,

function o

waveforms

sum or ave

forms are th

waveforms o

signals of

to mechan

rom some

2

ted from

he chest.

regional

is of EIT

ting EIT

minations

analysis.

t (ROI).

region-

regional

of time,

s can be

erage of

he sums

originate

multiple

ical and

medical

devices

may be

from in

baseline

Figure

periodic

mechan

500 dev

are larg

inspirat

three d

relative

Region

A regio

purpose

regiona

wavefor

various

, such as p

e caused by

terferences

e changes d

E3.2. Exa

c signals d

nically venti

vice (Dräge

ger than th

tory peaks o

ecremental

impedance

ns of intere

on of intere

e. In EIT im

al physiologi

rms for all

sources,

pulsating air

y spontaneo

s such as b

due to EIT h

ample of a

uring a de

ilated patie

er Medical, L

ose induce

of the EIT s

PEEP level

e change.

est

est (ROI)

mages, ROIs

ical aspects

pixels in th

it is neces

r suspensio

ous or vent

body movem

hardware dr

a global im

ecremental

ent with AR

Lübeck, Ger

d by cardia

signal. Dash

ls highlight

is a selecte

s have been

s. The EIT w

e ROI. Sinc

ssary to iso

on mattresse

tilator-induc

ments, chan

rift.

mpedance w

positive en

RDS. (The d

rmany).) Im

ac activity.

hed lines at

the stepwi

ed subset

n chosen to

waveform in

ce EIT wave

olate the d

es and pac

ced ventilat

nges of con

waveform c

nd-expirator

data was a

mpedance c

The latter

end-expira

se fall in en

of an ima

o group the

n a ROI is th

eforms cont

different co

emakers. N

tion maneu

ntact imped

containing

ry pressure

cquired usi

changes ind

can be disc

tory lung im

nd-expirato

ge identifie

image pixe

he sum or a

tain regiona

omponents

Non-periodic

uvers, or ca

dance over

periodic an

e (PEEP) tr

ing the Pul

duced by ve

cerned at t

mpedance v

ory volume.

ed for a p

els which re

average of t

al informati

of EIT sig

3

c signals

an result

time, or

nd non-

rial in a

lmoVista

entilation

the end-

values at

rel. ΔZ,

articular

epresent

the pixel

on from

gnals to

4

quantitatively analyze the images using relevant EIT measures (EOS 5). By the definition of

the ROI, image pixels are chosen to reflect the regional changes associated with relevant

physiological or pathological effects. The other approach to isolate regional information from

the images is filtering, discussed in the section “Frequency filtering” later in this EOS.

In the most common case, ROIs are defined to assess regional ventilation. For

example, ROIs of horizontal layers are appropriate for monitoring ventilation-related

pathology which affects both lungs equally, and for assessing dependent vs. non-dependent

lung behavior. ROIs that separate the left and right lungs are suitable for monitoring

pathology likely to affect each lung differently. ROIs can also be defined to assess cardiac-

related impedance changes. The anatomical location of breathing and heart beat-related

impedance changes has been validated to correspond to the correct anatomical locations,

when using anatomically accurate reconstruction algorithms (1).

Geometrical ROIs

The simplest ROIs are defined geometrically as equal horizontal and/or vertical divisions of

the image region, as illustrated in figure E3.3 (top row). Such regions are defined by

horizontal and vertical slices which divide the image region into equally sized rectangles.

In cases where the physiology of interest is distributed in the anteroposterior

direction, ROIs divide the image region with horizontal divisions. The minimum is a division

into two regions, the anterior and the posterior (i.e., the upper and the lower in supine

subjects). This approach has been used to calculate the ratio of anterior-to-posterior (i.e.,

upper-to-lower ventilation ratio) (2, 3), a measure originally called the "impedance ratio" (4)

(see also EOS 5). The image region may be further divided into smaller ROIs if necessary

(5-7). The maximum number of horizontal ROIs is the image resolution, where each

horizontal ROI is a pixel height (typically 32 pixels for most common reconstruction

algorithms) (8-10). This maximum vertical resolution increases the sensitivity of EIT data

analysis for ventilation distribution monitoring (11). In figure E3.3, a division into two regions

and into a larger numbers of regions is shown.

A further division of ROIs is to consider the left/right division between image regions

(3, 12). Another common type of ROIs is generated by dividing the image regions into four

quadrants (13).

There are a few drawbacks of horizontal ROIs defined on the global image. First,

given the variability in thorax shape and the position of lungs within it, it is common that the

most dependent ROIs contain only a small or no lung region. This is especially likely in

obese patients where there is a larger layer of tissue surrounding the lungs. In this case, no

5

ventilation signal will be present in these dependent ROIs. The low signals in dependent

ROIs can also be caused by lung collapse in these regions. One way to address this issue is

to define the geometrical ROIs not within the whole image but only in a ROI representing

the lungs (see the section “Lung ROIs” below). In this case, the lung ROI is first detected,

and then horizontal or vertical geometrical divisions are identified in that region (14). In

Figure E3.3, the top row shows the global image region and its subdivisions, while the

bottom row shows subdivisions of the lung ROI.

Figure E3.3. Different types of ROIs used to characterize the distribution of ventilation. The

top row shows ROIs for the global image, while the bottom row illustrates ROIs for the

division of the lung region only. From left to right: 1) global region, 2) anterior (upper) and

posterior (lower) regions, 3) image quadrants (anterior right, anterior left, posterior right,

posterior left), 4) multiple layers (six illustrated), and 5) multiple layers with left/right

division. (Areas with oblique lines are contained within the respective ROIs.)

Lung ROIs

As discussed above, some EIT image analysis methods require the identification of the lung

regions (14-19). Most of the lung ROI identification methods identify lungs as regions with

large values in functional EIT images (see also EOS 4). In these methods, the lung is the

region in which fEIT image values are above a threshold, defined as a fraction of the

maximum image value. Use of a smaller threshold increases the ROI size. It makes the

identification of lung regions more complete, however, some nearby non-lung regions are

detected as well. A larger threshold has the opposite effect. It better rejects the non-lung

regions, but it is less able to detect all lung regions. Pulletz et al. and Becher et al. analyzed

the influence of threshold settings for this purpose (11, 20). The definition of the ROI using

a threshold of 20–35% of the maximum pixel values was recommended. Higher threshold

values have been shown to obscure the differences in the degree of ventilation homogeneity

6

between ARDS patients and patients with healthy lungs (20). The impact of the type of

functional images used to generate the lung ROIs on the quantitative analysis of ventilation

distribution was also examined (21). Dynamic determination of threshold values is also

possible (22). However, there is no optimal threshold value that can separate lung regions

from other thoracic tissues.

Figure E3.4. EIT examination of an anesthetized supine patient ventilated at different end-

expiratory pressure (PEEP) values using the PulmoVista 500 device (Dräger Medical, Lübeck,

Germany). The two tidal images (top right) show the ventilation distribution at PEEP of 5

and 15 cmH2O, the differential image (top left) highlights the loss (orange) and gain (blue)

in regional ventilation between these two time points (cursor 1 and cursor 2). The lung ROIs

identified from the two tidal images are plotted as blue (PEEP 5 cmH2O) and red (PEEP

15 cmH2O) lung contours at the right top edge of the figure. Note that the ventral and

dorsal boundaries at PEEP 5 are positioned above the corresponding PEEP 15 boundaries.

A further issue that needs to be taken into account when lung ROI is defined is that

the ventilation distribution depends on the ventilator setting, for instance the end-expiratory

pressure (PEEP). At low PEEP, ventilation occurs mainly in non-dependent regions, and at

7

high PEEP levels, ventilation shifts toward dependent regions (Figure E3.4). Thus, to

approximate the total lung area, it is useful to combine the regions identified at low and at

high PEEP. However, this solution is still unable to identify collapsed lung regions. To

identify such non-ventilated lung tissue, some other approaches have been developed.

Mirroring the lung areas from one side to the other will include collapsed regions which are

present in one lung but not the other (23). Since perfusion may still be present in collapsed

lung tissue, ROI definitions from the EIT perfusion signal may be added to the ventilation-

defined lung ROI (24). Given that lung size in healthy adults depends on the height (25) and

weight, (26) built a database of lung sizes and locations in EIT images from CT images and

correlated them with height and weight measurements. Using this database, it is possible to

estimate the lung ROI and thorax shape in an image from a patient’s height and weight.

The heart region is more difficult to identify than the lungs because the cardiac-

related signals are smaller, and there are cardiac-frequency contributions throughout the

image. This means that a simple frequency filter will identify a larger region than the heart.

Several more sophisticated techniques have been proposed (1, 16, 23, 27, 28). ROIs have

been identified with ECG gating (29, 30) or principal component analysis (24, 31), and

separate the heart area and the perfused lung tissue region. A common approach used by

many of these techniques is an initial identification of the lung ROI, and then the use of this

region to exclude areas which would otherwise be identified by cardiac-frequency filtering.

Examples of ROIs and associated waveforms

Three typical ROI definitions for regional lung ventilation and their associated waveforms are

illustrated in figure E3.5. In this example, a linear regression functional EIT image (fEIT) is

shown, where colors are normalized to the maximum image value. Two divisions of the

global fEIT image are shown. Figure E3.5 (left) shows four quadrants. Figure E3.5 (right)

shows four anteroposterior layers with equal height. A lung ROI is shown (Figure E3.5,

middle). Here the lung region is defined as those pixels with values above a threshold of

20% of the maximum.

Figure

healthy

Goe-MF

quadran

anterop

normali

changes

impeda

EIT sca

Ava

Modern

systems

frames

means

that ea

reductio

at high

E3.5. Com

volunteer

F II device

nts, lung

posterior lay

zed to 100%

s within th

nce change

an rate

ailable sca

n medical E

s offer muc

per second

that transie

ach measur

on in the si

scan rates

mmonly use

during pulm

(CareFusio

regions w

yers with e

%. Regiona

he correspo

es.

n rates fro

EIT equipm

ch faster rat

d (32). Inc

ent phenom

rement mu

gnal quality

also results

ed regions

monary fun

on, Höchbe

ith a thre

equal heigh

al impedanc

onding ROI

om typical

ent offers

tes, includin

creasing the

mena can b

ust be mad

y. For applic

s in large da

of interest

ction testin

erg, Germa

eshold of

hts. The g

ce waveform

Is during

l EIT equip

a scan rat

ng an indust

e scan rate

be better re

de more q

cations of p

ata files.

t (ROIs) de

ng. (The raw

any).) ROIs

20% of m

lobal imped

ms and perc

the forced

pment

te from 10

trial system

e increases

esolved. Sin

uickly, high

prolonged m

efined in an

w data was

s from left

maximum i

dance wave

centages re

maneuver

-50 scans

m with a fram

the tempo

nce a highe

her rates m

monitoring,

n fEIT ima

s acquired w

to right a

mage valu

eform (left

epresent imp

r. rel. ΔZ,

per second

me rate up

oral resoluti

er scan rate

may also c

recording E

8

age of a

with the

are four

ue, four

top) is

pedance

relative

d. Some

to 1000

ion, and

e means

cause a

EIT data

Sca

The EIT

analyze

minute

accordin

second

related

features

classica

(measu

required

Figure

PulmoV

pig dur

and EIT

represe

change

frequen

because

lungs m

These e

We rec

an rate req

T scan rate

ed. Respirat

(~0.67 Hz)

ng to the

should be

signals. In

s in the ca

al sampling

rements in

d sampling

E3.6. Glo

Vista 500 EI

ring high-fre

T scan rat

ent the card

.

Thus, in pra

ncy details o

e lung ven

may have slo

effects incre

commend a

quirements

e requireme

tory frequen

). Assuming

Nyquist–Sh

sufficient fo

n practice,

ardiac activi

theory is no

an EIT sc

rate to avo

bal impeda

IT device (D

equency os

te 40 Hz. S

diac-related

actice, high

of these EIT

tilation and

ow and fas

ease the sca

a scan rat

s for typica

ents depend

ncy in adult

g a heart r

hannon sam

or the samp

such as sc

ity within e

ot directly a

an are mad

id aliasing (

ance wavefo

Dräger Med

scillatory ve

Slower per

signal (the

her scan rat

T signals. T

d heart act

t compartm

an rate requ

e above 1

al monitor

d on the fre

ts and child

rate lower t

mpling theo

pling and s

can rate is

each beat,

applicable, s

de sequent

(34).

form (left)

dical, Lübec

entilation. T

riodic imped

e heart rate

tes are reco

The respirat

tion are no

ments (35) a

uirements fo

10 Hz for

ring tasks

equency of

dren is usua

than 180 be

orem, a sc

separation o

far too lo

which shou

since EIT sa

tially (33)).

and fEIT im

ck, German

The oscillato

dance chan

was 1.3 Hz

ommended,

ory and hea

t strictly pe

and thus be

or respirato

analyzing v

the signal

ally lower th

eats per m

an rate ab

of the respir

w. First, th

uld be reso

amples both

The effect

mage (right

y) in an a

ory ventilat

nges in the

z). rel. ΔZ,

in order to

art rate ma

eriodic. Add

e ventilated

ory and card

ventilation-

componen

han 40 brea

inute (3 Hz

bove 6 fram

ratory and

here are im

olved. Furth

h in space a

t is to incre

t) acquired

anesthetized

tion rate w

e global w

relative imp

capture th

ay vary dram

ditionally, d

d inhomoge

diac-related

related imp

9

ts to be

aths per

z), then,

mes per

cardiac-

mportant

hermore,

and time

ease the

d by the

d supine

as 9 Hz

aveform

pedance

e higher

matically

diseased

neously.

signals.

pedance

10

changes and 25 Hz for cardiac-related impedance changes. For high frequency oscillatory

ventilation, which contributes much higher frequency signals, the scan rate of EIT should

correspondingly larger (Figure E3.6). High scan rates are also needed for the assessment of

rapid gas volume changes at high air-flow rates (35-41).

Frequency filtering

Definition and application

Cardiac action and ventilation occur at different rates in humans, and their EIT signals

components can thus be discriminated using frequency filtering (29, 42, 43). The breathing

rate (12 -16 breaths/min in a spontaneously breathing adult) is typically much lower than

cardiac rate (60 – 90 beats/min). It is therefore possible to design a digital filter to separate

the ventilation and cardiac activity components provided by EIT. In the simplest case, a low-

pass filter (which lets low-frequency components of a signal “pass” through while removing

higher frequencies) identifies the ventilation signal while a high-pass filter with an opposite

effect identifies cardiac-related information. In practice, however, a band-pass filter (which

lets a range of frequencies pass while removing higher and lower frequencies) is used. It

removes some low-frequency disturbances (e.g. baseline drift related to hardware or drying

of the electrode gel) as well as high-frequency error sources (e.g. other interfering electrical

equipment). The band-pass filters can be generated by defining the expected normal

frequency ranges for breathing and cardiac activity. Other band-pass filters are designed

centered around a single frequency with an appropriate bandwidth. The center frequency of

the filter may be obtained directly from the ventilator or ECG monitor, from a human

operator analyzing the signal spectrum or from an automated algorithm that identifies the

highest peaks from the frequency spectrum (corresponding to the first harmonics) in the

frequency ranges expected for ventilation or cardiac activity.

Digital filtering can be performed in two ways. The raw EIT data can be filtered, and

then reconstructed, or the pixel waveforms in a sequence of raw EIT images can be filtered.

The results of both of these operations are identical when the reconstruction algorithm is

linear, which is the case for most of the commonly used reconstruction algorithms for chest

imaging. The digital filters can be implemented either in the time or frequency domains. The

design of digital filters is beyond the scope of this document, but many textbooks and

software toolboxes are available to assist in their design. In the design of digital filters, there

are several issues, such aliasing and the effects of scan time and non-periodic signals, that

can contribute undesired effects. They are discussed in the following paragraphs.

11

Other techniques such as ECG-gated EIT acquisition (17) and principal component

analysis (PCA) (24) have been proposed to circumvent some of the limitations observed with

frequency filtering. Some papers combine one of these two techniques with frequency

filtering to achieve better separation of the ventilation and cardiac activities (24). ECG-gating

can be performed at the hardware level where a series of EIT measurements are

automatically triggered whenever a QRS complex is detected in the patient’s ECG. When the

ECG is acquired simultaneously and synchronized with the EIT measurements, ECG-gating

can also be performed by post processing the acquired EIT data to identifying those EIT

frames where a QRS complex was synchronously detected in the ECG. Ensemble averaging

can then be performed on ECG-gated EIT measurements to improve the signal-to-noise ratio

(SNR) of the cardiac-related information from the EIT measurement. The same gating

process can also be performed using the ventilator trigger instead of the QRS complex in

order to increase the SNR of the ventilation-related information.

Aliasing

The most important signal processing issue to consider when using digital frequency filtering

is aliasing. Aliasing is a signal artefact in which content at one frequency in the original data

is represented at a different frequency in the filter output. It occurs when the sampling

frequency is not high enough for the frequency content of the signal. According to the

Nyquist-Shannon sampling theorem, aliasing occurs when the sampling rate is less than

twice the maximum frequency component of the signal. One special worry is that frequency

components near the sampling frequency can appear via aliasing as low frequency artefacts

in the image. As mentioned, EIT has a mix of spatial and temporal aliasing which is more

complicated than that described by the Nyquist-Shannon theory (33, 34). Extra care should

be taken when the heart rate is an exact multiple of the ventilation rate (or the vice versa,

with high-frequency ventilation) which could occur for instance if ventilation occurs in

neonates at e.g. 40 breaths/min and a heart rate of 120 beats/min. The cardiac frequency

would then be contaminated with the third harmonic of the ventilation making it impossible

to discriminate between them with frequency filtering alone.

Inadequate sample time

Unlike the sampling frequency, a parameter that is sometimes neglected in design of data

acquisition is the duration of sample acquisition, or sample time (T ). In cases where data

are acquired only for a very short sample time, it is impossible to discriminate between close

adjacent frequencies, and they will appear superposed in the spectrum. The frequency

12

resolution, or the difference between adjacent frequency bins, is given by the inverse of the

sample time, 1/T. The frequency resolution is also the lowest frequency difference that is

distinguishable in the signal. For example, if 30 seconds of EIT data are acquired, it will not

be possible to distinguish frequencies which differ by 1/30 s (2 per minute), making

ventilation frequencies of 6 and 8 breaths per minute appear identical.

Non-periodic changes

Frequency filtering works well in simple cases where the respiratory and heart rates are well

separated and do not vary during the whole acquisition time (in the signal processing

literature, this is referred to as a “stationary” signal). A mechanically ventilated patient with

a constant tidal volume (VT) and breathing rate can be considered a stationary EIT signal

source, assuming the heart rate was also relatively constant over the period of time. During

spontaneous breathing, the breathing rate can greatly vary and including pauses in

breathing (e.g. sleep apnea). The heart rate might also vary, especially if the subject is

performing tasks which include exercise. In such cases, an adaptive filtering strategy would

be recommended, in which filters adapt over time to best match the frequencies of the heart

and breathing activity.

Other (patho-)physiological events that may be visible in EIT data, such as dynamic

hyperinflation, produce non-periodic changes of impedance that might be missed if high-

pass frequency filtering alone is used, since it typically removes very slow changes. This is

true also of changes in ventilator settings, such as PEEP or VT. Slow changes of the EIT

signal can also occur due to hardware electronic drift, drying of the electrode gel or patient

movement (resulting in posture or electrode position changes). Although these slow

impedance changes are often not clinically relevant, it might be difficult in some cases to

distinguish them from those occurring from physiological events such as the onset of

atelectasis or pulmonary edema.

Description of expected waveforms

Ventilation-related changes in EIT waveforms

The shape of EIT waveforms is altered by any frequency filtering applied to the signals.

Commercially available EIT devices often set filters automatically to enhance visualization of

ventilation waveform (i.e., by low-pass filtering).

In the case of a completely passive mechanically ventilated patient, ventilation

impedance-time waveforms usually present a stable end-expiratory level. After an

impeda

corresp

value ty

tidal im

mode),

observa

hand, d

regiona

regiona

Figure

PulmoV

(blue) a

global E

and ma

level ca

can incr

the sha

similar t

cases, s

informa

nce increas

onding to e

ypically reac

There is a

mpedance ch

the shape

ation time w

during pres

al impedanc

al respiratory

E3.7. EIT

Vista 500 d

and volume

EIT wavefor

For sponta

agnitude of

an vary ove

rease or de

ape of eac

to that in c

several min

ation.

se during

end–inspirat

ches the pre

close corre

hanges. If V

and dimen

would be e

ssure-contro

ce wavefor

y system co

waveform

evice (Dräg

e (red) signa

rm. Paw, air

neously bre

f the ventil

er time, and

ecrease end

h deflection

controlled ve

nutes of obs

inspiration,

tion. The w

evious end-

elation betw

VT does not

nsion of eac

enough to o

olled ventil

ms could y

ompliance a

of a passiv

ger Medica

als obtained

rway pressu

eathing pat

ation EIT w

d variations

d-inspiratory

n of the g

entilation, t

servation m

the ventil

waveform fa

-expiratory

ween global

t change ov

ch ventilatio

obtain clini

ation, chan

yield clinica

and/or airwa

ve mechanic

l, Lübeck,

d from the

ure; rel. ΔZ,

ients (Figur

waveform.

s of respirat

y lung volum

global and

time interva

might be nee

lation wave

alls at end-e

level (Figur

and region

ver time (e.g

on wavefor

cally releva

nges in the

al informat

ay resistanc

cally ventila

Germany).

ventilator d

, relative im

re E3.8), th

The baseli

tory rate an

me and, co

regional im

als and peak

eded to obt

eform reach

expiration, a

e E3.7).

nal VT and g

g., during a

rm will be s

ant informa

e peak val

ion on cha

ce.

ated patient

Left: Airw

during the e

mpedance ch

here is varia

ne end-exp

nd respirato

nsequently,

mpedance w

k values are

tain stable a

hes a peak

and the imp

global and

a volume-co

stable, and

ation. On th

ues of glo

anges of g

t acquired w

way pressur

examination

hange.

ability in th

piratory imp

ory muscle

, impedanc

waveforms

e irregular.

and compre

13

k value,

pedance

regional

ontrolled

a small

he other

obal and

lobal or

with the

e (Paw)

n. Right:

e shape

pedance

activity,

e. While

appears

In such

ehensive

Figure

in a spo

rel. ΔZ,

In s

multitud

anxiety,

inspirat

tidal im

represe

collapse

Car

The EIT

pulsatili

identifie

the con

wavefor

lung reg

the lun

field of

view of

increase

(Figure

the glo

E3.8. Glob

ontaneously

relative im

spontaneou

de of facto

, body tem

tory and tro

mpedance c

ent higher a

ed regions (

rdiac-relat

T signal ch

ity compone

ed by using

ntributions a

rm is comp

gions. Durin

gs through

view. Blood

a thoracic

e in the h

E3.8, left).

obal signal

bal EIT wav

y breathing

pedance ch

usly breathi

ors (e.g.,

perature, e

ough expirat

changes acr

air content

(i.e., after a

ted change

anges indu

ent in EIT d

g frequency

at ventilatio

plicated bec

ng systole,

the pulmo

d from the

EIT electro

eart and a

When the

and only o

veform with

g patient (P

hange.

ing patients

ventilation

tc.) and thi

tory impeda

ross the im

t, allowing

atelectasis o

es in EIT w

uced by car

data is sma

y filtering. T

on rate and

cause of the

blood from

onary arteri

left ventric

de placeme

an impedan

heart rate

one phase

h ventilation

PulmoVista

s, inspirato

support,

is can lead

ance values

maging field

monitoring

or pneumoth

waveforms

rdiac activit

all compare

Typically, a

it higher ha

e propagat

m the right v

es. This co

cle moves t

ent. During

nce decreas

is high, the

is evident,

n-related ch

500, Dräge

ory effort c

arterial res

to variation

s and, conse

d. Higher p

of sighs o

horax resolu

s

ty are refer

ed to the ve

band pass

armonics (F

ion of bloo

ventricle lea

omponent o

hrough the

systole, the

se which p

e two deflec

similar to

hanges in im

r Medical,

can vary ov

spiratory g

ns in global

equently, in

peak levels

or re-expan

ution).

rred to as t

entilation sig

filter is ap

Figure E3.9)

d from the

aves the he

of blood flow

aorta, and

erefore, the

propagates

ctions can b

that of ven

mpedance o

Lübeck, Ge

ver time d

gas concen

l and region

n the distrib

at end-ins

nsion of pr

the pulsatil

gnal and it

pplied which

). The pulsa

heart thro

eart and mo

w stays in

d leaves the

ere is an imp

through th

be superimp

ntilation. H

14

obtained

ermany).

ue to a

trations,

nal peak

bution of

spiration

reviously

lity. The

is often

h rejects

atile EIT

ough the

oves into

the EIT

e field of

pedance

he lungs

posed in

However,

regiona

right).

Figure

heart a

single p

lung re

relative

(Dräger

Orig

While t

sources

volume

constan

change

extreme

an arte

extreme

changes

the lun

affects

inducing

good co

Using f

regions

al waveform

E3.9. Puls

nd pulmon

pixel in the

egion. There

impedance

r Medical, L

gin of puls

he major c

s affecting t

changes a

nt blood flo

(and thus

e, blood vol

rial blood c

e difference

s and is no

gs does ha

the impeda

g red blood

orrelation b

frequency f

in the thor

m analysis c

satile EIT w

ary perfusio

heart regio

e is a phas

e change. (

übeck, Germ

satility in t

ontribution

these signa

and not dire

ow or flow

no pulsatilit

lume oscilla

lot) would g

es illustrate

ot a direct m

ave a majo

ance distrib

d cell alignm

etween the

filtering an

rax (1).

can separat

aveforms. L

on-related s

on. Right: re

se shift bet

(The EIT d

many).)

the EIT sig

to the puls

ls (44). Pul

ectly to the

in a comp

ty) as volum

ation in a ve

give an imp

e that pulsa

measure of

or contribut

bution by cy

ment with

e pulsatile im

d ROI ana

te these tw

Left: global

signals. Mid

egional imp

tween the

ata was ac

gnal

satile EIT s

lsatile impe

e blood flow

pletely rigid

me in the v

essel closed

pedance cha

atile impeda

lung perfus

tion from p

yclic expan

the blood f

mpedance w

alysis, Ferra

wo wavefor

relative im

ddle: region

pedance wav

two latter

cquired usin

signal is blo

edance chan

w. For exa

d vessel, th

vessel would

d by an occl

ange over t

ance origina

sion. Howev

perfusion, s

sion of pul

flow. Sever

waveform a

ario et al d

rms (Figure

pedance wa

nal impedan

veform of a

regional w

ng the Pulm

ood perfusio

nges are pro

mple, in th

here would

d be consta

usion (e.g.,

ime in abse

ates in regi

ver, the EIT

since blood

monary blo

ral in vivo s

nd perfusio

discriminate

e E3.9, mid

waveform co

nce wavefo

a single pixe

waveforms.

moVista 500

on, there ar

roportional t

he extreme

be no imp

ant. In the o

, in the pres

ence of flow

ional blood

T pulsatile s

flow in th

ood vessels

studies cou

on (16, 28,

ed heart a

15

ddle and

ombining

orm of a

el in the

rel. ΔZ,

0 device

re many

to blood

case of

pedance

opposite

sence of

w. These

volume

signal in

he lungs

and by

uld show

45, 46).

nd lung

16

Common artifacts in EIT waveforms

While EIT signals contain much useful physiological information, there are many sources of

artifacts which can corrupt the EIT waveforms. Artefacts can originate in the raw EIT data,

or from the signal processing, where, for example, inappropriate filtering might cause

superimposition of waveforms from different origins. A number of effects (such as body

movement and posture change) can alter baseline end-expiratory impedance. As waveforms

are usually calculated as relative changes to baseline, this will alter the magnitude and

shape of the ventilation impedance waveforms.

Changes in electrode-skin contact impedance influence the calculated waveforms. The

electrode-skin impedance is affected by the contact force, drying of the gel or contact fluid

and other factors. EIT system vendors provide a measure of electrode-skin contact

impedance as a measure of signal quality.

Possible interference with other medical devices must be taken into account in

monitoring of intensive care patients, where electrical “noise” is produced by many other

devices in the ICU. An overview of most common interferences encountered during EIT

examinations in the ICU like pulsation therapy with air suspension mattresses, continuous

cardiac output monitoring and impedance pneumography has been provided by Frerichs et

al. along with the recommendations on EIT data analysis under these circumstances (47).

Other devices emitting electromagnetic signals may potentially interfere with EIT

measurements. However, overall, modern EIT devices are capable of generating high-quality

data in the majority of experimental and clinical settings.


The first draft of this online document was prepared by Z. Zhao with collaboration of H.

Gagnon, O. Stenqvist, T. Mauri, I. Frerichs and A. Adler. It was reviewed and approved by

all other authors and collaborators.

17

References

1. Ferrario D, Grychtol B, Adler A, Sola J, Bohm SH, Bodenstein M. Toward

morphological thoracic eit: Major signal sources correspond to respective organ locations in

CT. IEEE Trans Biomed Eng 2012;59:3000-3008.

2. Rossi Fde S, Yagui AC, Haddad LB, Deutsch AD, Rebello CM. Electrical impedance

tomography to evaluate air distribution prior to extubation in very-low-birth-weight infants: A

feasibility study. Clinics (Sao Paulo) 2013;68:345-350.

3. Schramel J, Nagel C, Auer U, Palm F, Aurich C, Moens Y. Distribution of ventilation

in pregnant shetland ponies measured by electrical impedance tomography. Respir Physiol

Neurobiol 2012;180:258-262.

4. Kunst PW, Vazquez de Anda G, Bohm SH, Faes TJ, Lachmann B, Postmus PE, de

Vries PM. Monitoring of recruitment and derecruitment by electrical impedance tomography

in a model of acute lung injury. Crit Care Med 2000;28:3891-3895.

5. Odenstedt H, Lindgren S, Olegard C, Erlandsson K, Lethvall S, Aneman A, Stenqvist

O, Lundin S. Slow moderate pressure recruitment maneuver minimizes negative circulatory

and lung mechanic side effects: Evaluation of recruitment maneuvers using electric

impedance tomography. Intensive Care Med 2005;31:1706-1714.

6. Meier T, Luepschen H, Karsten J, Leibecke T, Grossherr M, Gehring H, Leonhardt S.

Assessment of regional lung recruitment and derecruitment during a PEEP trial based on

electrical impedance tomography. Intensive Care Med 2008;34:543-550.

7. Victorino JA, Borges JB, Okamoto VN, Matos GF, Tucci MR, Caramez MP, Tanaka

H, Sipmann FS, Santos DC, Barbas CS, Carvalho CR, Amato MB. Imbalances in regional

lung ventilation: A validation study on electrical impedance tomography. Am J Respir Crit

Care Med 2004;169:791-800.

8. Frerichs I, Dargaville PA, van Genderingen H, Morel DR, Rimensberger PC. Lung

volume recruitment after surfactant administration modifies spatial distribution of ventilation.


9. Zick G, Elke G, Becher T, Schadler D, Pulletz S, Freitag-Wolf S, Weiler N, Frerichs I.

Effect of PEEP and tidal volume on ventilation distribution and end-expiratory lung volume: A

prospective experimental animal and pilot clinical study. PLoS One 2013;8:e72675.

10. Tingay DG, Wallace MJ, Bhatia R, Schmolzer GM, Zahra VA, Dolan MJ, Hooper SB,

Davis PG. Surfactant before the first inflation at birth improves spatial distribution of

ventilation and reduces lung injury in preterm lambs. J Appl Physiol 2014;116:251-258.

11. Pulletz S, van Genderingen HR, Schmitz G, Zick G, Schadler D, Scholz J, Weiler N,

Frerichs I. Comparison of different methods to define regions of interest for evaluation of

regional lung ventilation by EIT. Physiol Meas 2006;27:S115-127.

18

12. Frerichs I, Hahn G, Hellige G. Gravity-dependent phenomena in lung ventilation

determined by functional EIT. Physiol Meas 1996;17 Suppl 4A:A149-157.

13. Muders T, Luepschen H, Zinserling J, Greschus S, Fimmers R, Guenther U,

Buchwald M, Grigutsch D, Leonhardt S, Putensen C, Wrigge H. Tidal recruitment assessed

by electrical impedance tomography and computed tomography in a porcine model of lung

injury. Crit Care Med 2012;40:903-911.

14. Frerichs I, Dargaville PA, Rimensberger PC. Regional respiratory inflation and

deflation pressure-volume curves determined by electrical impedance tomography. Physiol

Meas 2013;34:567-577.

15. Elke G, Fuld MK, Halaweish AF, Grychtol B, Weiler N, Hoffman EA, Frerichs I.

Quantification of ventilation distribution in regional lung injury by electrical impedance

tomography and xenon computed tomography. Physiol Meas 2013;34:1303-1318.

16. Schibler A, Pham TM, Moray AA, Stocker C. Ventilation and cardiac related

impedance changes in children undergoing corrective open heart surgery. Physiol Meas

2013;34:1319-1327.

17. Vonk Noordegraaf A, Kunst PW, Janse A, Marcus JT, Postmus PE, Faes TJ, de

Vries PM. Pulmonary perfusion measured by means of electrical impedance tomography.

Physiol Meas 1998;19:263-273.

18. Zhao Z, Moller K, Steinmann D, Frerichs I, Guttmann J. Evaluation of an electrical

impedance tomography-based global inhomogeneity index for pulmonary ventilation

distribution. Intensive Care Med 2009;35:1900-1906.

19. Becher TH, Bui S, Zick G, Blaser D, Schadler D, Weiler N, Frerichs I. Assessment of

respiratory system compliance with electrical impedance tomography using a positive end-

expiratory pressure wave maneuver during pressure support ventilation: A pilot clinical

study. Crit Care 2014;18:679.

20. Becher T, Vogt B, Kott M, Schadler D, Weiler N, Frerichs I. Functional regions of

interest in electrical impedance tomography: A secondary analysis of two clinical studies.

PLoS One 2016;11:e0152267.

21. Pulletz S, Elke G, Zick G, Schadler D, Scholz J, Weiler N, Frerichs I. Performance of

electrical impedance tomography in detecting regional tidal volumes during one-lung

ventilation. Acta Anaesthesiol Scand 2008;52:1131-1139.

22. Gomez-Laberge C, Arnold JH, Wolf GK. A unified approach for EIT imaging of

regional overdistension and atelectasis in acute lung injury. IEEE Trans Med Imaging

2012;31:834-842.

23. Zhao Z, Steinmann D, Muller-Zivkovic D, Martin J, Frerichs I, Guttmann J, Moller K. A

lung area estimation method for analysis of ventilation inhomogeneity based on electrical

impedance tomography. J Xray Sci Technol 2010;18:171-182.

19

24. Deibele JM, Luepschen H, Leonhardt S. Dynamic separation of pulmonary and

cardiac changes in electrical impedance tomography. Physiol Meas 2008;29:S1-14.

25. Crapo RO, Morris AH, Clayton PD, Nixon CR. Lung volumes in healthy nonsmoking

adults. Bull Eur Physiopathol Respir 1982;18:419-425.

26. Woitzik C, Grychtol B, Ferrario D, Böhm S, Robitaille N, Adler A, Alzen G. A priori

Informationen für die elektrische Impedanztomografie (EIT) aus CT-daten des Thorax –

Erhebung und Bearbeitung anthropometrischer Daten und Konturanalysen unter besonderer

Berücksichtigung von pädiatrischen Patienten. Fortschr Rontgenstr 2013;185:V28.

27. Fagerberg A, Stenqvist O, Aneman A. Monitoring pulmonary perfusion by electrical

impedance tomography: An evaluation in a pig model. Acta Anaesthesiol Scand

2009;53:152-158.

28. Frerichs I, Pulletz S, Elke G, Reifferscheid F, Schadler D, Scholz J, Weiler N.

Assessment of changes in distribution of lung perfusion by electrical impedance tomography.

Respiration 2009;77:282-291.

29. Leathard AD, Brown BH, Campbell J, Zhang F, Morice AH, Tayler D. A comparison

of ventilatory and cardiac related changes in EIT images of normal human lungs and of

lungs with pulmonary emboli. Physiol Meas 1994;15 Suppl 2a:A137-146.

30. Smit HJ, Vonk-Noordegraaf A, Marcus JT, van der Weijden S, Postmus PE, de Vries

PM, Boonstra A. Pulmonary vascular responses to hypoxia and hyperoxia in healthy

volunteers and COPD patients measured by electrical impedance tomography. Chest

2003;123:1803-1809.

31. Hahn G, Dittmar J, Just A, Quintel M, Hellige G. Different approaches for quantifying

ventilation distribution and lung tissue properties by functional EIT. Physiol Meas

2010;31:S73-84.

32. Wilkinson AJ, Randall EW, Cilliers JJ, Durrett DR, Naidoo T, Long T. A 1000-

measurement frames/second ERT data capture system with real-time visualization. IEEE

Sens J 2005;5:300-307.

33. Yerworth R, Bayford R. The effect of serial data collection on the accuracy of

electrical impedance tomography images. Physiol Meas 2013;34:659-669.

34. Gagnon H, Grychtol B, Adler A. A comparison framework for temporal image

reconstructions in electrical impedance tomography. Physiol Meas 2015;36:1093-1107.

35. Pulletz S, Kott M, Elke G, Schadler D, Vogt B, Weiler N, Frerichs I. Dynamics of

regional lung aeration determined by electrical impedance tomography in patients with acute

respiratory distress syndrome. Multidiscip Respir Med 2012;7:44.

36. Bodenstein M, Boehme S, Bierschock S, Vogt A, David M, Markstaller K.

Determination of respiratory gas flow by electrical impedance tomography in an animal

model of mechanical ventilation. BMC Pulm Med 2014;14:73.

20

37. Vogt B, Pulletz S, Elke G, Zhao Z, Zabel P, Weiler N, Frerichs I. Spatial and temporal

heterogeneity of regional lung ventilation determined by electrical impedance tomography

during pulmonary function testing. J Appl Physiol 2012;113:1154-1161.

38. Zhao Z, Muller-Lisse U, Frerichs I, Fischer R, Moller K. Regional airway obstruction

in cystic fibrosis determined by electrical impedance tomography in comparison with high

resolution CT. Physiol Meas 2013;34:N107-114.

39. Zhao Z, Fischer R, Frerichs I, Muller-Lisse U, Moller K. Regional ventilation in cystic

fibrosis measured by electrical impedance tomography. J Cyst Fibros 2012;11:412-418.

40. Frerichs I, Zhao Z, Becher T, Zabel P, Weiler N, Vogt B. Regional lung function

determined by electrical impedance tomography during bronchodilator reversibility testing in

patients with asthma. Physiol Meas 2016;37:698-712.

41. Vogt B, Zhao Z, Zabel P, Weiler N, Frerichs I. Regional lung response to

bronchodilator reversibility testing determined by electrical impedance tomography in chronic

obstructive pulmonary disease. Am J Physiol Lung Cell Mol Physiol 2016;311:L8-L19.

42. Zadehkoochak M, Blott BH, Hames TK, George RF. Pulmonary perfusion and

ventricular ejection imaging by frequency domain filtering of EIT (electrical impedance

tomography) images. Clin Phys Physiol Meas 1992;13 Suppl A:191-196.

43. Smallwood RH, Hampshire AR, Brown BH, Primhak RA, Marven S, Nopp P. A

comparison of neonatal and adult lung impedances derived from EIT images. Physiol Meas

1999;20:401-413.

44. Braun F, Proença M, Rapin M, Alba X, Lekadir K, Lemay M, Sola J, Frangi AF,

Thiran JP. 4D heart model helps unveiling contributors to cardiac EIT signal. In: Sola J,

Braun F, Adler A, editors. 16th International Conference on Biomedical Applications of

Electrical Impedance Tomography. Neuchatel; 2015. p. 107.

45. Nguyen DT, Jin C, Thiagalingam A, McEwan AL. A review on electrical impedance

tomography for pulmonary perfusion imaging. Physiol Meas 2012;33:695-706.

46. Fagerberg A, Sondergaard S, Karason S, Aneman A. Electrical impedence

tomography and heterogeneity of pulmonary perfusion and ventilation in porcine acute lung

injury. Acta Anaesthesiol Scand 2009;53:1300-1309.

47. Frerichs I, Pulletz S, Elke G, Gawelczyk B, Frerichs A, Weiler N. Patient

examinations using electrical impedance tomography-sources of interference in the intensive

care unit. Physiol Meas 2011;32:L1-L10.








ONLINE SUPPLEMENT 4

Functional EIT images

Functi

This ele

generat

recorde

mathem

The ma

measur

onset o

process

Figure

EOS 4 (

EOS, ele

center o

R, right

Genera

EIT is a

large nu

the clin

distribu

speed, f

The

obtaine

algorith

ional EIT

ectronic on

ted from ra

ed in paral

matical oper

athematical

re specific

or regional r

ses involved

E4.1. Sequ

(gray backg

ectronic on

of ventilatio

t; L, left.

al principle

an imaging

umber of ra

nical inform

ted over a

frequency a

e generatio

ed from a s

hm is applie

images

nline supple

aw EIT imag

lel during

ration on a

l operation

regional ph

respiratory s

d in EIT exa

uence of pro

ground) des

line supplem

on; U/L, upp

e of functi

technique w

aw images

mation on v

a number o

and localiza

on of fEIT

sequence o

ed to calcul

ement (EOS

ges (and, in

the EIT ex

sequence o

s determin

hysiological

system com

minations a

ocesses inv

cribes the g

ment; tdEIT

per-to-lower

onal EIT im

with a high

(EOS 2) w

ventilation

of raw ima

tion of impe

images req

of raw EIT

ate a physi

S) describes

n some cas

xamination)

of raw EIT i

ne the type

characteri

mpliance (Cr

and the plac

volved in EIT

generation o

T, time-diffe

r ventilation

maging

temporal r

within each

and cardia

ages and re

edance cha

quires the

images, o

iologically r

s how func

ses, also us

). An fEIT

images and

e of fEIT

stics, such

rs). Figure E

ce of fEIT im

T chest exa

of functiona

erence EIT;

n ratio; RVD

resolution th

breath. No

c activity b

equires ana

nges in a se

availability

n which a

relevant par

ctional EIT

ing other p

image is

d the resulti

image, and

as delay

E4.1 shows

mages in th

mination an

al EIT image

ROI, region

D, regional v

hat can qui

single EIT

by itself. S

alyzing the

equence of

of EIT wa

mathemati

rameter for

(fEIT) ima

physiologica

generated

ing EIT wav

d are desig

in local ve

the sequen

his scheme.

nd data ana

es. Abbrevia

n-of-interes

ventilation d

ckly provide

image con

Such inform

character,

raw image

aveforms (

ical functio

r each imag

2

ages are

l signals

from a

veforms.

gned to

entilation

nce of all

alysis.

ations:

st; CoV,

delay;

e a very

ntains all

mation is

timing,

s.

EOS 3),

n or an

ge pixel.

The res

generat

types o

system,

otherwi

ventilat

during

opening

Var

and cha

is poss

differen

example

visualize

We des

and pro

Figure

sequenc

(EOS 3)

(describ

wavefor

obtaine

or volum

sult is a s

ting an fEIT

of fEIT ima

, can be ge

ise noted,

ted newborn

an increm

g pressure w

rious types

aracterize re

sible to gen

nt times thro

e, an imag

ed by subtr

scribe some

ovide refere

E4.2. Conc

ce of raw E

) from the

bed in the t

rm (B); D)

ed by compu

me recordin

single new

T image is

ages, addr

enerated fr

are based

n pig using

ental and

were simult

of fEIT ima

egional lung

nerate seco

oughout an

ge of the c

racting the

e of the mo

nces for oth

ceptual diag

EIT images

e image se

text) a fun

) Some ap

uting the su

ngs from the

image cal

illustrated

ressing diff

rom a sing

on a sing

g the Goe-M

decrementa

aneously m

ages have b

g ventilation

ondary func

n experimen

hange in v

fEIT image

ost common

hers, thoug

gram of the

is obtained

quence (A)

ctional EIT

pproaches

um of all im

e patient ve

lled a func

in Figure E

ferent funct

le EIT exam

gle dataset

MF II EIT d

al PEEP tr

measured.

been propos

n and/or pe

ctional ima

ntal or clinic

ventilation d

es obtained

n approache

h the list is

e process o

d (EOS 2); B

); C) Using

image pixe

require a

mage pixels,

entilator or

ctional imag

E4.2. To hi

tional char

mination, a

t (1). Thes

device (Care

ial, during

sed in the l

erfusion. Fo

ages, by co

cal interacti

distribution

before and

es to fEIT i

by no mea

f computing

B) A wavefo

g various s

el value is

global wav

, or other a

spirometer.

ge, The ge

ghlight the

racteristics

all figures in

se data we

eFusion, Hö

which bot

iterature an

or most type

omparing r

on with the

after an in

d after the

mage gene

ns compreh

g a function

orm is obta

signal proc

calculated

veform (tim

uxiliary dat

eneral proc

fact that d

of the res

n this EOS

ere acquire

öchberg, Ge

th EIT and

nd used to

es of fEIT i

results achi

e subject. T

ntervention

intervention

eration in th

hensive.

nal EIT ima

ained for ea

cessing app

based on t

me course)

ta such as p

3

cess for

different

spiratory

S, unless

ed on a

ermany)

d airway

quantify

mage, it

eved at

Thus, for

can be

n (2, 3).

his EOS,

ge. A) a

ach pixel

proaches

the pixel

that is

pressure

4

Types of functional EIT images

We group fEIT imaging approaches according to their target physiological parameters:

ventilation distribution, ventilation- and cardiac-frequency impedance changes, aeration

change, respiratory system mechanics, ventilation timing, and tissue classification (i.e. for

regions of repeated opening and closing and for overdistended and atelectatic tissue). The

different approaches differ in their requirements for additional simultaneous information in

addition to the EIT data (e.g. airway pressure), and specific clinical interventions (e.g.

constant low-flow inflation). We conclude by describing various approaches to combine

different types of fEIT image to label the state of different regions of the lung into states

(e.g. atelectatic or overdistended).

Because of the variety of fEIT approaches and techniques, many different terms have

been used for the same calculation. In some cases, early terminology has been replaced by

more precise terms. In Table E4.1, we show the fEIT imaging categories and the associated

imaging techniques. Technical details are described in subsequent sections.

5

Table E4.1. Overview of the various fEIT imaging techniques with older or alternative

taxonomy organized by fEIT imaging categories.

fEIT imaging category fEIT imaging technique Alternative terms

Ventilation

(distribution of)

Standard deviation fEIT SD fEIT

Tidal variation fEIT VT fEIT, TV fEIT, tidal fEIT

Normalized tidal variation fEIT Stretch fEIT

fEIT of other lung volumes than VT FVC fEIT, FEV1 fEIT, IVC fEIT

fEIT of ratios of lung volumes or

flows

FEV1/FVC fEIT, FEF75/FEF25 fEIT

Frequency analysis of

impedance changes

Ventilation-frequency fEIT

Cardiac-frequency fEIT,

Pulsatility fEIT

Cardio-synchronous fEIT, heart

beat-related fEIT, perfusion fEIT

Aeration change Volume-difference fEIT ∆EELV, ∆EELZ, ∆EELI

Respiratory system

mechanics

fEIT of dynamic Crs

fEIT of quasi-static Crs and landmark

pressure points

fEIT of lower and upper inflection

point pressures

fEIT of opening / closing pressures

Regression fEIT

Linear regression fEIT, slope fEIT,

filling capacity, polynomial fit fEIT,

regional curvilinearity fEIT, filling

index

Time constant fEIT

Ventilation timing Ventilation delay fEIT Opening / closing delay

Expiration time fEIT

Tissue classification fEIT of overdistension / atelectasis

Low tidal variation fEIT Silent spaces

Abbreviations: Crs, respiratory system compliance; EELI, end-expiratory lung impedance;

EELV, end-expiratory lung volume; EELZ, end-expiratory lung impedance; FEF25, forced

expiratory flow at 25% of FVC exhaled; FEF75, forced expiratory flow at 75% of FVC

exhaled; fEIT, functional EIT; FEV1, forced expiratory volume in 1 s; IVC, inspiratory vital

capacity; FVC, forced vital capacity; TV, tidal variation; VT, tidal volume; ∆, change.

Figure E4.3 provides an overview of these types of fEIT images, based on a set of

maneuvers or interventions during EIT data acquisition. During an initial PEEP setting, a

subject is ventilated and fEIT measures of (A) ventilation and Crs, and (B) delays and

6

respiratory time constants are calculated. A new PEEP setting is then applied and fEIT

measures of (C) changes in aeration (e.g. in end-expiratory lung volume (ΔEELV)) are

calculated. Last, a pressure-volume (P-V) loop is performed, from which (D) other fEIT

parameters of respiratory system mechanics are calculated.

Figure E4.3. Conceptual diagram of the types of fEIT images, based on raw images

acquired during an initial PEEP setting, an increased PEEP and a quasi-static pressure-

volume (P-V) inflation and deflation maneuver. Each letter indicates the type of fEIT images:

A) measures of the distribution of ventilation and of regional compliance (the illustration

shows a measure of tidal variation), B) measures of regional delays and time constants (the

illustration shows a measure of inspiratory time), C) measures of regional changes in

aeration due to an intervention (the illustration shows a slow inflation), and D) measures of

PV loop parameters (the illustration shows maximum curvature-change points).

fEIT images of ventilation distribution

One of the key advantages of EIT is its ability to provide real-time information on the

distribution of ventilation. Thus, the earliest (and still most common) use of fEIT was to

characterize the distribution of ventilation. Several different fEIT approaches have been

developed, which we discuss below.

Sta

The firs

(SD) fE

wavefor

largest

Figure E

wavefor

impeda

the corr

Figure

the wav

wavefor

represe

values

displaye

pulmon

from th

Thi

working

each b

generat

limitatio

andard dev

st type of f

EIT image (

rm of the

variation o

E4.4 illustra

rms are rep

nce values

responding

E4.4. Stan

veform SD

rms at the

ent the distr

in each hi

ed as pixel

ary and ca

e most ante

is simple a

g well at slo

reath, and

ting only a

ons to SD fE

viation fEI

functional E

(4). Each p

pixel in the

of impedanc

ates a SD fE

presented o

for each wa

pixel in the

ndard devia

of individua

indicated p

ributions of

istogram ca

values in th

rdiogenic s

erior pixel c

pproach ha

ow scan ra

was thus

few frames

EIT images

IT images

EIT image p

ixel of a SD

e image se

ce observed

EIT image. F

on the left.

aveform use

e functional

ation (SD) f

al pixels ov

pixel location

impedance

an be cha

he fEIT ima

signal sourc

close to the

as the adva

tes, when o

suitable fo

s per secon

. It is not a

presented in

D fEIT imag

equence. Th

d in a sequ

For each of

The middle

ed to comp

image.

functional E

ver the ana

ns are show

e values for

racterized

ge (right).

ces. This is

heart (top

antage of r

only a few

or the ear

nd. While sim

a specific m

n the literat

ge represen

his techniqu

uence of im

the three b

e histogram

pute the SD

EIT image (

alyzed time

wn. The his

each pixel

by a SD,

Note that t

seen in the

left).

requiring lit

data samp

ly EIT dev

mple to cal

measure, and

ture is the

nts the SD

ue relies on

ages is rela

black pixels,

s represent

value that

(right) obta

period (left

stograms in

waveform.

σ, shown.

he SD fEIT

e upper wa

ttle comput

ples (in time

vices, which

culate, ther

d thus is un

standard d

of the dat

n the fact t

ated to ven

, the corres

t the distrib

is then ass

ained by ca

ft). On the

the center

The distrib

These val

image rele

aveform ori

tational pow

e) are avail

h were cap

re are a nu

nable to dis

7

deviation

a in the

that the

ntilation.

sponding

bution of

igned to

lculating

left, the

r column

bution of

ues are

cts both

ginating

wer and

lable for

pable of

umber of

stinguish

8

ventilation-related changes in impedance from the cardio-synchronous ones, or artifacts in

the data (caused e.g. by poor electrode contact). Additionally, since, by its calculation, SD is

always positive, areas of increasing or decreasing air volume cannot be distinguished in such

images.

Tidal variation fEIT images

Tidal variation functional EIT (TV fEIT) images show the difference between EIT image

values at end-inspiratory and end-expiratory times for each pixel of the functional image.

The occurrences of end-inspiratory and end-expiratory events are typically automatically

identified on a global waveform that is obtained by computing the sum of all pixels of the

image sequence. The values of each pixel waveform at these time occurrences are then

averaged for each of the end-inspiratory and end-expiratory events. Each pixel of the TV

fEIT image is then calculated as the difference between its average end-inspiratory and

average end-expiratory value (Figure E4.5). Each of the three black dot pixel values is

obtained from the waveforms on the left. The red dots indicate the waveform values at each

end-inspiratory event while the blue dots indicate the waveform values at each end-

expiratory event. The red horizontal line represents the average of the end-inspiratory

values while the blue line represents the average of the end-expiratory values. The

difference between the red and the blue lines represents the pixel values in the TV fEIT

image.

The advantage of this method is that it is specific to ventilation changes. It is more

complicated computationally because the signal processing is required to identify the breath

boundaries. While automatic approaches to detect end-inspiratory and end-expiratory events

work well in many circumstances, they can occasionally give incorrect results when the

pattern of ventilation is unusual or when there is a difference in phase of ventilation among

regions. Incorrect detection of these breathing time points will disturb the fEIT images. One

correction is for the human operator to review and correct these detections (e.g. (5)).

Normalized tidal variation fEIT images

A “normalized tidal ventilation image” is an fEIT image in which TV image is normalized such

that each image is represented as a fraction of the one with the maximum value. In this

case, each image pixel is then a fraction or a percentage value. Such an image is useful to

visually present the spatial pattern of the distribution of ventilation within the lung. This has

also been referred to as the “stretch” fEIT image (6).

9

Figure E4.5. Tidal variation (TV) fEIT image (right). Time courses (left) from the indicated

pixels are analyzed to determine the values at each end-inspiratory (red dots) and end-

expiratory (blue dots) event, and subsequently averaged, yielding the red and blue lines.

Each fEIT image pixel value is obtained by subtracting the values corresponding to the red

and blue lines (also shown as a value of TV) and is visually expressed in the respective

shade of gray of the pixels,

fEIT images of other lung volumes than VT

Regional ventilation distribution has mostly been examined by EIT during either

spontaneous or mechanical tidal ventilation, with a focus on monitoring tidal breathing.

However, it is also possible to analyze the distribution of other lung volumes in the chest

cross-section than VT. Virtually all lung volumes typically determined during conventional

pulmonary function testing by spirometry as global values at the airway opening can be

assessed by EIT on a regional level. The prerequisite is that the subjects are examined by

EIT during well-known ventilation maneuvers like full inspiration and forced full expiration.

The resulting pixel EIT waveforms are analyzed by calculation of differences between the

impedance values at characteristic time points (e.g., full expiration to residual lung volume,

full inspiration to total lung capacity with the onset of forced expiration and the end of

forced full expiration). This allows the assessment of pixel lung volumes like inspiratory vital

capacity (IVC), forced vital capacity (FVC), expiratory (ERV) and inspiratory reserve volumes

(IRV) or forced expiratory volume in 1 s (FEV1) (7-11). The distribution of these lung

volumes is then visualized by plotting these values as fEIT.

10

fEIT images of ratios of lung volumes or flows

Regional lung function is conventionally assessed not only by determining the lung volumes

but also by several quotients determined as ratios between various lung volumes or

between air flows. The best known are the ratios of FEV1 and FVC or of maximum forced

flow rates at defined volumes. These ratios can be determined by EIT in individual image

pixels representing the chest cross-section. The corresponding fEIT images can be used to

determine the spatial heterogeneity of ventilation.

A few clinical studies have been performed using this approach. For instance, EIT-

derived FEV1/FVC values acquired in patients suffering from chronic obstructive lung disease

were not only significantly lower than in healthy subjects but they also exhibited a higher

inhomogeneity of their spatial distribution (8). These measures were able to differentiate

patients with asthma from age-matched controls (11) and to identify the regional effects of

an inhaled bronchodilator during reversibility testing (10). The fEIT images of pixel ratios of

forced expiratory flows after exhalation of 75% and 25% of FVC identified the spatial

distribution of airway obstruction in patients with cystic fibrosis (12).

fEIT images using frequency analysis

As described in EOS 3, frequency filtering can be used to extract ventilation and cardiac-

related information from EIT images. The underlying concept is to filter each EIT pixel

waveform to extract only the signal at the breathing (and the cardiac, if desired) frequency,

and then to measure the amplitude of this extracted signal. The breathing and heart rates

are determined either 1) automatically, by analyzing the power spectrum of the global

waveform – obtained by computing a Fourier Transform (FT) on the global signal and

identifying the highest peak in the expected frequency range for breathing and heart rates

or 2) from another instrument such as the patient ventilator, impedance plethysmograph,

ECG or pulse oximetry. Once these rates are known, the FT of each waveform

corresponding to individual pixels in the sequence of raw images is performed. The

frequency signal amplitudes at the breathing and heart rates are then each assigned to a

different fEIT image: the first corresponds to ventilation while the second represents heart-

frequency related impedance changes. Figure E4.6 shows examples of fEIT images obtained

by frequency analysis.

11

Ventilation-frequency fEIT images

The frequency-analysis fEIT method has many advantages in its ability to identify and

measure ventilation-related changes. It has a high specificity; by using a narrow, adapted

filter centered on the breathing frequency, it is able to reject non-ventilation contributions to

the fEIT image. However, there are also several practical challenges. First, it is sometimes

difficult to identify the peak corresponding to the ventilation rate in the frequency spectrum,

as it is not always the highest one. Second, it is not clear how harmonics of the main

respiratory and heart rate should be analyzed, since only a fraction of the amplitude of the

breathing signal is contained at the principle frequency component. (In a few studies,

frequency filtering is performed using a low or band-pass filter with the 1st and 2nd

harmonics of the breathing rate included (e.g. (13, 14)). Another difficulty with this method

is its assumption of a stable breathing and cardiac frequency. This assumption of a stable

breathing frequency is appropriate for some ventilation modes, but not for spontaneously

breathing patients or for some support ventilation modes. It is, however, the method of

choice for high-frequency oscillatory ventilation (HFOV), where the waveform is simpler but

only a few data points per breath are available (e.g. (15, 16)).

Cardiac-frequency fEIT images

Frequency-analysis fEIT can provide high quality images of the cardiac-related EIT image

contributions. Since the cardiac-frequency contributions are much smaller and closer to the

EIT frame rate, a very selective filter (such as provided by this technique) is required.

Clearly, if the heart rate is not stable, frequency analysis will not be an appropriate analysis

technique. Since this cardiac frequency-filtered image contains information on perfusion it

has been called a “perfusion image” in some publications (e.g. (13, 17, 18)). However, there

are several physiological reasons why the EIT image filtered at the cardiac frequency does

not correspond directly to the perfusion. First, this analysis assumes a constant and stable

heart rate, which may not be valid in most clinical scenarios. Next, overestimation of

perfusion is possible, because the movement of the heart via cardioballistic effects

independently contributes to EIT signals in addition to the perfusion contribution (19).

Underestimation of perfusion is also possible, because small blood vessels (especially

capillaries and veins in which blood flow is continuous) will not show a pulsatile component

(20). This latter effect would also presumably be gravity dependent. We thus recommend,

this type of fEIT image be referred to as “cardiac-related”, “heart beat-related”, “cardiac

frequency-filtered” or the “pulsatility” fEIT image. If the cardiac frequency is a multiple of

the ven

from ve

Figure

images,

left (the

lower d

graphs)

height

0.34 Hz

ventilat

of the

harmon

respirat

ntilation freq

entilation as

E4.6. Fun

, two pixels

e upper fra

ot, in the lu

) of individ

of the mai

z = 20.4 m

tion (bottom

breathing

nic of the ca

tory rate; PH

quency, the

s explained

nctional EIT

s are select

me shows t

ung region)

ual pixel w

n peaks, co

min-1 and

m-right) and

frequency

ardiac frequ

HR, power d

cardiac fre

in EOS 3.

T images o

ted (black d

the upper d

. Undernea

waveforms

orrespondin

2.81 Hz =

d cardiac-re

can be cle

uency can b

density at t

equency-filte

obtained vi

dots). The w

dot, in the

th each wa

(plots) is o

ng to the r

168.6 min

elated (top-

early seen

be seen at

the heart ra

ered image

ia frequenc

waveform a

heart regio

veform, the

obtained via

respiratory

n-1, respect

-right) func

starting at

about 5.7 H

te.

also contai

cy (spectra

at each pixe

n; the lowe

e power spe

a the Four

and heart

tively) are

tional imag

t about 0.6

Hz. PRR, po

ns aliasing

l) analysis.

el is plotted

er frame sh

ectrum dens

rier transfor

rates (in t

used to c

ges. The ha

6 Hz, and t

ower densit

12

artifacts

In the

d on the

ows the

sity (bar

rm. The

his case

calculate

armonics

the first

ty at the

fEIT im

When v

recruitm

aeration

the dis

propose

Vo

Volume

the res

PEEP.

interven

volume

Figure

in PEEP

wavefor

dots) ti

after th

calculat

interven

Fig

pixel va

mages of a

ventilation

ment maneu

n of the lun

stribution of

ed to perfor

lume-diffe

e-difference

spiratory cy

If the volu

ntion, the im

difference”

E4.7. Ima

P of 5 cmH

rm is displa

me points a

he PEEP int

ted from th

ntion.

ure E4.7 sh

alues from a

eration ch

therapy is

uver is perfo

ng. To help

f additiona

rm this calc

erence fEIT

fEIT image

ycle (e.g. e

ume differe

mage is ref

” or “end-ex

ge of chang

H2O, as a f

ayed (left),

are identifie

tervention

he differenc

hows such a

a sequence

hange

adjusted b

ormed, the

predict the

l or lost g

ulation.

T images

es represen

end-expiratio

ence is tak

ferred to as

xpiratory lun

ge in end-e

function of

at which e

ed. The ave

(blue and

ce in the va

an fEIT imag

e of raw EIT

by increasin

underlying

e effectivene

gas volume

nt the chang

on) followin

ken betwee

s a “change

ng impedan

expiratory lu

time (s). I

end-expirat

erage end-e

red lines).

alues highlig

ge. A globa

T images a

ng the tidal

clinical obj

ess of the t

es. Several

ge in lung v

ng an inter

en end-exp

e in EELV”, “

nce” fEIT im

ung volume

In each ind

ory (blue d

expiratory v

Each pixel

ghted as b

l waveform

nd is used

l volume o

jective is of

therapy, it i

fEIT appro

volume at a

rvention, su

piration bef

“ΔEELV”, “e

mage.

e (AU, right

dicated pixe

dots) and e

value is cal

in the ima

lue lines be

is obtained

to automat

or PEEP or

ften to incre

is useful to

oaches hav

a particular

uch as incr

fore and a

end-expirato

t), due to a

el in the im

end-inspirato

culated bef

age (right)

efore and a

d by summi

tically ident

13

when a

ease the

analyze

ve been

point of

rease in

after an

ory lung

change

mage, a

ory (red

fore and

is then

after the

ng of all

tify end-

14

expiratory events. Based on the global EIT waveform, the time of each end-expiratory event

is identified. Next, with each pixel waveform in the raw image sequences, end-expiratory

values are identified at each identified end-expiratory time. End-expiratory values before and

after the intervention are separately averaged, and then the difference between these two

averages calculated. The functional image pixel values therefore reflect the increase or

decrease in end-expiratory lung volume due to the intervention. An analogous calculation for

the end-inspiratory event yields a functional image showing the change of the distribution of

lung volume at end-inspiration. It should be noted that it is necessary for the calculation of

any of these functional images that the same baseline (reference) measurement frame is

used to reconstruct the difference EIT raw image before and after the intervention.

fEIT images of regional respiratory system mechanics

fEIT images of regional dynamic respiratory system compliance

Crs is a measure of the ease with which the chest wall and lung can increase in volume in

response to an increase in airways pressure. Mechanical measurements of Crs are thus in

units of L/cmH2O. EIT has seen much interest due to its potential to calculate regional

measures of Crs. Such an image of the regional distribution of Crs provides useful information

on the mechanical properties of lung tissue and can thus help guide therapeutic

interventions. Unfortunately, it is not possible in vivo to calculate the regional

micromechanical volumes and pressures from which true regional Crs could be calculated.

For lung pressure values, it is typical to assume that pressure is uniform throughout the

lung, and to use the pressure reading from a ventilator when flow values have reached zero

at both end-expiration and end-inspiration; this assumption may not be valid with

obstructive lung diseases, especially if there were regions of trapped gas.

fEIT images of regional Crs interpret gas volume as proportional to the EIT signal. This

can be either the EIT image (in arbitrary EIT image units, AU) (1) or with the raw EIT image

calibrated to some external measure of regional volume or global lung volume has been

sometimes used (e.g. (21-23)). In all cases, measurements of incremental changes of lung

volume and pressure are required to calibrate fEIT images of regional Crs. Without

calibration, such fEIT images would be considered to be images of ventilation.

Regional dynamic Crs has often been determined from single breaths under dynamic

ventilation conditions during an incremental and decremental PEEP trial (1, 22) and used to

assess regional alveolar recruitment and overdistension. It can identify the PEEP step at

which the most homogeneous distribution of regional Crs occurs.

15

It is worth noting that fEIT images of aeration change also provide an approximation of

Crs, when normalized by the difference in airway pressure between peak-inspiration and

peak-expiration.

fEIT images of regional quasi-static respiratory system compliance

The quasi-static pressure-volume (P-V) maneuver is a standard tool to study respiratory

system mechanics. As the lung inflates it changes from an initial low compliance (due to

collapsed or atelectatic regions) through a region of high compliance and finally returns to a

low compliance behavior (due to overdistension). The P-V curve is also able to measure the

hysteresis of lung properties when acquired during both inflation and deflation. EIT data

acquired during a P-V curve can provide regional functional information which can enhance

the clinical value of the procedure. Figure E4.8 illustrates the types of information available

from an EIT measurement during a P-V curve maneuver.

Figure E4.8. Illustration of the fEIT measurements available from data during a quasi-static

constant low-flow inflation and deflation P-V maneuver. The top row shows raw EIT images

during a period of ventilation, followed by inflation, deflation and then continued ventilation.

The middle row shows representative EIT waveforms, and the bottom row volume

(calculated from the EIT waveform, either calibrated to volume units, or in units of raw EIT

impedance changes (ΔZ)). Parameter types which can be measure are illustrated: A) time to

16

cross a threshold fraction of the total ΔZ (which may be represented as an opening or

closing pressure), B) time constant of an exponential curve fit to a step change in airway

pressure, C) sigmoid curve fit to the P-V loop hysteresis, and D) maximum curvature-change

values from the fitted curve (also referred to as “inflection” points).

When EIT data are recorded simultaneously with airway pressure either during a

constant low-flow or a stepwise P-V maneuver, a quasi-static pressure-impedance curve is

obtained for each pixel. Kunst et al. were the first to study fEIT measures from P-V curves

using a low-flow inflation protocol, and used the data to estimate upper (UIP) and lower

inflection points (LIP) in the EIT curves (24). A similar approach was used by van

Genderingen et al. (25). Further work was done by Frerichs et al who fit the sigmoid curves

of both inflation and deflation to EIT data from quasi-static P-V loops obtained in

experimental animal with healthy, acutely injured and surfactant-treated lungs (26).

Grychtol et al. proposed a method to automatically extract LIP and UIP, as well as Crs

assessment from such curves (27). Miedema et al. generated deflation pressure-impedance

curves before and after surfactant treatment in high-frequency oscillatory ventilated preterm

infants (28).

Note that the term “inflection point”, while commonly used, is in fact mathematically

incorrect. The points of interest are those at which a maximum change in slope of the curve

occurs, but this is not an “inflection” (which is a change in curvature from positive to

negative). Instead, we recommend the terminology “maximum curvature-change points”.

Opening/closing pressure fEIT images

As the lungs inflate from a low volume state, parts of the lung begin to open, and, similarly,

as the lungs deflate, parts of the lung can close. A functional image of the pressures at

which these events occur is a “regional opening (or closing) pressure fEIT image”. In

patients with acute respiratory distress syndrome, such images show significant

heterogeneity, which may help to guide therapy.

In (5), regional opening and closing pressures were defined through a low-flow inflation

and deflation maneuver. As the lungs were slowly inflated (allowing the pressure in lung

regions to equalize to quasi-static conditions), the waveform in each image pixel was

analyzed to determine when it began to increase from baseline (opening pressure) or when

it returned to baseline (closing pressure). In this work, regional opening and closing

pressures in each pixel were defined by the crossing of a threshold value in each pixel of the

EIT image sequence, above which the lung region corresponding to that pixel is considered

17

open. A functional image of regional opening or closing pressure is then composed by

assigning each pixel the value of airway pressure at the moment at which that threshold is

crossed, respectively during inflation or deflation. This work defined the threshold in each

pixel as a percentage (commonly 10%) of the total magnitude of changes recorded in that

pixel during a respiratory cycle. Results based on (5) are shown in Figure E4.9.

Figure E4.9. Calculation of regional opening and closing pressures fEIT images (5). Data

were from slow-flow inflation and deflation on a patient with acute respiratory distress

syndrome. The normalized global EIT and airway pressure (upper left) are shown, as well as

the EIT waveform at four pixels in the image (colored lines, lower left). Based on the time of

crossing a threshold, fEIT images of regional opening and closing pressures (units of

cmH2O) are shown.

It is possible to calculate opening and closing pressure fEIT images for each breath

during conventional ventilation (given a sufficiently high scan rate); however, such dynamic

conditions would not permit the assumption of quasi-static pressures in the lungs. An fEIT

showing regional opening pressure during tidal breathing is shown in Figure E4.10.

Figure

(dark s

expirati

of its m

maneuv

pressur

changes

lower o

Re

If a me

calculat

linear fi

a regre

images

“Filling

used, w

In

relation

represe

for som

global v

EOS 3)

E4.10. Re

pots), the

on (black d

maximum t

ver) is iden

re (bottom

s less than

opening airw

egression f

easure of p

te a regress

it of regiona

ession analy

and the gl

capacity”. I

we recomme

a regress

nship betwe

enting each

me pixels to

value increa

but in spe

gional open

waveforms

dot), the mo

idal change

ntified (gree

left) is used

n half the a

way pressur

fEIT image

ressure syn

sion functio

al to global

ysis betwee

obal wavef

It is describ

end the term

ion fEIT i

een each

pixel’s ave

show a sm

ases. Such

ecific situati

ning pressu

are shown

oment at w

e (or overa

en dot) and

d to compo

average we

es.

es

nchronous w

onal EIT ima

measures.

en the wav

orm. This t

bed in deta

minology “re

mage, eac

individual

rage contri

mall negativ

pixels are

ons such a

re fEIT ima

n at left du

which an ind

all change

d the corre

ose a functi

ere excluded

with the EIT

age, which

Regression

veform of e

technique w

ail by (30, 3

egression fE

ch pixel re

pixel wave

bution to th

ve slope, m

usually as

as one-lung

age (right).

uring conve

dividual wav

determined

esponding v

onal image

d (set to b

T signal is

gives a com

n fEIT image

each pixel

was develop

31). While t

EIT image”.

epresents t

eform and

he global w

eaning that

sumed to l

ventilation

For the ind

ntional ven

veform (top

through a

value of ins

e (right). Pix

lack). Brigh

not availab

mpliance m

es are obta

from a seq

ped by (29)

the original

.

the slope

the globa

waveform. It

t their value

lie outside

n, they may

icated imag

ntilation. Aft

p left) reach

a low-flow

stantaneous

xels exhibit

hter colors

ble, it is pos

measure bas

ained by per

quence of

) and initial

terminolog

of the reg

al waveform

t is not unc

e decrease

the lung R

y be located

18

ge pixels

ter end-

hes 10%

inflation

s airway

ing tidal

indicate

ssible to

sed on a

rforming

raw EIT

ly called

gy is still

gression

m, thus

common

s as the

ROI (see

d in the

(contral

While li

where

curviline

Z=a·Zglo

Zglob glo

Figure

the pixe

below (

global w

the reg

regress

Tim

A time

system

expone

mechan

pressur

complia

informa

lateral) lung

inear regre

the lung

earity of lu

obFI+c to th

obal impeda

E4.11. Re

el waveform

(left, bottom

waveforms

gression slo

ion slope as

me constan

constant c

with a re

ntial relaxat

nical proper

re and volu

ance. The s

ation on the

g region as

ssion is co

filling or e

ng filling h

e EIT wave

ance, a and

egression fE

m is plotted

m, rotated a

are plotted

pe, r, is in

s the value

nt fEIT im

characterize

esistive and

tion. The m

rties charac

ume chang

spatial distri

e spatial di

s well (30).

mmonly us

emptying c

has also bee

eforms on a

c constants

EIT image

as a functi

and reflecte

d and the b

dicated. Th

for each pix

ages

s the rate

d elastic e

most commo

cterized by

ges with a

ibution of t

stribution o

. Figure E4

sed, polyno

characterist

en describe

a pixel-by-pi

s, and FI th

(right). For

ion of time

ed). In the

best linear f

he fEIT ima

xel.

of relaxatio

element wi

on models o

a complian

time cons

time consta

of lung mec

4.11 represe

mial regres

ics are ve

ed by fitting

ixel basis (3

e filling ind

r three ima

(left) and t

left center,

fit calculate

age (right)

on of a sys

ll respond

of lung tissu

nce and res

stant, τ, pr

nts in a fEI

chanical pro

ents a regre

ssion is also

ry non-line

g an equat

34, 35). (Z

ex.)

ge pixel loc

the global w

, time point

ed. For each

is displaye

stem after

to a step

ue consider

sistance, an

roportional

IT image th

operties. O

ression fEIT

o possible

ear (32, 3

tion of the

is pixel imp

cations (ind

waveform is

ts on the p

h line, the

ed, using th

a change.

p change

r the tissue

nd thus res

to resistan

hus provide

One example

19

T image.

in cases

3). The

form of

pedance,

dicated),

s plotted

pixel and

value of

he linear

A linear

with an

to have

spond to

nce and

es useful

e of the

20

use of such images is to visualize the effects of surfactant treatment on the regional

dynamic tissue properties (36, 37).

To obtain time constant fEIT images, EIT data should be recorded before and after a

step increase or decrease in airway pressure is performed (36). An exponential decay curve

( / is then fitted to the EIT waveform after the pressure increase or decrease for

each pixel. The time constant is then calculated to each pixel value to form the time

constant fEIT image. Time constant analysis was also used by Miedema et al. who

performed small incremental and decremental pressure steps in preterm infants receiving

HFOV and analyzed the time constant of adaptation in the EIT signal in global, ventral and

dorsal regions (37). Time constants may be calculated for other physiological phenomena in

the lungs, including for the inspiratory and expiratory phase of each breath (38).

fEIT images of ventilation timing

Ventilation delay fEIT images

Regional ventilation delay measures a similar property to regional opening and closing

pressures. Lung regions which are less compliant will respond more slowly to an applied

pressure step and will begin opening later than more compliant regions. Regions with large

ventilation delays may thus be considered pathological, and the corresponding fEIT images

should provide a useful guide to therapy (39).

The regional ventilation delay fEIT image represents the regional ventilation delay

compared to the global waveform. Figure E4.12 represents a regional ventilation delay

functional EIT image for one breath obtained from a ventilated pig. The pixel values are

obtained by first finding, for each breath, the instant where each corresponding pixel

waveform reaches a threshold of its overall magnitude as defined by its end-expiratory and

end-inspiratory values. (Threshold values of up to 40% were used (39)). Where the

threshold was crossed at an instant between two EIT data samples, the estimate is

calculated by interpolation of the EIT waveform (using linear or higher order interpolation).

The same process of finding the instant where the values reach a threshold of the overall

magnitude is performed on the global waveform. The regional ventilation delay pixel values

correspond to the difference between each individual waveform and the global waveform. A

negative value means that the corresponding ventilation in this pixel leads the average

ventilation while a positive value means that the ventilation in this pixel is lagging the

average ventilation.

Reg

ventilat

dynamic

flow br

weight)

resistan

calculat

crosses

corresp

time.

Figure

(dark s

panels)

values

(20% o

axis at

dots) fo

breath

ventilat

expirati

pixels in

lower th

gional vent

tion; howev

c conditions

reath (e.g.

) can be pe

nce. As men

ted in simil

a thresho

onding pre

E4.12. Re

spots in the

. The globa

(blue dots)

of its maxim

right). Each

or an individ

(green lin

tion delay a

on and ins

n each brea

han half the

tilation dela

ver, the ca

s. To calcul

inspiratory

rformed to

ntioned, reg

lar ways. B

old; howeve

ssure, while

egional ven

e image) a

al EIT wave

are determ

mum excurs

h pixel wav

dual pixel (t

e). The fE

and the glo

spiration, re

ath cycle. C

e average ti

ay can be c

lculated im

late delays

gas flow 4

increase pr

gional vent

Both calcula

er, the ope

e the regio

ntilation del

are selected

eform is sho

mined. The

sion) is det

veform is th

top left) to

EIT image

obal delay.

espectively,

Color axis o

dal change

calculated f

mage will th

under quas

4 L/min, or

ressure reso

ilation delay

ations seek

ening-pressu

onal ventilat

ay EIT fun

d, and the

own (bottom

time for th

termined an

hen analyze

reach 20%

is then th

Blue and r

while blac

on image is

were exclu

from data

hen represe

si-static con

r tidal volu

olution and

y and open

k the point

ure fEIT im

tion delay i

nctional ima

waveform

m left) for w

he global w

nd represen

ed to determ

% of its over

he time dif

red dots ma

ck dots ma

s in milliseco

uded (blacke

collected d

ent the lun

nditions, a l

me 12 mL/

minimize t

ning-pressur

at which

mage is the

mage is pre

age (right).

shown at

which minim

waveform to

nts a delay

mine the tim

rall magnitu

fference be

ark the mo

rk the extr

onds. Pixels

ed out).

during conv

ng pressure

large consta

/kg predicte

the effect o

re fEIT ima

the EIT w

en mapped

esented in

Two imag

t left (left

mum end-ex

o cross a th

of zero (s

me required

ude change

etween eac

oment of m

rema for in

s with tidal

21

ventional

es under

ant low-

ed body

of airway

ages are

aveform

on the

units of

ge pixels

top two

xpiration

hreshold

ee color

d (green

in each

ch pixel

maximum

ndividual

change

22

In the presence of lung collapse, atelectatic areas may open up during inspiration and

collapse again during expiration, a phenomenon called tidal recruitment. Wrigge et al have

shown that the time-dependency of regional filling quantified as regional ventilation delay

correlates with tidal recruitment (40).

Expiration time fEIT images

Calculation of pixel expiration times can also be applied to assess the temporal

heterogeneity of ventilation. This approach has been applied in spontaneously breathing

subjects examined by EIT during conventional pulmonary function testing. The pixel

expiration times needed to exhale defined fractions (25%, 50%, 75% and 90%) of FVC

during forced full expiration were shown to be a robust measure to determine the high

temporal heterogeneity of exhalation in patients with airway obstruction (8, 10, 11).

fEIT images with classification of regional lung tissue state

Different types of functional EIT image may be jointly analyzed to diagnose regional lung

(patho-) physiology. Of particular clinical interest is the identification of hyper- or (over-)

distended and atelectatic lung regions, as this provides clinically useful physiological

information.

fEIT images of regional overdistention and atelectasis

Both ovedistended and atelectatic regions are characterized by low Crs, and thus low tidal

impedance change. They differ in that an increase in airways pressure may open a collapsed

area, leading to its opening and hence an increase in Crs, whereas in an overdistended lung

region increased pressure leads to a further reduction of Crs.

Based on this difference, two approaches for identifying lung tissue were proposed (27,

41). Costa et al. analyzed changes in functional images of Crs during a PEEP titration (41).

Using the assumption that airway pressure is uniform throughout the lung, local Crs is

calculated by dividing local tidal change by the difference in airway pressure between peak-

inspiration and peak-expiration. The amount of collapse or overdistention in the lung area

corresponding to each pixel is quantified by comparing its Crs at a given PEEP level with the

best Crs observed throughout the maneuver. This approach was also described in (42). In

(23) it was used to guide choice of PEEP settings in management of experimental lung

injury.

23

The approach of (27) is conceptually similar, but does not require a full titration

maneuver. Instead, a single step increase in PEEP is used by a fuzzy logic expert system to

identify lung areas which have become more open or more overdistended (or more

collapsed or less overdistended for case of a decrease in PEEP). Rather than assuming

uniform airway pressure, functional images of tidal changes were analyzed in tandem with

volume change images to determine the direction of each pixel’s movement along its P-V

curve, and thus allow detection of the events of interest.

fEIT images of low tidal variation

A simple approach to tissue classification is to determine all lung regions which have no – or

very small – contribution to the global image amplitude. The approach of (6) identified so

called “silent spaces” as those image pixels in which the normalized TV is less than a

threshold of 10%. We propose the terminology “low tidal variation” regions as a more

general synonym for “silent spaces”. The advantage of this approach is its simplicity over a

classification into collapsed or overdistended categories, since a lung region with low TV

may have many possible causes, including overdistension, collapse or pneumothorax.

Robustness of fEIT approaches

Functional imaging approaches are an exciting aspect of EIT image analysis, since these

images provide the connection between the raw EIT waveforms and the regional behavior of

lung tissue. There is an active development of fEIT strategies to make them more robust

and clinically useful. One key question in the selection of fEIT approach is the extent to

which a given approach is sensitive to the sources of noise and artefact (e.g. low electrode

contact quality) that may occur in EIT signals. There is clear evidence that some fEIT

strategies are more robust than others. Pulletz et al. studied the impact of fEIT on the

quantitative assessment of ventilation distribution (30). Hahn et al. compared their response

to noise in a simulation study (31).

To illustrate these differences, the figure E4.13 compares three types of fEIT

calculations used to measure regional ventilation amplitude, and described in the sections

above: SD, regression and TV fEIT images. When EIT raw data are of high quality and

interference is low, all these types of fEIT deliver similar images (Figure E4.13, top).

However, real EIT measurements, especially over prolonged times, often show sources of

interference (Figure E4.13, bottom). In this case, the most robust is the calculation of TV

24

fEIT images, since this type of fEIT updates its baseline with every breathing cycle by

subtracting end-expiratory EIT images from end-inspiratory images. Hence, it can reduce

the errors caused by measurement drift (Figure E4.13, bottom right).

Figure E4.13. Comparison of fEIT images of regional ventilation amplitude calculated from

EIT measurements on a patient during long-term monitoring during weaning. Three image

calculations are shown: standard deviation fEIT (left column), regression fEIT (middle

column), and tidal variation fEIT (right column). The top row shows images for the same

data set, over a short period when no signal drift was present. The bottom row shows a

subsequent long period with drift in the raw data (e.g. change of contact impedance,

electrode movement). While all fEIT approaches appear similar without signal drift, the tidal

variation fEIT with continuously updated baseline is the most robust in the presence of drift.


The first draft of this online document was prepared by B. Grychtol, Z. Zhao, A. Adler and

I. Frerichs with collaboration of H. Wrigge. It was reviewed and approved by all other

authors and collaborators.

25

References

1. Dargaville PA, Rimensberger PC, Frerichs I. Regional tidal ventilation and

compliance during a stepwise vital capacity manoeuvre. Intensive Care Med 2010;36:1953-

1961.

2. Frerichs I, Hahn G, Schroder T, Hellige G. Electrical impedance tomography in

monitoring experimental lung injury. Intensive Care Med 1998;24:829-836.




4. Hahn G, Šipinková I, Baisch F, Hellige G. Changes in the thoracic impedance

distribution under different ventilatory conditions. Physiol Meas 1995;16:A161-173.

5. Pulletz S, Adler A, Kott M, Elke G, Gawelczyk B, Schadler D, Zick G, Weiler N,

Frerichs I. Regional lung opening and closing pressures in patients with acute lung injury. J

Crit Care 2012;27:323 e311-328.

6. März A, Ukere A, Wodack K, Trepte C, Waldmann A, Böhm SH, Reuter D.

Perioperative assessment of regional ventilation during changing body positions and

ventilation conditions by electrical impedance tomography with increased spatial resolution

and signal quality. Critical Care 2015;19 (Suppl 1):P249.

7. Vogt B, Falkenberg C, Weiler N, Frerichs I. Pulmonary function testing in children

and infants. Physiol Meas 2014;35:R59-90.




9. Krueger-Ziolek S, Schullcke B, Kretschmer J, Muller-Lisse U, Moller K, Zhao Z.

Positioning of electrode plane systematically influences EIT imaging. Physiol Meas

2015;36:1109-1118.










26



Respiration 2009;77:282-291.







16. Wolf GK, Grychtol B, Frerichs I, Zurakowski D, Arnold JH. Regional lung volume

changes during high-frequency oscillatory ventilation. Pediatr Crit Care Med 2010;11:610-

615.

17. Smit HJ, Handoko ML, Vonk Noordegraaf A, Faes TJ, Postmus PE, de Vries PM,

Boonstra A. Electrical impedance tomography to measure pulmonary perfusion: Is the

reproducibility high enough for clinical practice? Physiol Meas 2003;24:491-499.



2009;53:152-158.

19. Proenca M, Braun F, Rapin M, Sola J, Adler A, Grychtol B, Bohm SH, Lemay M,

Thiran JP. Influence of heart motion on cardiac output estimation by means of electrical

impedance tomography: A case study. Physiol Meas 2015;36:1075-1091.

20. Borges JB, Suarez-Sipmann F, Bohm SH, Tusman G, Melo A, Maripuu E, Sandstrom

M, Park M, Costa EL, Hedenstierna G, Amato M. Regional lung perfusion estimated by

electrical impedance tomography in a piglet model of lung collapse. J Appl Physiol

2012;112:225-236.

21. Lowhagen K, Lundin S, Stenqvist O. Regional intratidal gas distribution in acute lung

injury and acute respiratory distress syndrome-assessed by electric impedance tomography.

Minerva Anestesiol 2010;76:1024-1035.

22. Gomez-Laberge C, Rettig JS, Smallwood CD, Boyd TK, Arnold JH, Wolf GK.

Interaction of dependent and non-dependent regions of the acutely injured lung during a

stepwise recruitment manoeuvre. Physiol Meas 2013;34:163-177.

23. Wolf GK, Gomez-Laberge C, Rettig JS, Vargas SO, Smallwood CD, Prabhu SP,

Vitali SH, Zurakowski D, Arnold JH. Mechanical ventilation guided by electrical impedance

tomography in experimental acute lung injury. Crit Care Med 2013;41:1296-1304.

24. Kunst PW, Bohm SH, Vazquez de Anda G, Amato MB, Lachmann B, Postmus PE,

de Vries PM. Regional pressure volume curves by electrical impedance tomography in a

model of acute lung injury. Crit Care Med 2000;28:178-183.

27

25. van Genderingen HR, van Vught AJ, Jansen JR. Estimation of regional lung volume

changes by electrical impedance pressures tomography during a pressure-volume

maneuver. Intensive Care Med 2003;29:233-240.



Meas 2013;34:567-577.

27. Grychtol B, Wolf GK, Adler A, Arnold JH. Towards lung EIT image segmentation:

Automatic classification of lung tissue state from analysis of EIT monitored recruitment

manoeuvres. Physiol Meas 2010;31:S31-43.

28. Miedema M, de Jongh FH, Frerichs I, van Veenendaal MB, van Kaam AH. Changes

in lung volume and ventilation during surfactant treatment in ventilated preterm infants. Am J

Respir Crit Care Med 2011;184:100-105.

29. Kuhnel G, Hahn G, Frerichs I, Schroder T, Hellige G. Neue Verfahren zur

Verbesserung der Abbildungsqualität bei funktionellen EIT-Tomogrammen der Lunge.

Biomed Tech (Berl) 1997;42 Suppl:470-471.




31. Hahn G, Dittmar J, Just A, Quintel M, Hellige G. Different approaches for quantifying

ventilation distribution and lung tissue properties by functional EIT. Physiol Meas

2010;31:S73-84.



Physiol 2001;91:39-50.

33. Hinz J, Gehoff A, Moerer O, Frerichs I, Hahn G, Hellige G, Quintel M. Regional filling

characteristics of the lungs in mechanically ventilated patients with acute lung injury. Eur J

Anaesthesiol 2007;24:414-424.

34. Rooney D, Friese M, Fraser JF, K RD, Schibler A. Gravity-dependent ventilation

distribution in rats measured with electrical impedance tomography. Physiol Meas

2009;30:1075-1085.

35. Grant CA, Fraser JF, Dunster KR, Schibler A. The assessment of regional lung

mechanics with electrical impedance tomography: A pilot study during recruitment

manoeuvres. Intensive Care Med 2009;35:166-170.




28

37. Miedema M, de Jongh FH, Frerichs I, van Veenendaal MB, van Kaam AH. Regional

respiratory time constants during lung recruitment in high-frequency oscillatory ventilated

preterm infants. Intensive Care Med 2012;38:294-299.

38. Róka P, Waldmann A, Müller B, Ender F, Bohm SH, Windisch W, Strassmann S,

Karagiannidis C. Breath-by-breath regional expiratory time constants by electrical

impedance tomography – a feasibility study. In: Sola J, Braun F, Adler A, editors. 16th

International Conference on Biomedical Applications of Electrical Impedance Tomography.

Neuchatel; 2015. p. 50.





40. Wrigge H, Zinserling J, Muders T, Varelmann D, Gunther U, von der Groeben C,

Magnusson A, Hedenstierna G, Putensen C. Electrical impedance tomography compared

with thoracic computed tomography during a slow inflation maneuver in experimental models

of lung injury. Crit Care Med 2008;36:903-909.

41. Costa EL, Borges JB, Melo A, Suarez-Sipmann F, Toufen C, Jr., Bohm SH, Amato

MB. Bedside estimation of recruitable alveolar collapse and hyperdistension by electrical




2012;31:834-842.








ONLINE SUPPLEMENT 5

EIT measures

EIT me

Overvie

EIT mea

are des

(global

generat

resulting

physiolo

EI

assess

contrast

parame

complem

EIT ima

Figure

data (E

image e

further

the phy

time-dif

ventilati

Types o

A large

measure

easures

ew

asures are

igned to re

measures)

ted from a

g EIT wav

ogical functi

IT measure

the presen

t to an fEI

ters which

mentary. Fig

age sequenc

E5.1. Sche

OS 1), a se

element (pix

calculations

ysiological st

fference EIT

ion ratio; RV

of EIT mea

number of

es can be

the values

present diff

) or large

mathemat

veforms. D

ion is repre

es are valu

nt state of

IT image,

represent

gure E5.1 s

ce, fEIT ima

ematic diag

equence of

xel) is analy

s are made

tate of the

T; ROI, reg

VD, regiona

asures

f EIT meas

classified

calculated f

ferent featu

image reg

tical operat

Depending

sented.

es derived

ventilation

an EIT me

t the phys

shows a sch

ages, and E

ram of the

f EIT image

yzed to pro

to obtain E

lung. Abbre

ion-of-inter

al ventilation

sures have

into three

from analys

ures of the l

gions. As d

tion on a

on the fE

from the

(and perfu

easure is a

siological st

hematic diag

IT measure

calculation

es is calcul

oduce an fEI

EIT measur

eviations: E

rest; CoV, c

n delay; R,

been propo

e groups: a

sis of functi

lung functio

described in

sequence

IT image

examinatio

usion) distr

a single pa

tate. Imag

gram of the

es.

of EIT mea

ated (EOS

IT image (E

res which ea

EOS, electro

center of ve

right; L, lef

osed. To h

averaged g

ional EIT (f

on at level o

n EOS 4,

of raw EIT

type, a d

ns that are

ribution and

rameter or

es and nu

relationshi

asures. First

2). The tim

EOS 4). Fro

ach represe

onic online s

entilation; U

ft.

elp categor

global or re

fEIT) image

of the entire

an fEIT im

T images a

ifferent as

e typically

d trends. T

r small num

umeric valu

p between

t, from the

me course

om this fEIT

ent some as

supplement

U/L, upper-t

rize this wo

regional me

2

es which

e image

mage is

and the

pect of

used to

Thus, in

mber of

ues are

the raw

raw EIT

of each

T image,

spect of

t; tdEIT,

to-lower

ork, EIT

easures,

3

measures for characterization of spatial distribution of ventilation, and examination-specific

measures.

1. Average regional EIT measures represent the functional measures used to generate

fEIT images, only that the pixel values of these measures are averaged (or summed) in the

whole image or its sections (e.g., quadrants or layers). For instance, pixel tidal impedance

changes summed in all image pixels provide a global measure of regional tidal variation in

the studied chest slice.

2. Measures for characterization of the spatial distribution of ventilation consist of

measures which aim to describe the location and spatial configuration of the ventilated

region. Examples are the center of ventilation and ventilation heterogeneity measures.

These measures are also calculated from pixel functional measures.

3. Examination-specific measures characterize specific aspects of lung physiology

using specific examinations or measurements of EIT in parallel with other signals, such as

airway pressure. This is a varied group of EIT measures which is sometimes calculated from

pixel functional measures, but can also by calculated using other approaches.

Figure E5.2. Classification of EIT measures into three principal groups, as well as the sub-

classifications of the types of measures within each group. PFT, pulmonary function testing.

4

Average regional EIT measures

The first type of EIT measures are those representing the average (or sum) of fEIT

measures. These measures are typically used to characterize the overall ventilation and

aeration and the changes in these quantities, e.g. tidal volume, and changes in end-

expiratory lung volume. Such EIT measures may be calculated in the whole image or in

ROIs dividing the whole image into regions, as discussed in EOS 3. ROIs may be defined on

the whole image or divide only the ventilated region identified as lungs.

In cases where the physiological effects of interest occur in the gravity direction and

the patient is supine or prone, ROIs are defined to divide the defined region in the

anteroposterior direction with horizontal divisions, into either two regions (an upper and

lower region) or into a larger number of regions. The highest degree of spatial resolution

offer the ventilation profiles calculated from the smallest layer ROIs with the height of one

image pixel. A further division of ROIs is to consider a left/right division between image

regions. These EIT measures allow description of physiological processes which differentially

affect the left and right lungs. The most commonly used divisions have been into two

(upper/lower divisions) or into four quadrant regions.

Figure E5.3. Illustration of the calculation of regional EIT measures, based on the tidal

variation fEIT image of Figure E4.5 in EOS 4. Left: Regional EIT waveforms in three image

pixels highlighted as small black squares in the images. Middle: Calculation of the average

image values in each quadrant, where resulting values are shown in each quadrant, below.

Right: Calculation of average image values in the left and right layers, resulting values are

shown as the length of bars below.

5

The process of calculation of an average regional EIT measure is illustrated in figure

E5.3, in which the tidal variation fEIT image is first calculated, and then the average pixel

value in each quadrant is calculated and displayed. Such average (or summed) regional EIT

measures may be applied to any of the fEIT images defined in EOS 4, as summarized in

Table E5.1.

Table E5.1. EIT measures calculated from average (or summed) regional fEIT images.

EIT measure fEIT image

Regional/global standard

deviation EIT measure

Standard deviation fEIT image

Regional/global tidal variation

EIT measure

Tidal variation fEIT image

Regional/global aeration change

EIT measure

Aeration change fEIT image

Regional/global regression

coefficient EIT measure

Regression coefficient fEIT

image

Regional/global time constant

EIT measure

Time constant fEIT image

Regional/global ventilation delay

EIT measure

Ventilation delay fEIT image

Measures for characterizations of the spatial distribution of ventilation

The second group consists of measures aimed at characterizing the spatial distribution of

ventilation in the image. Since a large part of the promise of EIT for ventilation monitoring is

driven by its ability to calculate a regional map of ventilation, such measures are key to

characterizing and summarizing this information.

As a descriptive statistical measure, the spatial distribution may be divided into

measures as follows:

1. Measures of the central position of the distribution of ventilation

2. Measures of the spatial extent of the distribution of ventilation

Calculation of measures of the spatial distribution of ventilation requires using

formulae which are more complicated than simply an average or sum of fEIT measures, and

6

are thus categorized as a different type of EIT measure. These types of measures are

described below.

Measures of the central position of the distribution of ventilation

Such measures globally characterize the orientation of the distribution of ventilation within

the chest. The most common measures characterize the distribution in the anteroposterior

direction. This direction is the clinically relevant when patients are examined in the supine or

prone positions; i.e. it corresponds to the orientation of the gravity vector. However, when

relevant to the physiology or the ventilation pattern, measures of right-left position have

been used, such as for postural changes (1-3) or unilateral ventilation (4, 5).

a. Anterior-to-posterior ventilation ratio (A/P ratio)

A common measure of the anterioposterior distribution of ventilation is the anterior-to-

posterior ventilation ratio (A/P ratio). This measure was originally called the "impedance

ratio" (6). Is has also been referred to as the upper-to-lower ventilation ratio (U/L ratio). As

illustrated in Figure E5.4, this index relates the ventral impedance changes to dorsal

impedance changes by computing

A/P ratio = (Anterior ventilation) / (Posterior ventilation)

where the upper (ventral) and lower (dorsal) ventilation are calculated

(Anterior ventilation) = Sum of pixel values in anterior half of image (ΣA), and

(Posterior ventilation) = Sum of pixel values in posterior half of image (ΣP).

The A/P ratio is thus the ratio between the sum of tidal impedance changes in the

anterior and posterior halves of the functional image or the lung ROI within it, as illustrated

in Figure E5.4. As the center of the distribution of ventilation moves ventrally, for instance

during volume-controlled ventilation when positive end-expiratory pressure (PEEP) is

decreased and dorsal atelectasis develops, the ventilation in the anterior region (ΣA) will

increase, while the ventilation in the posterior region (ΣP) will decrease, resulting in an

increase in A/P ratio.

While the A/P ratio is relatively widely used, we do not consider it to be the most

robust measure of the vertical distribution of ventilation, and instead, recommend the center

of ventilation (CoV), described below. Because A/P is a ratio, its value can become arbitrarily

large if the posterior region (ΣP) has a low value. This means that the change in the A/P

ratio is not proportional to the change in the position of ventilation. Additionally, the lung

areas are not identical in the upper and lower halves of the image, which means that the

A/P ratio is not expected to be equal to one, even in the case of uniform ventilation.

7

Figure E5.4. Illustration of computation of the anterior-to-posterior (A/P) ventilation ratio.

The image is divided into two regions horizontally through the center, and the sum of the

functional image values in the anterior (ΣA) and posterior (ΣP) areas are calculated.

b. Center of ventilation (CoV)

Another common measure of anteroposterior distribution is the center of ventilation (CoV),

introduced by Frerichs et al. (7). The CoV is motivated by the concept of center of gravity

(CoG) which is used in mechanics, defined as the location in space which characterizes the

center of a distributed mass. (Indeed, the CoV parameter is sometimes called the CoG,

although we do not recommend this terminology). CoV has been applied as a measure of

ventilation distribution in several studies (8-12). For its application to an EIT image, the

center of ventilation may be computed by

CoV [%] = (Height weighted pixel sum) / (Pixel sum)

where

(Pixel sum) = Σ (ΔZj), and

(Height weighted pixel sum) = Σ (yj × ΔZj)

where y is the vertical position (y-axis) in the fEIT image, yj is vertical position of pixel j and

is scaled to increase in the gravity direction, and ΔZj is the image sum in the horizontal ROI

at position yj. The calculation of these values is illustrated in figure E5.5. CoV yields a value

between 0 and 100% where 0% would indicate all image amplitude at the top, and 100%

all amplitude at the bottom of the image. As the center of the distribution of ventilation

moves dorsally, CoV increases. It is worth noting that a few authors have scaled the CoV

differently (e.g. to scale it between −1 and +1 (13), or to reverse the direction such that

100% is the top (14)). For the horizontal axis, it is possible to define an analogous

horizontal CoV, x CoV.

The CoV has been used in many studies, as has been shown to be a sensitive index to

describe opening and closing of the lung during incremental and decremental PEEP trials

8

(15). For many applications in which an EIT measure of the vertical position of ventilation is

required, the CoV has been used in both experimental and clinical studies (8, 16-19). We

recommend the use of CoV, instead of A/P ratio, since CoV is a linear measure of the

position; thus a change in the position of the distribution of ventilation will introduce a

proportional change in CoV, but not in the A/P ratio. Additionally, CoV is a function of each

pixel layer in the image, and thus more sensitive than a simple division into upper and lower

regions.

Figure E5.5. Illustration of computation of the center of ventilation (CoV). The image is

divided into horizontal regions (of pixel width, thinner than illustrated). In each horizontal

region, yj, the image sum ΣZj is determined and the CoV calculated using the formula

provided in the text above. Z, impedance.

Measures of the spatial extent of the distribution of ventilation

These measures are designed to describe variations in the pattern of ventilation which are

not simply a change in the position of the center of the distribution. Such characterizations

are needed for several features of the image, depending on the physiology of interest.

The most common measures of this type are used to describe the overall degree of

spatial heterogeneity of ventilation. For some applications, it is important to characterize the

level of inhomogeneity in an EIT image, since many pathological processes compromise the

more uniform distribution of ventilation which exists under normal conditions. Such

measures are also calculated from the pixel functional measures. Examples are the global

inhomogeneity index or the coefficient of variation.

a. Global inhomogeneity index

One characterization of the degree of inhomogeneity is the global inhomogeneity index (GI),

introduced by Zhao et al. (20). The calculation of GI is based on the difference between

9

each pixel value and the median value of all pixels. These values are normalized by the sum

of impedance values within the lung area.

GI = Σ(pixel differences from median) / Σ(pixels)

where

Σ(pixels) = Σ (ΔZj), and

Σ(pixel differences from median) = Σ (ΔZj − ΔZmedian)

where ΔZj is the fEIT image value in pixel j, ΔZmedian is the median image value, and all sums

are calculated for all pixels j in the image. In (20-22), GI was based on the tidal variation

fEIT image; however, the GI measure is applicable to other fEIT image types as well.

b. Coefficient of variation

The coefficient of variation (CV) is a statistical measure which characterizes the relative

magnitude of the standard deviation of a distribution with respect to the distribution mean.

Frerichs et al. (23) introduced the concept of using the CV to characterize the global

heterogeneity in all image pixels. CV was defined for an fEIT image as

CV = SD ( fEIT ) / Mean ( fEIT )

where SD ( fEIT ) and Mean ( fEIT ) represent the standard deviation and mean of a given

fEIT image across all image pixels.

In (23), CV was used to characterize the degree of ventilation heterogeneity during

high-frequency oscillation ventilation when compared with conventional ventilation. In the

study by Vogt et al. (24), CV was calculated from the fEIT pixel values of inspiratory vital

capacity (IVC), forced expiratory volume in 1 s (FEV1), forced vital capacity (FVC) and tidal

variation. Based on these measures, an analysis was performed to determine the power of

the EIT-based measures to discriminate between patients with obstructive pulmonary

disease and healthy subjects.

Examination-specific EIT measures

These measures have been created in order to characterize a specific aspect of thoracic

physiology, and require either specific examination procedures or measurements of EIT in

parallel with other signals, such as airway pressure (or, in some cases, both). For

convenience of description, we thus divide examination-specific EIT measures as follows:

1. EIT measures which require EIT recordings in parallel with other signals

2. EIT measures which require specific examination procedures.

10

Examples of the first type are regional respiratory system compliance or opening and

closing pressure values, while an example of the second type are measures of pulse transit

times. The types of measures are further described below.

EIT measures using simultaneously measured signals

This subgroup consists of measures that are derived from the EIT data registered in parallel

with other signals, mostly airway pressure. They aim at characterizing regional respiratory

system mechanics, with the main focus on regional respiratory system compliance. The high

EIT scan rates allow the assessment of intratidal changes in regional compliance. Another

EIT measure of respiratory mechanics is the average regional respiratory time constant

introduced by (25, 26), which is described above under "average regional EIT measures".

a. Regional respiratory system compliance

As described in EOS 4, an EIT-derived compliance fEIT image may be calculated, in which

Cj = ΔZj / ΔP

where Cj is a regional compliance value in an fEIT image pixel, ΔP is the change in pressure,

typically measured at the airways opening, and ΔZj is the corresponding change in EIT

image at pixel j. Regional compliance measures are averages or sums, or the maximum or

minimum values within regions.

Regional EIT-derived values of compliance have predominantly been analyzed under

dynamic conditions (12, 27-29), in which case the value of ΔP at the airway opening

represents not only the effect of compliance, but also that of flow-resistive pressure. This

means that a dynamic measurement of Cj is an underestimate of its true value. Several

authors have studied respiratory system mechanics during quasi-static conditions identifying

the landmark pressures at maximum compliance changes on the inflation (and sometimes

also deflation) limbs of the low-flow pressure-volume maneuver (30-33). An EIT-based

approach to determine regional respiratory system compliance in ventilated patients with

spontaneous breathing activity using a short PEEP wave maneuver was described in (34).

b. Regional intratidal gas distribution

Regional intratidal gas distribution is an approach to characterize which parts of the lungs

are filled at each stage of an inflation protocol. The goal of this EIT measure is to help

identify intratidal regional recruitment and overdistention. To calculate regional intratidal gas

distribution, the volume of gas (as reflected by the global impedance change) during

11

inflation is divided into a number of equal volume parts. During each volume part, the

fraction of volume going into each lung ROI is calculated. These data are then graphed as

the regional volume fraction vs. the inflation step.

Regional intratidal gas distribution was first introduced by Lowhagen et al. (29). In

the original study, the authors divided inflation into eight steps and the lung into four

horizontal ROIs. This work showed that the contribution of the dependent lung regions to

the inspiration increases at higher PEEP levels. The same measure was used to assess the

effect of different assist levels during pressure-support ventilation (PSV) and neurally-

adjusted ventilatory assist (NAVA) on ventilation distribution (35), however, it was analyzed

using only two (anterior and posterior) image layers. This measure was further used as a

strategy to optimize PEEP settings in experimental (36) and clinical settings (16).

EIT measures using specific examinations

a. Regional pulmonary opening/closing pressures

As described in EOS 4, the pixel opening and closing pressures can be calculated by

detecting the time that the pixel waveform crosses a threshold. This approach was

described by Pulletz et al. (37) and requires the use of a low-flow inflation and deflation

maneuver. The fEIT image for each pixel is taken from the pressure at the time that the

threshold is crossed. The use of a slow inflation and deflation means that pressures can be

considered quasi-static, and that the opening and closing pressures reflect the physiological

state of lung tissue without the flow-resistive effects. Based on the fEIT images of opening

or closing pressures, regional average values can be calculated.

b. Regional values of pulmonary function testing (PFT) measures

During conventional pulmonary function testing using spirometry or whole body

plethysmography, global lung function parameters like inspiratory vital capacity (IVC),

forced expiratory volume in 1 s (FEV1), forced vital capacity (FVC), FEV1/FVC or mean

maximum expiratory flow between 25 to 75% of FVC, are measured at the airways opening

and used to assess the lung status of the patients. EIT enables the assessment of such

measures on a regional basis by analysis of pixel EIT waveforms acquired during the

examination of full inspiration from residual volume and subsequent forced full expiration.

This approach has been used in a few studies to show the spatial and temporal

heterogeneity of ventilation in patients with cystic fibrosis, chronic obstructive lung disease

(COPD) or asthma (24, 38-42).

12

The EIT image pixel impedance values are proportional to expiratory volumes (43);

thus, derivatives of pixel value can be considered proportional to expiratory flows. Without

calibration, a direct comparison with spirometric parameters is not possible. To remove the

requirement for calibration, ratios of maximum expiratory flows at 25% and 75% of vital

capacity (MEF25/MEF75) during forced expiration were calculated for each pixel in lung

regions in the EIT images (38, 39). This fEIT image was called the “regional obstruction

map”; however, we propose the term “expiratory flow ratio fEIT image”. To create this

image, MEF25/MEF75 values are shown in each pixel of the image. Patients with obstructive

lung diseases usually have a mean MEF25/MEF75 value lower than 0.2 with high variations

among pixels (39). Similar results have been observed for the regional EIT-derived FEV1/FVC

values (24, 41, 42). Healthy subjects had the highest values of pixel FEV1/FVC with a narrow

peak in the distribution histogram. The highest degree of heterogeneity with a flat

distribution of the histogram of pixel values was noted in patients with COPD. The spatial

distribution of pixel lung function measures of IVC, FEV1, FVC quantified as the coefficient of

variation was also identified as most heterogeneous in these patients.

Another EIT measure based on PFT measures is based on the fEIT image of expiration

times. This image is composed of pixel expiration times required to exhale certain amounts

of volume of pixel FVC (e.g. 75%, denoted as t75). These measures are effective to

characterize the temporal distribution of regional lung ventilation (24, 41, 42). In COPD and

asthma, delayed and more heterogeneous emptying than in the healthy lungs was noted.

c. Overdistension / Atelectasis [%]

As described in EOS 4, an image of regional overdistension and atelectasis/collapse may be

calculated from analysis of a stepwise recruitment protocol (i.e. decremental PEEP trial after

previous recruitment). The approach was defined by Costa et al. (44) and has been used

also by (45), or a similar approach by (46). This approach classifies the fraction of

overdistended or atelectatic units in each image pixel, F(P ), at each pressure level, P, as

the fraction of maximum compliance of each pixel.

F(P ) = 1 − Cdyn(P ) / Cmax

where Cdyn(P ) is the EIT-derived dynamic compliance and Cmax is the maximum compliance

seen at that pixel throughout the maneuver. An fEIT image pixel is classified as atelectatic if

the current pressure P is below the one at which its maximum compliance occurs, while it is

classified as overdistended in the converse case. This fEIT measure thus yields a fractional

state of each image pixel. In order to calculate an EIT measure of global or regional

13

overdistension or atelectasis, the average value in the lung ROI (or an image region) is

taken.

d. Measures of pulse transit time

Solà et al. defined a measure of pulse transit time from the region of the heart to the

descending aorta (47). Pulse transit time measures are correlated with systemic blood

pressure; however, typical measures use a peripheral site which means that the measure is

modulated by vascular tone in the peripheral arteries, as well as other factors. In this work,

the authors demonstrated that EIT is able to measure pulse transit time centrally, in a

location (the aorta) where no modulation by vascular tone occurs. First, ROIs are identified

corresponding to the heart and aorta, and in the EIT raw image sequence, pixel waveforms

are calculated for these ROIs. The pulse transit time is then the time difference between the

cardiac pulse in the waveform in the aortic ROI compared to that in the heart region.

Because the aortic region is small, its signal is relatively weak. In order to improve the

robustness of the measure, the average pulse transit time in the EIT waveform was

calculated by robust parametric detection algorithm. This work showed a strong correlation

between EIT measures of pulse transit time, and arterial blood pressure measurements in

an experimental animal.

e. Measures of perfusion with contrast agents

Cardiac output and pulmonary blood flow represent important physiological parameters with

clinical significance for patient management. As discussed in EOS 4, EIT signals contain a

pulsatility component, which is associated with the contribution of blood flow at the cardiac

rate to the EIT signals. The pulsatility gives useful information on the perfusion, but can

deviate from it for several reasons (see EOS 4).

In order to obtain more accurate measures of regional lung perfusion, it is necessary

to use EIT contrast agents. This approach was initially proposed by Frerichs et al. (48), in

which a bolus of hypertonic saline was administered to visualize blood flow through lung

regions. In the original work boli of 15 or 20ml of 5.85% NaCl were injected through the

proximal or distal openings of a Swan-Ganz catheter. Figure E5.6 shows the EIT waveforms

in two pixels following a saline bolus. There is a clear delay between the signal in the heart

and lungs due to the propagation time through the pulmonary arteries.

Additional work with contrast agent-based perfusion measures was done by Borges et

al. (49). Since the electrical conductivity of fluid is determined by the ionic concentration,

hypertonic saline will increase the conductivity and thus be visible in the raw EIT images

14

(50). Some work to consider other contrast agents, such as temperature decreases or

sucrose solutions (which decrease conductivity) has not been successful (51). While such

images offer extremely useful information, their utility is limited by the relative invasiveness

of the procedure. It is possible only in patients who already have a central catheter, and can

be repeated only as often as is safe considering the effects of the additional saline on the

patient.

Figure E5.6. Regional EIT waveforms obtained in one pixel located in the anterior heart

and posterior lung region following a bolus injection of 10% NaCl solution via a central

venous line. The location of the pixels is indicated in the fEIT standard deviation image (left)

with small squares and waveform numbers. The data were obtained in a mechanically

ventilated supine pig using the Goe-MF II EIT device (CareFusion, Höchberg, Germany)

before and after a breath-hold and bolus injection (52). The waveforms show the regional

fall in impedance resulting from the passage of electrically conductive saline through the

heart and vessels. Note the earlier onset of in the heart waveform as compared to the

posterior lung. The dashed lines visualize the time delay.


The first draft of this online document was prepared by A. Adler with collaboration of I.

Frerichs, Z. Zhao, B. Grychtol, S. Leonhardt and D. Gommers. It was reviewed and approved

by all other authors and collaborators.

15

References



Physiol 2001;91:39-50.

2. Heinrich S, Schiffmann H, Frerichs A, Klockgether-Radke A, Frerichs I. Body and

head position effects on regional lung ventilation in infants: An electrical impedance


3. Schnidrig S, Casaulta C, Schibler A, Riedel T. Influence of end-expiratory level and

tidal volume on gravitational ventilation distribution during tidal breathing in healthy adults.

Eur J Appl Physiol 2013;113:591-598.




5. Steinmann D, Stahl CA, Minner J, Schumann S, Loop T, Kirschbaum A, Priebe HJ,

Guttmann J. Electrical impedance tomography to confirm correct placement of double-lumen

tube: A feasibility study. Br J Anaesth 2008;101:411-418.










9. Schibler A, Yuill M, Parsley C, Pham T, Gilshenan K, Dakin C. Regional ventilation

distribution in non-sedated spontaneously breathing newborns and adults is not different.

Pediatr Pulmonol 2009;44:851-858.

10. Miedema M, de Jongh FH, Frerichs I, van Veenendaal MB, van Kaam AH. Changes

in lung volume and ventilation during surfactant treatment in ventilated preterm infants. Am J


11. Karsten J, Luepschen H, Grossherr M, Bruch HP, Leonhardt S, Gehring H, Meier T.

Effect of PEEP on regional ventilation during laparoscopic surgery monitored by electrical


16



bedside during a decremental positive end-expiratory lung pressure trial. Crit Care

2011;15:R193.

13. Wolf GK, Grychtol B, Frerichs I, Zurakowski D, Arnold JH. Regional lung volume

changes during high-frequency oscillatory ventilation. Pediatr Crit Care Med 2010;11:610-

615.

14. Radke OC, Schneider T, Heller AR, Koch T. Spontaneous breathing during general

anesthesia prevents the ventral redistribution of ventilation as detected by electrical

impedance tomography: A randomized trial. Anesthesiology 2012;116:1227-1234.

15. Luepschen H, Meier T, Grossherr M, Leibecke T, Karsten J, Leonhardt S. Protective

ventilation using electrical impedance tomography. Physiol Meas 2007;28:S247-260.

16. Blankman P, Hasan D, Groot Jebbink E, Gommers D. Detection of 'best' positive

end-expiratory pressure derived from electrical impedance tomography parameters during a

decremental positive end-expiratory pressure trial. Crit Care 2014;18:R95.




18. Pham TM, Yuill M, Dakin C, Schibler A. Regional ventilation distribution in the first 6

months of life. Eur Respir J 2011;37:919-924.

19. Rooney D, Friese M, Fraser JF, K RD, Schibler A. Gravity-dependent ventilation

distribution in rats measured with electrical impedance tomography. Physiol Meas

2009;30:1075-1085.




21. Zhao Z, Steinmann D, Frerichs I, Guttmann J, Moller K. PEEP titration guided by

ventilation homogeneity: A feasibility study using electrical impedance tomography. Cril Care

2010;14:R8.

22. Becher T, Kott M, Schädler D, Vogt B, Meinel T, Weiler N, Frerichs I. Influence of

tidal volume on ventilation inhomogeneity assessed by electrical impedance tomography

during controlled mechanical ventilation. Physiol Meas 2015;36:1137-1146.




17












1961.







30. van Genderingen HR, van Vught AJ, Jansen JR. Estimation of regional lung volume

changes by electrical impedance pressures tomography during a pressure-volume

maneuver. Intensive Care Med 2003;29:233-240.



Meas 2013;34:567-577.

32. Grychtol B, Wolf GK, Arnold JH. Differences in regional pulmonary pressure-

impedance curves before and after lung injury assessed with a novel algorithm. Physiol

Meas 2009;30:S137-148.

33. Hinz J, Moerer O, Neumann P, Dudykevych T, Frerichs I, Hellige G, Quintel M.

Regional pulmonary pressure volume curves in mechanically ventilated patients with acute

respiratory failure measured by electrical impedance tomography. Acta Anaesthesiol Scand

2006;50:331-339.

34. Becher TH, Bui S, Zick G, Blaser D, Schadler D, Weiler N, Frerichs I. Assessment of

respiratory system compliance with electrical impedance tomography using a positive end-

expiratory pressure wave maneuver during pressure support ventilation: A pilot clinical study.

Crit Care 2014;18:679.

18

35. Blankman P, Hasan D, van Mourik MS, Gommers D. Ventilation distribution

measured with EIT at varying levels of pressure support and neurally adjusted ventilatory

assist in patients with ali. Intensive Care Med 2013;39:1057-1062.

36. Bikker IG, Blankman P, Specht P, Bakker J, Gommers D. Global and regional

parameters to visualize the 'best' PEEP during a peep trial in a porcine model with and

without acute lung injury. Minerva Anestesiol 2013;79:983-992.



Crit Care 2012;27:323 e311-328.






40. Vogt B, Elke G, von Bismarck P, Ankermann T, Weiler N, Frerichs I. Regional

pulmonary function testing by electrical impedance tomography in healthy children and

children with asthma. Eur Respir J 2012;40 Suppl. 56:3296.







43. Marquis F, Coulombe N, Costa R, Gagnon H, Guardo R, Skrobik Y. Electrical

impedance tomography's correlation to lung volume is not influenced by anthropometric

parameters. J Clin Monit Comput 2006;20:201-207.






2012;31:834-842.




19

47. Sola J, Adler A, Santos A, Tusman G, Sipmann FS, Bohm SH. Non-invasive

monitoring of central blood pressure by electrical impedance tomography: First experimental

evidence. Med Biol Eng Comput 2011;49:409-415.

48. Frerichs I, Hinz J, Herrmann P, Weisser G, Hahn G, Quintel M, Hellige G. Regional

lung perfusion as determined by electrical impedance tomography in comparison with

electron beam CT imaging. IEEE Trans Med Imaging 2002;21:646-652.

49. Borges JB, Suarez-Sipmann F, Bohm SH, Tusman G, Melo A, Maripuu E,

Sandstrom M, Park M, Costa EL, Hedenstierna G, Amato M. Regional lung perfusion

estimated by electrical impedance tomography in a piglet model of lung collapse. J Appl

Physiol 2012;112:225-236.

50. Brown BH, Leathard A, Sinton A, McArdle FJ, Smith RW, Barber DC. Blood flow

imaging using electrical impedance tomography. Clin Phys Physiol Meas 1992;13 Suppl

A:175-179.

51. Mamatjan Y, Grychtol B, Gaggero P, Justiz J, Koch VM, Adler A. Evaluation and

real-time monitoring of data quality in electrical impedance tomography. IEEE Trans Med

Imaging 2013;32:1997-2005.

52. Elke G, Hintze C, Meybohm P, Biederer J, Weiler N, Frerichs I. Regional lung

perfusion measured by electrical impedance tomography – a comparison of the bolus

contrast and frequency-filtering approaches. Am J Respir Crit Care Med 2013;187:A1531.


terminology, clinical use and recommendations: consensus statement of the

TRanslational EIT developmeNt stuDy group



Ola Stenqvist, Tommaso Mauri, Vinicius Torsani, Luigi Camporota, Andreas Schibler, Gerhard K.

Wolf, Diederik Gommers, Steffen Leonhardt, Andy Adler, TREND study group

ONLINE SUPPLEMENT 6

EIT definitions and nomenclature

2

EIT definitions

Table E6.1 EIT terms and definitions Term Alternative

terms Abbreviation(Alternative abbreviation)

Explanation

Absolute EIT aEIT (A-EIT) An EIT imaging mode in which each image of the impedance distribution is calculated from a single frame of EIT data. It differs from a difference EIT image (tdEIT or fdEIT) which are calculated from two measurement frames. aEIT is significantly more sensitive than difference EIT to electrode contact quality, movement and EIT system changes, and based on current technology, is currently not yet ready for clinical use.

Adjacent current stimulation pattern

Sheffield stimulation pattern

A current stimulation pattern in which electrical current is sent sequentially through adjacent pairs of electrodes. For example, first, electrodes 1 and 2 are used, followed by 2 and 3, and so on. Historically, this is the most common current stimulation pattern.

Aeration change image

Aeration change map

The Aeration change map is an EIT image that represents the change in regional distribution of lung impedance between two points in time. When regional values at end-expiratory time points are compared the corresponding Aeration change image shows regional changes in end-expiratory lung volumes.

Anterior-to-posterior ventilation ratio

Upper-to-lower ventilation ratio

A/P ratio (U/L ratio) A functional EIT measure used to quantify the distribution of ventilation in the antero-posterior direction. A/P ratio is calculated as the ratio of A (the sum of the ventilation measured in the image pixels in anterior (ventral) half of the image) to P (the sum of the ventilation measured in the image pixels in the posterior (dorsal) half of the image). (In supine posture, the anterior equals the upper, non-dependent and the posterior the lower, dependent regions.)

Applied potential tomography

APT An obsolete name of electrical impedance tomography used until the early nineties of the last century.

Arbitrary units A.U. (AU) The measurement unit commonly used in EIT to quantify the amplitude of impedance changes within EIT image pixels. Current tdEIT image reconstruction algorithms do not produce images with well-defined units, and their sensitivity can depend on many factors such as subject posture and electrode placement details. It is thus common to express the image units as “arbitrary”. As an alternative to A.U., EIT images can be calibrated to the patient and measurement configuration in order to generate physiological image units.

3

Term Alternative terms

Abbreviation(Alternative abbreviation)

Explanation

Baseline

Baseline measurement, (Reference, Reference measurement)

BL Time difference EIT determines impedance changes against a reference (or baseline) measurement. Various choices of BL measurement can be used, depending on the nature of clinical information to be derived. Its definition has a big impact on the calculation of impedance changes and derived EIT based parameters. Some common definitions of baseline are: mean of EIT data, data at end-expiration or end-inspiration, a moving average of EIT data or a moving average of end-expiratory events.

Center of ventilation

Center of gravity

CoV A functional EIT measure used to quantify the distribution of ventilation in relation to the anteroposterior or right-to-left chest diameter and expressed as percentage. The geometric center of fEIT pixels is calculated and expressed as a fraction of the vertical or horizontal image size. In the former case, values lower than 50% imply that ventilation is predominantly directed towards ventral (anterior) regions, in the latter one towards the right lung.

Change in end-expiratory lung impedance

End-expiratory lung impedance change (Change in end-expiratory impedance)

ΔEELZ (ΔEELI)

The difference in end-expiratory EIT values between two points in time.

Chest EIT examination

Chest EIT imaging (chest EIT scanning)

Use of EIT with electrode placement on the chest to obtain information on regional lung ventilation, aeration and function and/or heart action and/or lung perfusion.

Current stimulation frequency

Stimulation frequency

Frequency of the alternating drive current applied to the body through the electrodes. Common devices use frequencies of 10 kHz - 250 kHz. The measured tissue impedance properties vary with current stimulation frequency.

Current stimulation pattern

Current injection pattern

The configuration and order in which electrical current is input into electrodes by the EIT device. Most EIT systems use a “pair drive” current stimulation pattern, in which current flows through a pair of electrodes, while the other electrodes are passive. During each current stimulation a set of voltage measurements is acquired using the measurement pattern.

Differential image Vendor-specific term for an image generated by subtraction of two functional images characterizing tidal ventilation distribution. The differential image can show positive or negative values indicating an increase or decrease in regional tidal ventilation over time.

4



Explanation

Dorsal ROI Posterior ROI D (ROI 4) The region of interest within an EIT image that represents the back or spinal part of the body. When an EIT image is subdivided into 4 equal horizontal layers, the dorsal (posterior) part is the lowermost layer in the image.

Dynamic image Real-time display of continuously updated relative impedance changes within the electrode plane shown as a series of tomograms in a movie-like style. The relative impedance changes are calculated using a baseline either at end of the last expiration, a moving average measurement, or a different user-selected value. (In the Sheffield Mark I device, this term was a synonym for time-difference image.)

EIT Electrical impedance tomography. EIT refers to any method of generating tomographic images from electrical impedance measurements made with body surface electrodes.

EIT data file File created by an EIT device, which contains all voltage measurements before data processing and image reconstruction. Many formats of EIT data files have been proposed. Some formats contain additional information, including reconstructed EIT images, details of the measurement configuration, patient details or other monitoring data.

EIT device The electronic system which connects to the electrodes attached to the body surface, applies EIT stimulation currents and performs and records EIT measurements.

EIT electrode EIT sensor EIT electrodes provide electrical contact between the EIT device and the body. Many different types of electrodes are used, including individual ECG-type electrodes and electrodes integrated into belts or stripes. Some devices use active electrodes, in which electronic amplifiers and other hardware are physically placed on or near the electrode.

EIT examination The process of performing EIT measurements on a subject.

EIT image Collective term for any kind of image created from EIT measurements, including dynamic, frequency difference, functional, raw, or time-difference images.

EIT image pixel A pixel (or picture element) is the smallest image element in a 2D EIT image. EIT images are calculated on a grid of image pixels. For 3D EIT images, the corresponding elements are image voxels.

EIT measurement 1) The process of EIT data acquisition.2) The measured data resulting from an EIT data acquisition.

EIT sensitivity region

The roughly lens-shaped intra-thoracic volume the impedance changes of which contribute to the generated EIT images. The thickness of this plane depends on the placement and size of the electrodes used, the dimension, the bioelectric properties, and the shape of the thorax, and particularly on the morphological structures within the chest. The sensitivity region is relatively small close to the periphery, but increases in size towards the center.

5



Explanation

EIT raw data EIT data set (EIT data file)

Voltages obtained during EIT measurement together with the measurement settings. The set of raw data acquired during one cycle of current applications and voltage measurements is typically called frame.

Electrode belt

Sensor belt, electrode stripe, patient interface

The patient interface onto which EIT electrodes are mounted to facilitate placement on a patient´s body. The form is specific to the EIT device, and can take the form of a belt or a more complex arrangement. It contains all components needed to attach EIT electrodes to the patient, including the electrodes themselves.

Electrode numbering

The numerical order in which body surface electrodes are addressed by the EIT device. In most applications, EIT electrodes are enumerated sequentially from the sternum to the left and subsequently the right hemi-thorax.

Electrode skin contact impedance

The impedance to electrical current flowing across the contact between the electrode and the body. It is affected by the skin properties, skin preparation (including contact agents), and electrode contact area. Low electrode skin contact impedance is necessary to obtain high quality EIT recordings. Many EIT devices provide continuous information on this value, in order to ensure data quality.

Electrode plane The plane intersecting the centers of the body surface EIT electrodes. The most common ones are the transverse or the oblique thoracic planes.

Frame Scan A complete set of voltage measurements collected by application of one complete cycle of current stimulations in a specified pattern. A frame of EIT data is used to reconstruct one raw EIT image.

Frequency-difference EIT

fdEIT (FD-EIT) An EIT imaging mode in which two data frames are acquired at different current stimulation frequencies. The reconstructed image represents the difference in tissue properties between the two stimulation frequencies. fdEIT requires an EIT system capable of multi-frequency operation.

Functional EIT image

fEIT image An EIT image calculated from a time-series of EIT images, designed to characterize a particular physiological feature. Each fEIT image pixel is calculated from that pixel’s time-series values. Many types of fEIT images have been defined. Two broad categories can be identified: 1) fEIT images based on the EIT data alone, such as tidal images, and 2) fEIT images based on EIT and additional monitoring data, such as images of regional opening pressures.

Global Global information

A value calculated from an EIT image, which represents the entire EIT plane. The global value can be calculated either as the sum or the average of the pixel values in the EIT image.

6



Explanation

Global change inend-expiratory lung impedance

ΔEELIglobal The global value of an aeration change image generated from differences in end-expiratory EIT values between two points in time. (Some vendors normalize the value and calculate it as the fraction of tidal impedance variation at the first of the two time points.)

Global impedance signal

Global impedance waveform, (Global impedance curve, Global impedance time course)

The global impedance signal is the global impedance change as a function of time.

Global minute tidal variation

MTVglobal The average of the global tidal variation over the previous minute.

Global tidal variation

Global tidal impedance variation, Global ventilation-related impedance change

TVglobal

(TIVglobal)

The global value of a tidal variation image.

Image orientation EIT images are shown in the conventional medical orientation, i.e. looking at them from the patient´s feet towards the head. This means that anterior sections of the thorax are imaged at the top and its right side on the left side of the image.

Impedance Z A measure of the opposition that an electrical circuit (for EIT, the body tissue) presents to the stimulation current. Gases (e.g. air in the lungs) have high impedance, while fluids with high ion concentrations (e.g. blood and saline) have low impedance. Electrical impedance is measured as a magnitude and a phase offset with respect to the stimulation current. Most current tdEIT image reconstruction algorithms currently do not use the phase offset value.

Impedance change

Impedance difference

∆Z Changes in the value of impedance.

Measurement pattern

The spatial configuration and order by which voltage measurements are performed by the EIT device during current stimulation with a particular stimulation pattern. Most EIT systems make measurements only on passive electrodes (i.e. those which are not used for current stimulation).

7



Explanation

Mid-dorsal ROI Mid-posterior ROI

MD (ROI 3) The region of interest within an EIT image that represents the body part above the most dorsal (posterior) section of the body. When an EIT image is subdivided into 4 equal horizontal layers, the mid-dorsal layer is represented by ROI 3 (i.e. above ROI 4).

Mid-ventral ROI Mid-anterior ROI

MV (ROI 2) The region of interest within an EIT image that represents the body part below the most ventral (anterior) section of the body. When an EIT image is subdivided into 4 equal horizontal layers, the mid-ventral part is represented by ROI 2 (i.e. below ROI 1).

Minute image Vendor-specific term which characterizes tidal volume distribution. It displays pixel tidal impedance variation values averaged over one minute.

Normalized tidal variation

Relative tidal stretch

Tidal variation normalized to maximum tidal variation in the image. Image regions with very low values of normalized tidal variation imply low ventilation. (In a vendor-specific taxonomy, pixels with values <10% are called ‘silent spaces’.)

Pixel The smallest element in a digital image. EIT images are reconstructed and displayed with an image size that depends on the algorithm and EIT system. The image size is represented by the number of horizontal pixel rows and vertical pixel columns (e.g. 32x32).

Pixel impedance signal

The signal obtained by representing the calculated value in a single image pixel over the time for which it was measured.

Pixel impedance tidal variation

The magnitude of the variations in a pixel impedance signal due to tidal ventilation.

Stimulation current Excitation current (Drive current, Injection current)

The electrical current applied to the body during EIT scanning. Currents are applied at the current stimulation frequency across pairs of stimulation electrodes in order to measure the resulting voltage on the body surface from which the impedance is then calculated. Stimulation currents are designed to be below the human threshold of perception, and safe against macroshock.

Stimulation pattern Drive pattern The spatial pattern or order by which electrodes are used for current stimulation. Pair drive EIT systems apply current across one pair of electrodes at a time. The adjacent current stimulation pattern (the most common) applies current sequentially through adjacent pairs of electrodes.

8



Explanation

Region of interest ROI An area within an image on which further analysis is performed. Based on the definition of the ROIs, regional information expressed as waveforms or parameters can be displayed. ROIs can be defined by user definition, such as horizontal layers or quadrants, or automatically, such as the ventilated region of an image. The smallest possible ROI is a single image pixel. Anatomy-related ROIs (i.e. lung contours) can be derived from other medical imaging techniques. For layer- and quadrant-based ROI definitions, the following conventions are used: when arranged as horizontal layers, ROI 1 represents the ventral (anterior) part; when arranged as quadrants, ROI 1 represents the upper left, ROI 2 the upper right, ROI 3 the lower left, ROI 4 the lower right quadrant.

Regional Regional information

Information derived from a predefined ROI within an EIT image.

Regional change in end-expiratory lung impedance

ΔEELIROI The regional value of an aeration change image generated from differences in end-expiratory EIT values between two points in time. (Some vendors normalize the value and calculate it as the fraction of tidal impedance variation at the first of the two time points.)

Regional impedance signal

Regional impedance waveform, (Regional impedance curve, Regional impedance time course)

The regional impedance signal is the impedance change in a ROI as a function of time.

Regional minute tidal variation

MTVROI The average of the regional tidal variation over the previous minute.

Regional tidal variation

TVROI The regional value of a tidal variation image.

Relative impedance change

Normalized impedance difference

rel. ∆Z A measurement unit commonly used to quantify the amplitude of impedance changes within EIT image pixels. A synonym to arbitrary units (A.U.). The term “relative” signifies that images are reconstructed with respect to a baseline and are not presented in any well-defined physiological units.

9



Explanation

Right-to-left ventilation ratio

R/L ratio A functional EIT measure used to quantify the distribution of ventilation in the right-to-left direction. R/L ratio is calculated as the ratio of R (the sum of ventilation image pixels in the left half of the image) to L (the sum of ventilation image pixels in the right half of the image). (EIT uses the conventional radiological image orientation with the right side of the body shown on the left side of the image and vice versa.)

Scan Frame A complete set of EIT measured data from which a single image can be reconstructed. A scan requires measurement at all electrodes in the measurement patterns for each stimulation pattern.

Scan rate Frame rate The scan rate is the frequency at which EIT scan data are measured and expressed as frames (or scans) per second. Typical scan rates used to monitor ventilation are between 20 and 50 images per second; however, there can sometimes be a trade-off between EIT data quality and high scan rates.

Signal quality SQ Indicator of the accuracy and reliability of displayed EIT data. Low quality EIT data can result in poor or erroneous images. There are several common causes of impaired signal, such as poor electrode-skin contact, patient movement, interference from other medical devices, and other electromagnetic noise. There is currently no standard for the calculation of signal quality.

Status image Vendor-specific term which characterizes tidal volume distribution. It can reflect the instantaneous values of tidal impedance variation, or their averaged values over one minute.

Tidal variation Difference between the EIT data at end-inspiration and end-expiration for an individual breath.

Tidal variation image

Tidal image (tidal ventilation map)

An image representing the regional distribution of tidal volume, calculated as difference of the EIT images at end-inspiration and end-expiration for an individual breath.

Time-difference EIT

tdEIT (TD-EIT) An EIT imaging mode based on data acquired in sequence over time. The reconstructed image represents the difference in tissue properties between a measurement at a particular time and a baseline measurement. tdEIT is the most common EIT imaging mode.

Ventral ROI Anterior ROI V (ROI 1) The region of interest within an EIT image that represents the anterior or sternal part of the body. When an EIT image is subdivided into 4 equal horizontal layers, the ventral part is the uppermost layer in the image represented by ROI 1.

10

Table E6.2 Medical terms relevant for chest EIT Term Alternative

terms Abbreviation(Alternative Abbreviation)

Explanation

Aeration Volume of gas contained within the lung at a given moment of time. (EIT allows the assessment of regional aeration changes induced e.g. by variation in positive end-expiratory pressure (PEEP).)

Atelectasis Lung collapse A condition where alveoli are deflated and collapsed which may be caused by a blockage of airways and subsequent gas resorption and / or excessive external pressure on the alveoli. (Regional EIT signals in atelectatic lung regions do not exhibit ventilation-related impedance changes or these are very low.)

Alveolar recruitment

Alveolar recruitment describes the sustainable re-expansion of previously collapsed alveoli which results in aeration and ventilation of the recruited parts of the lung. Re-expansion of collapsed lung tissue is one of the primary goals of respiratory care and is aimed at improving pulmonary gas exchange and lung mechanics, but also at protecting the lungs from ventilator-associated lung injury. (In EIT, regional alveolar recruitment can be identified by an increase in regional end-expiratory impedance and by the re-occurrence of previously absent ventilation-related impedance changes in regional waveforms e.g. after a PEEP increment.)

Derecruitment Atelectasis, lung collapse

Loss of end-expiratory lung volume leading to a lack of gas and respective ventilation within a lung area. Strictly speaking, atelectasis or lung collapse is the result of derecruitment but in clinical practice all terms are used interchangeably. (In EIT data, the effects of derecruitment can be observed during a decremental PEEP trial that includes sufficiently high initial inspiratory pressures and PEEP levels, where the recruitable parts of the lung are opened. At a certain PEEP level further PEEP reduction may cause a decrease of regional respiratory system compliance which is then interpreted as derecruitment.)

Dependent lung Area of the lungs most exposed to gravity induced pressure, where the lung itself (and other intrathoracic organs like the heart) above this area apply a superimposed pressure. In supine position, the dependent lung regions are located in the dorsal part of the lung.

11


Abbreviation(Alternative Abbreviation)

Explanation

Dorsal Posterior D Related to the position of the subject’s spine. In EIT imaging of a subject in the supine position, the lower aspect of the image represents the part closest to the subject’s spine.

End-expiratory lung volume

EELV Sometimes synonymously used with the term functional residual capacity (FRC). FRC equals the volume of gas in the lungs by the end of passive tidal expiration in spontaneously breathing subjects. At this volume, the elastic recoil forces of the lungs and chest wall are in equilibrium. However, mechanically ventilated patients exhale against PEEP rather than ambient pressure, which is why the term end-expiratory lung volume (EELV) should be used instead of FRC; the latter term to be used for subjects breathing at ambient pressure, only. EELV (resp. FRC) describes the gas volume which can contribute to gas exchange between two breaths. Adequate PEEP settings during mechanical ventilation help in maintaining a sufficient EELV and thus in keeping alveoli and airways open. (EIT can trace regional changes in EELV elicited for instance by PEEP variation.)

Functional residual capacity

FRC The gas volume present within the lungs at the end of a passive expiration against ambient pressure.

Non-dependent lung

Area of the lungs without superimposed pressure caused by gravity. In supine position, the non-dependent lung regions are located in the ventral (anterior) part of the lung.

Overdistension Overinflation, Hyperinflation, Hyper-distension

Excessive expansion of the lungs at the end of inspiration, commonly caused by either high tidal volumes or high end-expiratory lung volumes resulting from high PEEP levels. (In EIT, the presence of overdistension must be anticipated whenever a lack of ventilation is observed in the non-dependent lung areas. However, overdistension is usually systematically diagnosed by a decremental PEEP trial, causing regional compliance to increase after PEEP reduction which is then interpreted as the presence of overdistension at the previous higher PEEP levels.)

Peak inspiratory pressure

PIP Maximum pressure applied during positive pressure ventilation to deliver air into the lungs.

12



Explanation

Positive end-expiratory pressure

PEEP Pressure at end expiration used in ventilated patients to oppose passive emptying of the lung and to keep the airway pressure above the atmospheric pressure. PEEP is used to maintain a sufficiently large end-expiratory lung volume and to keep airways open. (PEEP-induced changes in regional aeration can be mapped by EIT.)

PEEP titration PEEP trial Intervention aimed at identifying the most appropriate level of PEEP for an individual mechanically ventilated patient. PEEP is usually titrated systematically downward starting at high levels. Often, a recruitment maneuver is performed prior to this intervention to assure, that recruitable lung regions are open before the titration is started. PEEP can also be titrated upward to find the best PEEP to keep airways open.

Plateau pressure Pplat Pressure that is present during the inspiratory plateau phase.

Recruitment maneuver

RM Intervention, usually performed with a mechanical ventilator, aimed at re-expanding collapsed lung tissue. In order to recruit collapsed lung tissue, sufficiently high peak pressures and PEEP levels must be temporarily imposed to exceed the critical opening pressure of the affected lung region. After the recruitment maneuver PEEP levels must be kept high enough to prevent subsequent derecruitment. Recruitment maneuvers also have a time related component as the time required to open (heterogeneous) lung regions varies. (The effects of a recruitment maneuvers on regional lung aeration and ventilation can be assessed by EIT during both conventional and high-frequency oscillatory ventilation.)

Tidal recruitment Cyclic opening and closing (repeated alveolar opening and collapse)

Describes a condition where alveoli and small airways collapse during each expiration and are re-opened with each inspiration. This condition can cause ventilator associated lung injury (VALI). (EIT can estimate the amount of tidal recruitment by analysing regional ventilation delay inhomogeneity and regional respiratory system compliance.)

Ventilation Movement of gas into and out of the lung during breathing. (EIT can trace ventilation-induced variation in regional EIT waveforms.)

13



Explanation

Ventilator-associated lung injury

VALI Lung injury that resembles acute respiratory distress syndrome (ARDS) and that occurs in patients receiving mechanical ventilation. VALI may be associated with pre-existing lung pathology such as ARDS. However, due to the pre-existing lung pathology it cannot be determined whether VALI was caused by mechanical ventilation. Therefore it is considered associated with rather than caused by mechanical ventilation. (VALI modifies regional respiratory system mechanics and leads to regional changes in aeration and ventilation that can be mapped by EIT. The impact of therapy and/or changed ventilator settings can be assessed by EIT.)

Ventilator-induced lung injury

VILI Acute lung injury directly induced by mechanical ventilation in animal models. Since VILI is usually indistinguishable from the diffuse alveolar damage of other causes, it can only be discerned definitely in animal models. (VILI has been used as a model of acute lung injury in multiple EIT studies. Because VILI results in typical changes in regional aeration and ventilation these can be tracked by EIT.)

14


The first draft of this online document was prepared by E. Teschner, S. H. Böhm, I. Frerichs and

A. Adler. It was reviewed and approved by all other authors and collaborators.








ONLINE SUPPLEMENT 7

Clinical use of EIT in adult patients

2

Clinical use of EIT in adult patients

Introduction

This electronic online supplement (EOS) 7 provides a structured overview of clinical issues

pertaining to adult patients examined by EIT in a clinical setting mainly in the fields of

mechanical ventilation monitoring and pulmonary function testing as mentioned in the main

document. These and other clinical issues are listed in three tables (E7.1-E7.3).

All tables have the same basic structure: they show a brief description how the EIT

data specific for each issue need to be acquired and analyzed. In some cases, specific

ventilation maneuvers like a positive end-expiratory pressure (PEEP) trial or constant low-

flow inflation are needed to generate the information necessary for the clinical issue in

question to be answered. These maneuvers are specified when applicable. The typical EIT

measures generated by the individual analyses are also provided. Finally, hints on the

interpretation of the findings and their use in clinical decision making are made available.

The relevant references to clinical studies involving human subjects are provided in the last

column of each table.

We only refer to those EIT studies here that were conducted in adult patients. We do

not mention studies performed in neonatal and pediatric patients, albeit these often address

clinical issues (e.g. EIT-based guidance of mechanical ventilation) relevant for the adult

population as well. With this respect, we kindly ask the readers to also consult the EOS8

where the specific information on the EIT use in neonates and children is presented in detail.

The clinical EIT studies carried out in adult subjects most often involved patients

suffering from various lung diseases. However, patients with other diseases not involving the

respiratory system were examined as well, for instance anesthetized patients undergoing

abdominal or cardiac surgery.

The study cohorts in the EIT clinical trials are generally rather small, typically about

10-20 patients. (The exact numbers of studied subjects are stated in the tables.) In addition

to the patients´ studies, we provide information and references to studies on healthy

volunteers, especially in those cases where clinical trials on patients are lacking or when

methodological concepts in EIT data acquisition, analysis and interpretation relevant for the

postulated later clinical use were described in these healthy study populations. We also refer

to a few pioneering animal experimental studies with clinically relevant findings.

3

EIT use for monitoring lung ventilation

The use of EIT for monitoring regional lung ventilation has been the most studied clinical EIT

application so far. Its major benefit is seen in the continuous bedside assessment of regional

lung aeration and ventilation in patients undergoing ventilator therapy. The primary aim of

EIT use in these patients is the identification of deleterious phenomena like regional

overdistension, derecruitment, cyclic alveolar collapse and EIT-guided optimization of

ventilator settings minimizing or eliminating their occurrence and preventing the

development of ventilator-associated lung injury. We identify several clinical issues relevant

in this context and present the so far proposed and studied EIT approaches to address them

in Table E7.1 presented below.

4

Clinical issue

Method of analysis / EIT measure

Brief description / EIT findings

Maneuvers / interventions besides tidal ventilation

Normal vs. pathological results (if applicable)

Suggested impact on clinical decisions

References / assessment of evidence

Pulmonary ventilation / monitoring of ventilation / ventilator-induced lung injury

Lung recruitment

Regional tidal ΔZ variation (1-6) with the calculation of U/L ratio, dorsal displacement of the center of ventilation (7-10), increase in regional Crs in the most dependent lung region (10-12), intratidal gas distribution in dorsal lung regions (13)

After an increase in PEEP, Pinsp or low-flow inflation maneuver, lung recruitment is confirmed by the increase in regional dorsal lung ventilation assessed by different indices of regional ventilation distribution.

Step increments in PEEP with a fixed driving pressure until lung opening pressures are reached. Low-flow inflation maneuver. Open-lung ventila-tion approach. Assisted ventilation mode.

In lung pathology or with inadequate ventilator settings lung ventilation tends to decrease in dependent lung regions.

EIT offers the potential to monitor lung recruitment and individualize opening pressures at the bedside.

(2) n=16 patients (3) n=20 patients (4) n= 10 patients (7) n=10 patients (13) n=16 patients (14) n=12 patients (5) n=15 patients (6) n=20 patients (10) n=5 patients (1) n=6 pigs (8) n=6 pigs (9) n=16 pigs (11) n=12 pigs (12) n=16 pigs

Regional delay in ventilation / regional opening and closing pressures

During a low-flow inflation maneuver, the delay until an increase in regional impedance above a set threshold is detected. The time delay between this time point and the actual beginning of the inflation is then calculated in each pixel or ROI. Simultaneous airway pressure measurement allows the assignment of the delay to a respective regional opening pressure.

Constant low-flow inflation (i.e. 2 l/min until 1.5 l are inflated or airway pressure of 50 mbar is reached) (15), step increase and decrease in airway pressure (16)

Normal: synchronous onset of regional filling, pathological: regional delay in ventilation implying recruitment, amount of delay / assignment to parallel pressure measurement indi-cating recruitability: long delay implies high pressure necessary to recruit, short delay increases the probability of tidal recruitment

Exclude pleural effusion or reversible bronchial occlusion as a reason for delay. Use high PEEP strategy when recruitment occurs early (open up the lung and keep it open) but do not try to open the lungs if the necessary ventilator invasiveness exceeds ARDS network limits.

(15) n=26 patients (16) n=8 patients (17) n=16 pigs (18) n=12 pigs

5

Clinical issue







‘Best‘ PEEP

Decreasing or increasing or stable end-expiratory lung impedance change (EELZ)

During an incremental and decremental PEEP trial, the EELZ trend is followed over several breaths. An increa-se in EELZ reflects ongoing recruitment or overdisten-sion (PEEP too high or too low), a decrease shows de-recruitment (PEEP is too low), stable EELZ implies no recruitment or de-recruitment (this does not exclude unrecruited, fixed atelectasis or over-distension).

PEEP trial (incremental / decremental)

Normal: steady EELZ, pathological: increasing or decreasing EELZ from breath to breath

Use PEEP with stable EELZ during a decremental PEEP trial

(19) n=10 patients (20) n=12 patients (21) n=15 patients (22) n=20 patients (23) n=1 pig (lavage, surfactant) (1) n=6 pigs (lavage model)

Homogeneity of ventilation (GI index)

Fall in the GI index along with increased percentage of open lung regions during inflation imply recruitability (24). GI index falls at the high compared with the low PEEP (below LIP) (25).

Constant low-flow inflation (24) Variation in VT and PEEP (25)

GI index is lower in patients with healthy lung than in patients with ARDS

Use EIT during a constant low-flow inflation to test the lung for recruitability

(24) n=26 patients (25) n=9 patients

Changes in regional Crs (10, 12, 14, 22, 26), the GI index (22, 27, 28) or CoV (22)

Assessment of changes in regional Crs or regional redistribution of ventilation (homogeneity of ventilation)

Incremental PEEP trial or lung recruitment maneuver and decremental PEEP trial

Regional Crs is lower in dependent lung regions in the pathological case

Set "optimal" PEEP at level which results in lowest GI index value in fixed ROIs or which minimizes the amount of lung collapse (<10%)

(26) n=2 patients (10) n=5 patients (14) n=12 patients (27) n=50 patients (28) n= 10 patients (22) n=20 patients (12) n=16 pigs

6

Clinical issue







Atelectasis / lung collapse / derecruit-ment

Changes in regional Crs in dorsal lung regions. Increase in the U/L ventilation ratio or ventral displacement of the center of ventilation.

Assessment of changes in Crs in response to a PEEP step expressed as a percent change in regional (pixel) Crs in relation to the best Crs reached during a PEEP titration trial.

Decremental PEEP trial. Step changes in PEEP, inspiratory pressure or tidal volume

Normal: more homogeneous distribution of ventilation between ventral and dorsal regions of the lung. Pathologic: decreased relative regional ventilation in dorsal lung regions

Helps to identify lung-protective ventilator settings resulting in minimum lung collapse and overdistension.

(26) n=2 patients (4) n=10 patients (29) n=14 patients (12) n=16 pigs (8) n=6 pigs

Distribution of tidal EIT variation between dependent and non-dependent regions (U/L ventilation ratio)

Increase in U/L ventilation ratio after suctioning implies derecruitment

Open suctioning Patients with a fall in functional residual capacity after suctioning exhibit an increase in U/L ventilation ratio

Stable U/L ratio after suctioning implies that patients do not require a recruitment maneuver

(30) n= 59 patients

CoV CoV shifts towards non-dependent lung regions after the induction of: - general anaesthesia during PSV, PCV and VCV - and pneumoperitoneum

Mechanical ventilation with laryngeal mask or endotracheal tube

Spontaneous breathing and PEEP preserve the preoperative ventilation distribution during general anaesthesia

n.a. (7) n=10 patients (31) n=30 patients (32) n=32 patients

Homo-geneity or hetero-geneity of regional respiratory system mechanics

Determination of inhomogeneous respiratory system mechanics, calculation of EIT-derived measures like the GI index, center of ventilation, U/L ventilation ratio.

The inhomogeneity of regional ventilation can be characterized by several gross EIT measures or by indices describing the ventilation distribution in the anteroposterior direction.

PEEP trial, ongoing tidal ventilation

Normal: homogeneous respiratory system mechanics

Inhomogeneity may lead to an adjustment of ventilator settings until homogeneity is reached. Severe un-correctable inhomogeneity may lead to rescue ECMO therapy.

(27) n=50 patients (28) n=10 patients (33) n=2 patients

7

Clinical issue







Spatial and temporal heterogeneity of ventilation during pulmonary function testing measured by EIT

Regional IVC, FEV1, FVC, FEV1/FVC and expiration times needed to exhale set percentages of FVC. The heterogeneity of their spatial distribution characterized by the CV or GI indices

Full inspiration to TLC and forced full expiration to RV

Pathological: higher spatial and temporal ventilation heterogeneity

n.a. (34) n=33 patients and n=26 healthy human subjects (35) n=7 patients and n= 7 healthy subjects (36) n=14 patients and n=14 healthy subjects

Assessment of homo-geneity/heterogeneity of ventilation distri-bution (PSV and NAVA: (37), HFOV: (38-41), CMV (42)) by various EIT measures

Regional distribution/changes in ventilation or lung volume in set regions of interest during various modes of mechanical ventilation.

Measurement during CMV, HFOV, PSV, NAVA

n.a. Set mechanical ventilation settings to ensure homogeneous distribution of ventilation/lung volume in set regions of interest

(37) n=10 patients (39) n=10 patients (40) n=1 patient (42) n=10 patients (38) n=9 pigs (41) n=18 pigs

Regional compliance / regional impedance-pressure curves

Estimation of regional ΔZ/ΔP (26) or fractional VT/ΔP (10, 12); construction of regional ΔZ-P curves (43-45)

Assumption of uniform pressure distribution in the ventilated lung.

Crs: step changes in PEEP under constant ventilator settings. Regional ΔZ-P curves or fractional VT: constant low-flow lung inflation (and deflation). Step increases/decrea-ses in airway pressure.

Normal: homogenous ΔZ/ΔP

Optimal PEEP selection. Detection of regional lung recruitment, overdistension and collapse.

(26) n=2 patients (44) n=9 patients (10) n=10 pigs (12, 43) n=16 pigs (45) n=9 pigs (46) n=12 pigs

8

Clinical issue







Effect of high frequency oscillatory ventilation

Homogenization of lung volume distribution during HFOV, assessment of regional lung volume distribution during HFOV

After calibration of tidal impedance to tidal air volume.

calibration Normal: homogeneous lung volume distribution

Change of CDP and oscillation frequency

(39) n=10 patients (38) n=8 pigs (lavage injury)

Effect of sighs

Distribution of EELZ and tidal Z variation

Sigh increases EELZ and decreases ventilation heterogeneity

Sigh: 35 cm H2O for 3-4 s applied at different rates

Normal: homogeneous lung volume distribution

Sighs reduce strain and ventilation heterogeneity during assisted ventilation in ARDS

(47) n=20 patients

Over-distension

Regional pressure-volume (ΔZ) curves

Determination of regional UIP by an automatic algorithm.

parallel measurement of airway pressure, better: transpulmonary pressure; low-flow inflation maneuver

Normal: homogenous UIP

Reduction of mean airway pressure

(44) n=9 patients

Deterioration of ventilation in lung units

Decrease in pixel Crs as surrogate for amount of alveolar collapse within that pixel. Weighted average at each PEEP step gives an estimate of cumulative collapse.

Lung recruitment maneuver and decremental PEEP trial

n.a. Set PEEP at level with lowest cumulated collapse

(26) n=2 patients

9

Clinical issue







Regional gas flow

Determination of regional pulmonary gas flow

First derivative of regional ΔZ(t) is proportional to regional flow. V(t) can be imaged after calibration of ΔZ(t)

Measurement during tidal ventilation or during forced full expiration/ recruitment maneuver

Normal: homogenous distribution of regional gas flow

None at the moment (36) n=14 patients and n=14 healthy adults (35) n=7 patients and n=7 healthy adults (48) n=5 patients (49) n=13 pigs

EIT-guided lung-protective ventilation

Use of EIT ventilation indices to reduce heterogeneity of ventilation distribution or optimize regional Crs (11, 50), EIT determined center of ventilation calculated in a Fuzzy controller (8)

Regional EIT-derived Crs used to maximize recruit-ment of the dependent and minimize overdistention of the nondependent lung areas. EIT proposed in automated (closed-loop) ventilation algorithms to determine the necessity of lung recruitment maneuvers

Incremental PEEP titration

Not applicable Use EIT-derived Crs to set PEEP at the level which balances dependent lung recruitment and nondependent overdistention.

(11) n=12 pigs (50) n=4 pigs (8) n=6 pigs

Single-lung ventilation

Monitoring qualitatively or quantitatively left and right-sided ventilation

During tidal artificial ventilation, a ROI-based analysis is performed either in a fEIT ventilation tomogram or in the EIT right and left lung waveforms. A loss or gain in ventilation during manipulation of the endotracheal tube shows either SLV or DLV.

Measurement during tidal artificial ventilation

SLV should be avoided during standard situations of ventilations (it can occur due to too deep intubation). But SLV can also be desired (during lung/thorax surgery, ventilation in a patient with a large bronchopleural fistula).

Correct tube position either to induce SLV or DLV whatever is intended

(51) n=10 patients (52) n=40 patients (53) n=5 pigs

10

Clinical issue







Monitoring of tidal recruitment of atelectasis

Monitoring: regional determination of tidal recruitment (54-56)

Proof of tidal recruitment is given by fast non-invasive measurement of SpO2 oscillations, changes in regional VT, EELZ and gas flow are followed by EIT in different ROIs

Parallel measurement of intratidal changes of gas exchange by fast, non-invasive SpO2 (54)

Normal: no regional recruitment of atelectasis

Lung recruitment, increase of PEEP, reduction of VT

(56)n=9 patients (54) n=21 pigs (55) n=9 pigs

Detecting a potential for tidal recruitment: Ventilation delay in the dorsal regions of the lung (17, 18)

During a low-flow inflation the delay until a threshold increase of regional Z in dorsal lung regions is reached (in comparison with global or ventral regions) indicates the late "opening" of lung units that collapsed during expiration.

Low-flow inflation maneuver

Normal: no regional recruitment of atelectasis

Lung recruitment, increase in PEEP, reduction of VT

(17) n=16 pigs (18) n=12 pigs

Detection of tidal recruitment by increased Crs in dependent regions on VT increase (10)

Higher Crs at higher VT at identical PEEP implies tidal recruitment

Temporary increase in VT

Regional Crs remains stable when tidal recruitment is noit present

Lung recruitment, increase in PEEP

(10) n=5 patients and 10 pigs

Table E7.1. Clinical issues of EIT use in adult patients focussing on monitoring lung ventilation and guidance of mechanical ventilation.

CDP: continuous distending pressure; CoV: center of ventilation; CMV: controlled mechanical ventilation; CoV: center of ventilation; Crs: respiratory system compliance; CV: coefficient of variation; DLV: double-lung ventilation; EELZ: end-expiratory lung impedance; ΔEELZ: end-expiratory lung impedance change; fEIT: functional EIT; FEV1: forced expiratory volume in one second; FVC: forced expiratory vital capacity; GI: global inhomogeneity, HFOV: high-frequency oscillatory ventilation; IVC: inspiratory vital capacity; n.a.: not applicable; LIP: lower inflection point of a pressure-volume curve; NAVA: neutrally adjusted ventilation assist; P: pressure; Pinsp: inspiratory pressure; ΔP: pressure change; PEEP: positive end-expiratory pressure, PCV: pressure-controlled ventilation; PSV: pressure-support ventilation; ROI: region of interest; RV: residual volume; SLV: single-lung ventilation; SpO2: oxygen saturation; TLC: total lung capacity; UIP: upper inflection point of a pressure-volume curve; U/L: upper-to-lower ventilation ratio; VCV: volume-controlled ventilation; VT: tidal volume; Z: impedance; ΔZ: impedance change

11

EIT use for monitoring cardiac activity, lung perfusion and fluid status

It has long been realized that the physiological activity of the heart leads to periodic changes in

regional thoracic electrical bioimpedance that can be traced by EIT (57, 58). This information

can be used to study for instance the heart performance, by measuring the stroke volume, or

lung perfusion. Compared with the ventilation-oriented chest EIT, the cardiovascular aspects of

EIT use have been studied less often and the number of involved research groups is lower. The

explanation may be that the analysis of EIT data specifically aiming at cardiovascular

phenomena in the chest is challenging because of the small amplitude of the corresponding

heart-beat related signal variations.

Table E7.2 summarizes the efforts in EIT research focussing on such clinical issues as

the assessment of stroke volume, ventricular diastolic performance, pulmonary perfusion,

pulmonary hypertension, ventilation-perfusion matching and heart-lung interactions. The table

also contains information on a few studies targeting the measurement of the fluid content of

the lung tissue using EIT. The clinical issues comprise the diagnosis and monitoring of lung

edema development and its therapy and the related assessment of extravascular lung water.

12

Clinical issue







Pulmonary perfusion and fluid state

Pulmonary perfusion

EIT lung pulsatility (59-61) or regional relative distribution of RPBF by hypertonic saline first-past kinetics (62, 63) or indirect assessment of RPBF by EIT measurement of regional O2 uptake (64)

Cardiac-related impedance changes can be detected by EIT by ECG-triggered measurement, measure-ment during apnea or frequency-domain filtering. The signal is not directly related to stroke volume but more to vessel pulsations. Pulsatility: detects cardiac-related impedance changes related to changes in local blood volume. RPBF: The high conductivity of hypertonic saline makes it suitable for an electric contrast agent. Injected in the right atrium a first-pass kinetics analysis can be performed.

Administration of saline bolus, ECG triggered EIT data sampling, frequency domain filtering. Pulsatility: Frequency domain filtering and/or ECG-gated EIT acquisition. During apnea or tidal ventilation, RPBF distribution: Administration of a small saline bolus into a central vein during apnea.

Normal: homogenous distribution of heart-beat related signal. No interference with inspiration or expiration. Pathology (pulmonary embolism): regional perfusion deficit

None at the moment (60) n=10 patients (61) n=10 healthy subjects and 20 patients (65) n=8 human subjects (ECG trigger) (64) n=12 patients (59) n=8 pigs (apnea) (62) n=3 pigs (saline bolus) (63) n=6 pigs (saline bolus) (53) n=5 pigs (saline bolus)

Pulmonary hyper-tension

EIT pulsatility amplitude: Maximal impedance change during systole (ΔZsys)

Changes in EIT pulsatility amplitude can be related to the interaction between cardiac-related changes in vascular volume and local vascular mechanics.

Measurement during spontaneous breathing or mechanical ventilation

Patients with IPAH have a reduced volume pulse of the pulmonary vascular bed manifested as a reduced systolic impedance amplitude

Diagnosis screening of IPAH. Evaluation of responses to treatment.

(66) n=21 patients and n=30 healthy human subjects (67) n=8 patients

13

Clinical issue







Ventilation-perfusion mismatch

EIT regional ventilation and pulsatility estimated perfusion

Regional mapping of ventilation and perfusion and ventilation-perfusion ratio

Analysis of functional ventilation and perfusion maps

Normal: matched ventilation and perfusion

None at the moment (68) n=12 healthy subjects (69) n=11 pigs (70) n=6 pigs

Cardiac output

Cyclic variation in Z in the cardiac (ventricular) region, HR derived from the EIT signal

Cyclic variation in Z correlates with SV

No specific interventions (variation in SV in validation studies)

n.a. None at the moment (57) n=4 healthy subjects (71) n=10 healthy subjects (72) n=10 healthy subjects (73) n=26 patients and n=10 healthy subjects (74) n=14 pigs

Right ventricular diastolic function

ROI-based analysis of the diastolic part of the heart-beat related EIT waveform

EIT-derived RAEV correlates with MRI findings

No specific interventions

(75) n=27 patients and n=7 healthy subjects

Heart-lung interaction

EIT assessment of pulmonary stroke volume variation

ΔZ pulse arrival time Mechanical ventilation at different VT and PEEP settings, hypovolemia

Ventilation induces stroke volume variation that is detected by EIT

n.a. (76) n=8 pigs

14

Clinical issue







Lung edema / extra-vascular lung water

Estimation of extravascular lung water by U/L ventilation ratio (77), reduction in ΔEELZ (78), mean Z (79, 80) or ΔZ during full inspiration (81), fdEIT (82) or aEIT imaging (83), EIT lung water ratio computed from the IM coefficient (84)

EVLW decreases bio-impedance. Increased EVLW can be detected by analysis of ventilation-induced impedance changes in the anterior and posterior parts of the lung (U/L ventilation ratio) or decreased ΔEELZ or reversed slope of IM coefficient during lateral positioning

Normal mechanical ventilation. Calculation of U/L ratio. Fluid administration. Right and left tilt and calculation of imbalance between right and left lung ventilation.

An U/L of 0.64 has a 93% sensitivity and 87% specificity to detect a supranormal EVLW (77). IM coefficients exhibit a negative trend slope in normal lungs during body rotation and a positive one in injured lungs.

Diagnosis of lung edema. Aid in clinical decision making regarding fluid/diuretic therapy. Diagnosis high-altitude pulmonary edema. (Changes in pulmonary fluid content may impact the EIT-derived information on lung aeration and should be taken into account.)

(77) n=14 patients (79) n=9 healthy subjects (80) n=20 patients (83) n=4 patients (81) n=20 healthy subjects (78) n=14 pigs (82) n=3 goats (83) n=5 pigs (84) n=30 pigs

Table E7.2. Clinical issues of EIT use in adult patients focussing on heart activity, lung perfusion and fluid status.

aEIT: absolute EIT imaging; EVLW: extravascular lung water; EELZ: end-expiratory lung impedance; ΔEELZ: end-expiratory lung impedance change; fdEIT: frequency-difference EIT imaging; HR: heart rate; IM: imbalance ventilation coefficient; IPAH: idiopathic pulmonary hypertension; n.a.: not applicable; PEEP: positive end-expiratory pressure, RAEV: right atrium emptying volume; RPBF: regional pulmonary blood flow; SV: stroke volumeU/L: upper-to-lower ventilation ratio; VT: tidal volume; Z: impedance; ΔZ: impedance change; ΔZsys: maximum impedance change during systole

15

EIT examinations in specific lung disease

The most widespread clinical use of EIT is noted in critically ill, mechanically ventilated patients.

The most frequent lung disease that is assessed and monitored by EIT in these patients is

ARDS. Nonetheless, there exist several other pulmonary diseases where EIT has been applied

and where the potential for routine clinical use has been identified. These comprise diseases

like pneumothorax and pleural effusion that can also be encountered in mechanically ventilated

patients but also other diseases where EIT examinations take place during spontaneous

breathing. The number of studies focussing on chronic lung diseases like COPD, asthma and

cystic fibrosis is increasing, the results imply that EIT might be of benefit, e.g. in disease

staging or monitoring of disease progression and therapy effects. EIT has also been applied in

patients suffering from pneumonia or lung cancer. A few case reports on EIT examinations in

less common diseases, like tracheobronchomalacia and lung transplant, have been reported as

well. Table E7.3 gives the detailed information on EIT use in these specific lung diseases.

16

Clinical issue







Specific diagnoses

Pleural effusion/ empyema

Ventilation distribution, ΔEELZ, ventilation asynchrony

R/L lung ventilation asymmetry in unilateral affections, changes in ventilation distribution


Ventilation reduced at the affected side of the chest in predominantly unilateral pathology

The improved symmetry of right-to-left lung ventilation distribution can be assessed during aspiration of pleural fluid

(85) n=6 patients (86) n= 10 patients (87) n=22 patients

Pneumo-thorax

Ventilation distribution, ΔEELZ automated algorithm for detection of penumothirax

Online sequential analysis of perturbations seen in the EIT ventilation and aeration change maps constructed from two to three respiratory cycles.

Sequential increase in PEEP with a fixed inspiratory driving pressure (88)

Pneumothoraces as small as 20 mL could be detected with EIT (sensitivity 100%; specificity 95%)

EIT-based algorithm can be used to detect early signs and locations of pneumothoraces in high-risk situations

(88) n=39 pigs (89) n=1 pig

Pneumonia Ventilation distribution R/L lung ventilation asymmetry in unilateral disease


Ventilation reduced at the affected side of the chest

None at the moment (90) n=1 patient (91) n=24 patients

Lung cancer

Ventilation distribution R/L lung ventilation asymmetry in unilateral bronchial carcinoma


Ventilation reduced at the affected side of the chest

None at the moment (90) n=1 patient (92) n=5 patients (93) n=21 patients (94) n=14 patients

17

Clinical issue







COPD Spatial and temporal ventilation homogeneity (quantified by various EIT measures representing regional lung volumes (e.g. VT, FEV1, FVC), flow rates (e.g. PEF, FEF25-75) or expiration times)

Ventilation distribution is heterogeneous in COPD, regional responses to bronchodilator administra-tion can be traced by EIT

Measurement during spontaneous breathing (34, 95-97), forced full expiration (34, 97) and mechanical ventilation (39, 98)

Ventilation heterogeneity more pronounced in COPD patients compared with healthy subjects

Regional pulmonary function monitoring, identification of exacerbation and response to therapy and reversibility testing, phenotyping

(34) n=33 patients and n=26 healthy subjects (95) n=12 patients and n=15 healthy subjects (96) n=28 patients (39) n=10 patients (97) n= 35 patients (98) n=1 patient

Asthma Spatial and temporal ventilation homogeneity (quantified by various EIT measures representing regional lung volumes (e.g. VT, FEV1, FVC), expiratory flows (PEF, FEF25-75) or expiration times)

Ventilation distribution is heterogeneous in patients with chronic history of asthma, it improves after bronchodilator reversibility testing

Measurement during spontaneous breathing and forced full expiration

Ventilation heterogeneity is more pronounced in asthma patients compared with age-matched healthy subjects

Regional pulmonary function monitoring, identification of exacerbation and response to therapy and reversibility testing

(35) n=7 patients and n=7 healthy subjects

18

Clinical issue







Cystic fibrosis

Spatial distribution of EIT-derived maximum expiratory flows ratio MEF25/MEF75, GI indices of inhaled volumes at 25%, 50%, 75% and 100% of IVC (36) Fraction of ventilation and filling index in the right lung region (99)

Pixel MEF25/MEF75 are heterogeneously distributed, GI indices of inhaled volumes fall significantly in the course of full inspiration in patients with CF

Forced full inspiration and expiration (36), lateral posture and breathing aids (99)

Normal: Narrow distribution peaks of MEF25/MEF75, stable GI during deep inspiration Pathological: Broad MEF25/MEF75 distribu-tion, high GI indices at low inhaled volumesTemporal ventilation distribution is affected by breathing aids in healthy subjects but not in patients with CF

Regional pulmonary function monitoring, identification of exacerbation and response to drug and physical therapy

(36) n=14 patients and n=14 healthy subjects (99) n=9 patients and n=11 healthy subjects

Tracheo-broncho-malacia

Percentage of collapse and overdistension during PEEP titration (100), R/L ventilation distribution (101)

EIT identified the PEEP level where airway collapse was be prevented

Decremental PEEP trial

n.a. Adjustment of ventilator settings

(100) n=1 patient (101) n=1 patient

Lung transplant

Ventilation distribution

Dissimilar functional status of the native and the transplanted lungs

Spontaneous breathing or mechanical ventilation

Symmetric ventilation distribution in healthy subjects

Adjustment of ventilator settings depending on the R/L ventilation distribution

(102) n=1 patient

Table E7.3. EIT use in adult patients with specific pulmonary diagnoses.

CF: cystic fibrosis; ΔEELZ: change in end-expiratory lung impedance; FEF25-75: mean forced expiratory flow between 25% and 75% of FVC; FEV1: forced expiratory volume in one second; FVC: forced expiratory vital capacity; GI: global inhomogeneity; IVC: inspiratory vital capacity; MEF25: maximum expiratory flow at 25% of FVC; MEF75: maximum expiratory flow at 75% of FVC; PEEP: positive end-expiratory pressure, PEF: peak expiratory flow; R/L: right-to-left; VT: tidal volume

19

Recommendations for future direction

As documented by the findings of a large number of EIT studies summarized in the Tables

E7.1-E7.3, EIT can generate information on regional lung ventilation and lung function that

is in part unique or not easily accessible by other medical technology and from which the

examined patients might benefit. This multitude of EIT results, however, does not yet allow

the conclusion that EIT is already fully ready for the routine clinical use.

The current degree of knowledge implies that the routine use of EIT is most probable

in the personalized bedside guidance of ventilator therapy. Prospective clinical trials are

urgently needed to confirm the ability of EIT to reliably guide mechanical ventilation in

comparison with the current standard clinical practice. Based on these studies, consensus

algorithms for adjustment of ventilator settings should be developed based on selected EIT

measures in combination with the findings of other medical diagnostic and monitoring

methods. An automated generation of relevant EIT findings with easy accessibility for

immediate use by the clinical staff with alerts in the case of potentially adverse situations is

essential. Large multicenter outcome EIT studies should then follow. An implementation of

EIT monitoring into the standard ICU monitoring would be desirable.

The use of chest EIT in cardiovascular monitoring and regional pulmonary function

testing is less advanced. It still requires validation studies, for instance regarding the aspired

long-term monitoring of patients with chronic lung diseases where the proof of the

correlation between the disease severity and EIT findings is lacking. Methodological

advances are also needed, for instance allowing reliable EIT data acquisitions and analysis

under the more challenging conditions of spontaneous breathing in typically upright and

partially moving patients. The subsequent steps necessary for establishing EIT in these

additional clinical fields (i.e. the prospective mono- and multicenter studies mentioned

above) resemble the proposed steps needed in establishing EIT for optimization of ventilator

therapy.

The clinical use and acceptance of EIT would certainly benefit from further

technological advances. Wearable EIT would largely increase its applicability and potentially

allow home monitoring especially in pulmonology patients. EIT image reconstruction taking

individual a priori anatomical information (which often is available in patients) into account

might improve not only the EIT image quality but also increase the information content of

EIT chest examinations. The research in these areas is ongoing and it could foster the

clinical use of EIT even in not yet identified new application fields.

20


The first draft of this online document was prepared by M. Bodenstein with collaboration of

L. Camporota, I. Frerichs, F. Suarez-Sipman, E. Fan and N. Weiler. It was reviewed and

approved by all other authors and collaborators.

21

References




2. Lowhagen K, Lindgren S, Odenstedt H, Stenqvist O, Lundin S. A new non-

radiological method to assess potential lung recruitability: A pilot study in ALI patients. Acta


3. Kunst PW, de Vries PM, Postmus PE, Bakker J. Evaluation of electrical impedance

tomography in the measurement of PEEP-induced changes in lung volume. Chest

1999;115:1102-1106.

4. Mauri T, Bellani G, Confalonieri A, Tagliabue P, Turella M, Coppadoro A, Citerio G,

Patroniti N, Pesenti A. Topographic distribution of tidal ventilation in acute respiratory

distress syndrome: Effects of positive end-expiratory pressure and pressure support. Crit

Care Med 2013;41:1664-1673.

5. Cinnella G, Grasso S, Raimondo P, D'Antini D, Mirabella L, Rauseo M, Dambrosio M.

Physiological effects of the open lung approach in patients with early, mild, diffuse acute

respiratory distress syndrome: An electrical impedance tomography study. Anesthesiology

2015;123:1113-1121.

6. Blankman P, SM VDK, Gommers D. Tidal ventilation distribution during pressure-

controlled ventilation and pressure support ventilation in post-cardiac surgery patients. Acta





8. Luepschen H, Meier T, Grossherr M, Leibecke T, Karsten J, Leonhardt S. Protective

ventilation using electrical impedance tomography. Physiol Meas 2007;28:S247-260.







11. Wolf GK, Gomez-Laberge C, Rettig JS, Vargas SO, Smallwood CD, Prabhu SP, Vitali

SH, Zurakowski D, Arnold JH. Mechanical ventilation guided by electrical impedance

tomography in experimental acute lung injury. Crit Care Med 2013;41:1296-1304.

22



1961.




14. Blankman P, Hasan D, Groot Jebbink E, Gommers D. Detection of 'best' positive end-

expiratory pressure derived from electrical impedance tomography parameters during a

decremental positive end-expiratory pressure trial. Crit Care 2014;18:R95.



Crit Care 2012;27:323 e311-328.

16. Zhao Z, Frerichs I, Pulletz S, Muller-Lisse U, Moller K. The influence of image

reconstruction algorithms on linear thorax EIT image analysis of ventilation. Physiol Meas

2014;35:1083-1093.

17. Wrigge H, Zinserling J, Muders T, Varelmann D, Gunther U, von der Groeben C,

Magnusson A, Hedenstierna G, Putensen C. Electrical impedance tomography compared

with thoracic computed tomography during a slow inflation maneuver in experimental models

of lung injury. Crit Care Med 2008;36:903-909.





19. Hinz J, Hahn G, Neumann P, Sydow M, Mohrenweiser P, Hellige G, Burchardi H.

End-expiratory lung impedance change enables bedside monitoring of end-expiratory lung

volume change. Intensive Care Med 2003;29:37-43.



bedside during a decremental positive end-expiratory lung pressure trial. Crit Care

2011;15:R193.

21. Erlandsson K, Odenstedt H, Lundin S, Stenqvist O. Positive end-expiratory pressure

optimization using electric impedance tomography in morbidly obese patients during

laparoscopic gastric bypass surgery. Acta Anaesthesiol Scand 2006;50:833-839.

22. Karsten J, Grusnick C, Paarmann H, Heringlake M, Heinze H. Positive end-expiratory

pressure titration at bedside using electrical impedance tomography in post-operative cardiac

surgery patients. Acta Anaesthesiol Scand 2015;59:723-732.

23

23. Frerichs I, Dargaville PA, Dudykevych T, Rimensberger PC. Electrical impedance

tomography: A method for monitoring regional lung aeration and tidal volume distribution?

Intensive Care Med 2003;29:2312-2316.

24. Zhao Z, Pulletz S, Frerichs I, Muller-Lisse U, Moller K. The EIT-based global

inhomogeneity index is highly correlated with regional lung opening in patients with acute

respiratory distress syndrome. BMC Res Notes 2014;7:82.

25. Becher T, Kott M, Schadler D, Vogt B, Meinel T, Weiler N, Frerichs I. Influence of tidal

volume on ventilation inhomogeneity assessed by electrical impedance tomography during

controlled mechanical ventilation. Physiol Meas 2015;36:1137-1146.







28. Zhao Z, Steinmann D, Frerichs I, Guttmann J, Moller K. PEEP titration guided by

ventilation homogeneity: A feasibility study using electrical impedance tomography. Crit Care

2010;14:R8.

29. Schaefer MS, Wania V, Bastin B, Schmalz U, Kienbaum P, Beiderlinden M, Treschan

TA. Electrical impedance tomography during major open upper abdominal surgery: A pilot-

study. BMC Anesthesiol 2014;14:51.

30. Heinze H, Eichler W, Karsten J, Sedemund-Adib B, Heringlake M, Meier T.

Functional residual capacity-guided alveolar recruitment strategy after endotracheal

suctioning in cardiac surgery patients. Crit Care Med 2011;39:1042-1049.

31. Radke OC, Schneider T, Heller AR, Koch T. Spontaneous breathing during general

anesthesia prevents the ventral redistribution of ventilation as detected by electrical

impedance tomography: A randomized trial. Anesthesiology 2012;116:1227-1234.

32. Karsten J, Luepschen H, Grossherr M, Bruch HP, Leonhardt S, Gehring H, Meier T.

Effect of PEEP on regional ventilation during laparoscopic surgery monitored by electrical


33. Camporota L, Smith J, Barrett N, Beale R. Assessment of regional lung mechanics

with electrical impedance tomography can determine the requirement for ECMO in patients

with severe ARDS. Intensive Care Med 2012;38:2086-2087.




24



patients with asthma. Physiol Meas 2016; 37: 698-712.



37. Blankman P, Hasan D, van Mourik MS, Gommers D. Ventilation distribution

measured with EIT at varying levels of pressure support and neurally adjusted ventilatory

assist in patients with ALI. Intensive Care Med 2013;39:1057-1062.

38. van Genderingen HR, van Vught AJ, Jansen JR. Regional lung volume during high-

frequency oscillatory ventilation by electrical impedance tomography. Crit Care Med

2004;32:787-794.







41. Wolf GK, Grychtol B, Boyd TK, Zurakowski D, Arnold JH. Regional overdistension

identified with electrical impedance tomography in the perflubron-treated lung. Physiol Meas

2010;31:S85-95.

42. Victorino JA, Borges JB, Okamoto VN, Matos GF, Tucci MR, Caramez MP, Tanaka

H, Sipmann FS, Santos DC, Barbas CS, Carvalho CR, Amato MB. Imbalances in regional

lung ventilation: A validation study on electrical impedance tomography. Am J Respir Crit

Care Med 2004;169:791-800.



Meas 2013;34:567-577.

44. Hinz J, Moerer O, Neumann P, Dudykevych T, Frerichs I, Hellige G, Quintel M.

Regional pulmonary pressure volume curves in mechanically ventilated patients with acute

respiratory failure measured by electrical impedance tomography. Acta Anaesthesiol Scand

2006;50:331-339.

45. Kunst PW, Bohm SH, Vazquez de Anda G, Amato MB, Lachmann B, Postmus PE, de

Vries PM. Regional pressure volume curves by electrical impedance tomography in a model

of acute lung injury. Crit Care Med 2000;28:178-183.

46. Grychtol B, Wolf GK, Arnold JH. Differences in regional pulmonary pressure-

impedance curves before and after lung injury assessed with a novel algorithm. Physiol

Meas 2009;30:S137-148.

25

47. Mauri T, Eronia N, Abbruzzese C, Marcolin R, Coppadoro A, Spadaro S, Patroniti N,

Bellani G, Pesenti A. Effects of sigh on regional lung strain and ventilation heterogeneity in

acute respiratory failure patients undergoing assisted mechanical ventilation. Crit Care Med

2015;43:1823-1831.

48. Zhao Z, Muller-Lisse U, Frerichs I, Fischer R, Moller K. Regional airway obstruction in

cystic fibrosis determined by electrical impedance tomography in comparison with high


49. Bodenstein M, Boehme S, Bierschock S, Vogt A, David M, Markstaller K.

Determination of respiratory gas flow by electrical impedance tomography in an animal

model of mechanical ventilation. BMC Pulm Med 2014;14:73.

50. Gomez-Laberge C, Rettig JS, Smallwood CD, Boyd TK, Arnold JH, Wolf GK.

Interaction of dependent and non-dependent regions of the acutely injured lung during a

stepwise recruitment manoeuvre. Physiol Meas 2013;34:163-177.







53. Reinius H, Borges JB, Freden F, Jideus L, Camargo ED, Amato MB, Hedenstierna G,

Larsson A, Lennmyr F. Real-time ventilation and perfusion distributions by electrical

impedance tomography during one-lung ventilation with capnothorax. Acta Anaesthesiol

Scand 2015;59:354-368.

54. Bodenstein M, Boehme S, Wang H, Duenges B, Markstaller K. Hints for cyclical

recruitment of atelectasis during ongoing mechanical ventilation in lavage and oleic acid lung

injury detected by SpO2 oscillations and electrical impedance tomography. Exp Lung Res

2014;40:427-438.






2012;31:834-842.

57. Eyuboglu BM, Brown BH, Barber DC, Seagar AD. Localisation of cardiac related

impedance changes in the thorax. Clin Phys Physiol Meas 1987;8 Suppl A:167-173.

26

58. McArdle FJ, Suggett AJ, Brown BH, Barber DC. An assessment of dynamic images

by applied potential tomography for monitoring pulmonary perfusion. Clin Phys Physiol Meas

1988;9 Suppl A:87-91.



2009;53:152-158.



Respiration 2009;77:282-291.

61. Smit HJ, Vonk Noordegraaf A, Marcus JT, Boonstra A, de Vries PM, Postmus PE.

Determinants of pulmonary perfusion measured by electrical impedance tomography. Eur J

Appl Physiol 2004;92:45-49.

62. Frerichs I, Hinz J, Herrmann P, Weisser G, Hahn G, Quintel M, Hellige G. Regional

lung perfusion as determined by electrical impedance tomography in comparison with

electron beam CT imaging. IEEE Trans Med Imaging 2002;21:646-652.

63. Borges JB, Suarez-Sipmann F, Bohm SH, Tusman G, Melo A, Maripuu E, Sandstrom

M, Park M, Costa EL, Hedenstierna G, Amato M. Regional lung perfusion estimated by

electrical impedance tomography in a piglet model of lung collapse. J Appl Physiol

2012;112:225-236.

64. Elke G, Pulletz S, Schadler D, Zick G, Gawelczyk B, Frerichs I, Weiler N.

Measurement of regional pulmonary oxygen uptake - a novel approach using electrical

impedance tomography. Physiol Meas 2011;32:877-886.

65. Zadehkoochak M, Blott BH, Hames TK, George RF. Pulmonary perfusion and

ventricular ejection imaging by frequency domain filtering of EIT (electrical impedance

tomography) images. Clin Phys Physiol Meas 1992;13 Suppl A:191-196.

66. Smit HJ, Vonk-Noordegraaf A, Boonstra A, de Vries PM, Postmus PE. Assessment of

the pulmonary volume pulse in idiopathic pulmonary arterial hypertension by means of

electrical impedance tomography. Respiration 2006;73:597-602.

67. Smit HJ, Vonk Noordegraaf A, Roeleveld RJ, Bronzwaer JG, Postmus PE, de Vries

PM, Boonstra A. Epoprostenol-induced pulmonary vasodilatation in patients with pulmonary

hypertension measured by electrical impedance tomography. Physiol Meas 2002;23:237-

243.

68. Li Y, Tesselaar E, Borges JB, Bohm SH, Sjoberg F, Janerot-Sjoberg B. Hyperoxia

affects the regional pulmonary ventilation/perfusion ratio: An electrical impedance

tomography study. Acta Anaesthesiol Scand 2014;58:716-725.

27

69. Fagerberg A, Sondergaard S, Karason S, Aneman A. Electrical impedence

tomography and heterogeneity of pulmonary perfusion and ventilation in porcine acute lung

injury. Acta Anaesthesiol Scand 2009;53:1300-1309.

70. Fagerberg A, Stenqvist O, Aneman A. Electrical impedance tomography applied to

assess matching of pulmonary ventilation and perfusion in a porcine experimental model. Crit

Care 2009;13:R34.

71. Eyuboglu BM, Brown BH, Barber DC. In vivo imaging of cardiac related impedance

changes. IEEE Eng Med Biol Mag 1989;8:39-45.

72. Vonk Noordegraaf A, Faes TJ, Marcus JT, Janse A, Heethaar RM, Postmus PE, de

Vries PM. Improvement of cardiac imaging in electrical impedance tomography by means of

a new electrode configuration. Physiol Meas 1996;17:179-188.

73. Vonk-Noordegraaf A, 2nd, Janse A, Marcus JT, Bronzwaer JG, Postmust PE, Faes

TJ, De Vries PM. Determination of stroke volume by means of electrical impedance

tomography. Physiol Meas 2000;21:285-293.

74. Pikkemaat R, Lundin S, Stenqvist O, Hilgers RD, Leonhardt S. Recent advances in

and limitations of cardiac output monitoring by means of electrical impedance tomography.

Anesth Analg 2014;119:76-83.

75. Vonk Noordegraaf A, Faes TJ, Janse A, Marcus JT, Bronzwaer JG, Postmus PE, de

Vries PM. Noninvasive assessment of right ventricular diastolic function by electrical

impedance tomography. Chest 1997;111:1222-1228.

76. Maisch S, Bohm SH, Sola J, Goepfert MS, Kubitz JC, Richter HP, Ridder J, Goetz

AE, Reuter DA. Heart-lung interactions measured by electrical impedance tomography. Crit

Care Med 2011;39:2173-2176.

77. Kunst PW, Vonk Noordegraaf A, Raaijmakers E, Bakker J, Groeneveld AB, Postmus

PE, de Vries PM. Electrical impedance tomography in the assessment of extravascular lung

water in noncardiogenic acute respiratory failure. Chest 1999;116:1695-1702.

78. Bodenstein M, Wang H, Boehme S, Vogt A, Kwiecien R, David M, Markstaller K.

Influence of crystalloid and colloid fluid infusion and blood withdrawal on pulmonary

bioimpedance in an animal model of mechanical ventilation. Physiol Meas 2012;33:1225-

1236.

79. Noble TJ, Harris ND, Morice AH, Milnes P, Brown BH. Diuretic induced change in

lung water assessed by electrical impedance tomography. Physiol Meas 2000;21:155-163.

80. Noble TJ, Morice AH, Channer KS, Milnes P, Harris ND, Brown BH. Monitoring

patients with left ventricular failure by electrical impedance tomography. Eur J Heart Fail

1999;1:379-384.

81. Mason NP, Petersen M, Melot C, Imanow B, Matveykine O, Gautier MT, Sarybaev A,

Aldashev A, Mirrakhimov MM, Brown BH, Leathard AD, Naeije R. Serial changes in nasal

28

potential difference and lung electrical impedance tomography at high altitude. J Appl Physiol

2003;94:2043-2050.

82. Brown BH, Flewelling R, Griffiths H, Harris ND, Leathard AD, Lu L, Morice AH,

Neufeld GR, Nopp P, Wang W. EITS changes following oleic acid induced lung water.

Physiol Meas 1996;17 Suppl 4A:A117-130.

83. Hahn G, Just A, Dudykevych T, Frerichs I, Hinz J, Quintel M, Hellige G. Imaging

pathologic pulmonary air and fluid accumulation by functional and absolute EIT. Physiol

Meas 2006;27:S187-198.

84. Trepte CJ, Phillips CR, Sola J, Adler A, Haas SA, Rapin M, Bohm SH, Reuter DA.

Electrical impedance tomography (EIT) for quantification of pulmonary edema in acute lung

injury. Crit Care 2016;20:18.

85. Blaser D, Pulletz S, Becher T, Schadler D, Elke G, Weiler N, Frerichs I. Unilateral

empyema impacts the assessment of regional lung ventilation by electrical impedance

tomography. Physiol Meas 2014;35:975-983.

86. Campbell JH, Harris ND, Zhang F, Brown BH, Morice AH. Clinical applications of

electrical impedance tomography in the monitoring of changes in intrathoracic fluid volumes.

Physiol Meas 1994;15 Suppl 2a:A217-222.

87. Alves SH, Amato MB, Terra RM, Vargas FS, Caruso P. Lung reaeration and

reventilation after aspiration of pleural effusions. A study using electrical impedance

tomography. Ann Am Thorac Soc 2014;11:186-191.

88. Costa EL, Chaves CN, Gomes S, Beraldo MA, Volpe MS, Tucci MR, Schettino IA,

Bohm SH, Carvalho CR, Tanaka H, Lima RG, Amato MB. Real-time detection of

pneumothorax using electrical impedance tomography. Crit Care Med 2008;36:1230-1238.

89. Preis C, Luepschen H, Leonhardt S, Gommers D. Experimental case report:

Development of a pneumothorax monitored by electrical impedance tomography. Clin

Physiol Funct Imaging 2009;29:159-162.

90. Frerichs I, Golisch W, Hahn G, Kurpitz M, Burchardi H, Hellige G. Heterogeneous

distribution of pulmonary ventilation in intensive care patients detected by functional electrical

impedance tomography. J Intensive Care Med 1998;13:168-173.

91. Karsten J, Krabbe K, Heinze H, Dalhoff K, Meier T, Dromann D. Bedside monitoring

of ventilation distribution and alveolar inflammation in community-acquired pneumonia. J Clin

Monit Comput 2014;28:403-408.

92. Serrano RE, de LB, Casas O, Feixas T, Calaf N, Camacho V, Carrio I, Casan P,

Sanchis J, Riu PJ. Use of electrical impedance tomography (EIT) for the assessment of

unilateral pulmonary function. Physiol Meas 2002;23:211-220.

29

93. Serrano RE, Riu PJ, de Lema B, Casan P. Assessment of the unilateral pulmonary

function by means of electrical impedance tomography using a reduced electrode set.

Physiol Meas 2004;25:803-813.

94. Kunst PW, Vonk Noordegraaf A, Hoekstra OS, Postmus PE, de Vries PM. Ventilation

and perfusion imaging by electrical impedance tomography: A comparison with radionuclide

scanning. Physiol Meas 1998;19:481-490.

95. Eyuboglu BM, Oner AF, Baysal U, Biber C, Keyf AI, Yilmaz U, Erdogan Y. Application

of electrical impedance tomography in diagnosis of emphysema--a clinical study. Physiol

Meas 1995;16:A191-211.

96. Balleza M, Calaf N, Feixas T, Gonzalez M, Anton D, Riu PJ, Casan P. Measuring

breathing pattern in patients with chronic obstructive pulmonary disease by electrical

impedance tomography. Arch Bronconeumol 2009;45:320-324.




98. Mauri T, Bellani G, Salerno D, Mantegazza F, Pesenti A. Regional distribution of air

trapping in chronic obstructive pulmonary disease. Am J Respir Crit Care Med

2013;188:1466-1467.

99. Wettstein M, Radlinger L, Riedel T. Effect of different breathing aids on ventilation

distribution in adults with cystic fibrosis. PLoS One 2014;9:e106591.

100. Dias OM, Costa EL, Pereira DA, Chaves CN, Rached SZ, Barbas CS.

Tracheobronchomalacia in a patient on invasive mechanical ventilation: The role of electrical

impedance tomography in its detection and positive end-expiratory pressure titration. J Bras

Pneumol 2015;41:203-205.

101. Dostal P, Truhlar A, Polak J, Sedlak V, Turek Z. Assessment of regional ventilation

with the electrical impedance tomography in a patient after asphyxial cardiac arrest.

Resuscitation 2014;85:e115-117.

102. Guerin C, Frerichs I. Getting a better picture of the correlation between lung function

and structure using electrical impedance tomography. Am J Respir Crit Care Med

2014;190:1186-1187.





Bartłomiej Grychtol, Marc Bodenstein, Hervé Gagnon, Stephan H. Böhm, Eckhard Teschner, Ola

Stenqvist, Tommaso Mauri, Vinicius Torsani, Luigi Camporota, Andreas Schibler, Gerhard K.


ONLINE SUPPLEMENT 8

Clinical use of EIT in neonatal and pediatric patients

2

Clinical use of EIT in neonatal and pediatric patients

Introduction

EIT offers unique potential as a clinical tool in the pediatric and neonatal population, both to

monitor physiological states and to direct therapy and interventions. Currently there is need for

improved technologies to guide respiratory care in children. In general, bedside

cardiorespiratory monitoring is more difficult in pediatric patients (1). In addition, traditional

modes of cardiorespiratory imaging require patient cooperation, expose the patient to radiation

or are unacceptably invasive for children. Uncuffed endotracheal tubes (ETT) and the trend

towards exclusive or early non-invasive ventilation (NIV) modes, further complicates respiratory

monitoring in infants and children (2, 3). EIT, by virtue of being non-invasive, radiation-free,

portable, independent of an ETT and allowing multiple physiological information to be extracted

from a single recording, fulfils the criteria of an ideal pediatric lung monitor for the modern era.

The role of respiratory monitoring in improving outcomes in pediatric and neonatal lung

disease is well established (4-7). Increasing awareness of the differences in the regional

behaviour of the diseased pediatric lung, particularly in diseases characterized by atelectasis

(such as infant respiratory distress syndrome (RDS)), and a lack of existing lung monitoring

tools, has been the attention of previous pediatric studies of EIT in the clinical setting as well as

animal models. To date, most studies have focused on monitoring (regional) changes in lung

aeration and ventilation and, to a lesser extent, lung perfusion, with the aim of improving basic

knowledge on lung physiology and better understanding the interaction between the impact of

lung disease and clinical interventions on lung function. As of July 2015, we identified 48

articles reporting on EIT examinations of preterm and term neonates, infants and children with

the total of 1018 subjects studied. Our search was primarily based on the PubMed database

(National Center for Biotechnology Information, U.S. National Library of Medicine 8600 Rockville

Pike, Bethesda MD, 20894 USA).

This electronic online supplement (EOS) reviews the existing literature involving EIT in the

pediatric and neonatal populations, hindrances to clinical use and suggests areas for further

development.

3

Understanding the physiological principles of the respiratory system in the healthy

and diseased state

Regional patterns of tidal ventilation, end-expiratory lung volumes (EELV) and mechanics in the

newborn, infant and pediatric lungs are poorly understood. Safe and effective respiratory care

involves understanding the physiological state of the respiratory system. This is particularly

relevant in the diseased lung, where physiological states cannot be assumed to be uniform

throughout all lung units. The regional and global physiology of the respiratory system during

pediatric disease states has been well described in animal models (8-12) but the lack of

practical bedside tools to measure regional physiology has hampered confirmation of these

principles in human subjects. EIT has been extensively used to fill this research gap, and the

literature can be generally considered to demonstrate the ability of EIT to describe gravity-

dependent and right-to-left differences in volume states and mechanics. A few preliminary

studies investigating ventilation-perfusion relationships with EIT offer a new promising research

pathway and may lead to better understanding of the complex interaction of ventilation and

lung perfusion (13-16).

Gravity-dependent inhomogeneity

a. End-expiratory lung volume

The susceptibility of the acutely diseased adult lung to the gravity-dependent inhomogeneity of

EELV is well known (17). Radiological methods to describe gravity-dependency and volume

states are not practical in pediatric patients. The group of van Kaam and co-workers used EIT

to describe end-expiratory patterns in RDS of prematurity (18). This group further described the

interaction between applied pressure and regional volumes in 15 infants subjected to a stepwise

oxygenation-guided recruitment procedure during high-frequency oscillatory ventilation (HFOV).

During each incremental and decremental step in mean airway pressure the changes in EELV

were recorded and used to reconstruct the inflation and deflation limbs of the individual

pressure-volume relationship of the lung. The study confirmed that regional lung hysteresis is

present in preterm infants with RDS (18). It also showed that the regional (dependent versus

non-dependent) changes in EELV were highly variable and did not always follow the well-

defined gravity-dependent pattern seen in adult RDS. The hysteresis pattern identified was

similar to those described in the whole lung using respiratory inductive plethysmography (19).

This similarity was directly confirmed later when it was shown that EELV changes measured by

EIT in a c

measured

changes ov

time const

surfactant

Figure E8

overdistens

a nonrespo

The lung a

clinical rec

responders

exchange i

ross-section

by respirat

ver time, th

tant of the

therapy (21

8.1. EIT im

sion in blue

onder. The

area is divid

ruitment (re

s or reachin

in nonrespo

nal slice of

tory inducta

his group de

e lung in p

1).

mages of a

e. Left colum

correspond

ded into fo

eached sum

ng the maxi

onders. (Dat

the lung a

ance plethys

escribed, fo

preterm inf

pediatric l

mn shows a

ding platea

ur regions

m of arterial

mal plateau

ta originate

are represen

smography

r the first ti

fants with

lung recrui

a responder

u pressure

of interest.

partial pres

u pressure o

e from the s

ntative of t

(20). In a

ime, the gra

RDS; both

tment man

r to lung re

is shown t

“Pmax” on

ssures of O

of 50 cm H2

study (22).)

he changes

separate a

avity-depen

before an

neuver, atel

cruitment, r

to the right

n the left o

2 and CO2 o

2O without i

s in the wh

nalysis of t

ndent patter

nd after ex

lectasis in

right colum

t of each E

of the image

of > 400 mm

improvemen

4

hole lung

the EELV

rn of the

ogenous

red and

mn shows

EIT scan.

e means

m Hg) in

nt in gas

5

EIT has been used to define regional EELV states during a stepwise PEEP recruitment

maneuver in pediatric RDS (22, 23). In these studies EIT was able to classify lung areas as

atelectatic, overdistended or normally ventilated. Atelectasis was found to be predominant in

the posterior (dependent) lung areas. It was shown on EIT images that all posterior atelectatic

lung areas need to be completely recruited in order to achieve physiologic lung opening, as

defined by a significantly improved gas exchange. The pressures used to recruit posterior lung

areas always produced significant overdistension of anterior (nondependent) lung areas. Most

of the patients responded to a stepwise increase in plateau pressures (defined as

“responders”), and those responders had large amounts of posterior atelectasis prior to lung

recruitment. A few patients in these studies did not respond to high plateau pressures, as

evidenced by no improvement in gas exchange, and those patients had almost no atelectasis on

EIT. Overall, large amounts of atelectasis on EIT images significantly increased the chance of

responding to lung recruitment (Figure E8.1) (22).

In summary EIT has the potential to be a powerful monitoring tool for detecting changes

in EELV either as a result of an intervention or as a result of progression/improvement of lung

disease.

b. Tidal volume

EIT has been used to deepen our understanding of the distribution of tidal ventilation in infants.

Longstanding principles suggested a reversal of the adult pattern of tidal ventilation distribution

in infancy (24). EIT studies in spontaneously breathing healthy infants have repeatedly shown

this to be an oversimplification. The distribution of tidal ventilation is highly variable, almost on

a breath-to-breath basis and influenced by magnitude of tidal effort, respiratory rate and

pattern, EELV, body position and even neck position (25-27). Gravity was found to be a

significant factor in the ventilation distribution in 32 spontaneously breathing healthy infants

examined by EIT in the neonatal period, and at 3 and 6 months of life (28). Riedel and co-

workers demonstrated similar findings in a population of 17 healthy term infants, also

identifying that gravity had a greater effect on regional tidal ventilation patterns in

spontaneously breathing preterm infants, with greater relative tidal ventilation in the non-

dependent lung (29). This study compared EIT with the multiple breath washout technique for

measuring regional ventilation, finding higher repeatability and better discrimination of regional

patterns using EIT. Interestingly, a study by van den Burg and co-workers (30) in preterm

infants on nasal continuous positive airway pressure (CPAP) showed that changing from supine

to prone positioning increased tidal ventilation in the ventral, dependent lung regions. Lupton-

6

Smith and co-workers identified highly variable gravity-dependent patterns of ventilation in 55

infants and children between 6 months and 9 years (27). Lying on the left side resulted in

significantly better ventilation of the left, but lying on the right produced equal ventilation

patterns.

In the diseased infant lung, the feasibility of EIT to monitor tidal ventilation has been long

established (31). EIT has been used to demonstrate significant regional breath-to-breath

variability in the distribution of delivered tidal volume (VT) during synchronized mechanical

ventilation even using tightly controlled, and constant, VT delivery (volume-guarantee) modes in

preterm infants (32). This study supports the previous observations made in healthy infants

regarding the complex behaviour of the tidal ventilation within the mild to moderately diseased

preterm infant lung, and also demonstrated similar gravity dependent changes associated with

prematurity. Post-menstrual age was also found to influence the distribution of tidal ventilation

during conventional ventilation. The authors postulated that the age-related changes were due

to differences in airway calibre associated with the primary reason for respiratory support;

infant RDS in early life and evolving or established chronic lung disease of prematurity later

(32). In summary, EIT has changed our understanding of regional ventilation distribution in the

infant lung and is highly suitable to monitor ventilation inhomogeneities in these populations.

c. Lung mechanics

Understanding lung mechanics is essential to optimal delivery of therapy for pediatric lung

diseases. Studies in animal models of pediatric and neonatal RDS, many using EIT, have

demonstrated significant regional differences in respiratory system compliance (8, 10, 12, 33-

36), time constants and filling characteristics of the lung (37). To date these findings have not

been replicated in human studies and the use of EIT to better understand regional lung

mechanics has been limited to the observational study of time constants in 22 preterm infants

with acute RDS receiving an open lung maneuver during HFOV (21). This study identified

distinct differences in the time constant of the lung depending on whether ventilation was on

the inflation or deflation limb of the pressure-volume relationship, and the surfactant state of

the lung at the time. The authors postulated that these differences were due to changes in

respiratory system compliance, a reasonable conclusion although this could not be directly

measured, due to the difficulties in bedside quasi-dynamic compliance measurements during

HFOV (38). The findings though were similar to those in pediatric and neonatal animal models

receiving an open lung maneuver, where direct measures of lung mechanics were possible (8,

12, 39). In summary, the measurement of the volume pressure relationship using EIT needs

7

further exploration, particularly as significant differences in regional visco-elastic properties are

likely to be present in acute pediatric and infant lung diseases.

Right versus left lung asymmetry

Right to left lung asymmetry is not generally recognized as a component of pediatric lung

disease. EIT is well suited to defining right to left lung asymmetry and has allowed these

assumptions to be challenged. In particular, the observations in spontaneously breathing term

and preterm infants that right to left lung asymmetry is markedly influenced by relatively minor

events such as turning of the head from one side to another or body position are important (25,

26). In mechanically ventilated infants body and head position only influenced ventilation

symmetry in the left lung and only during prone (26).

In piglets, accurate identification of ETT position within the respiratory tree and

oesophagus is possible with EIT, and superior to current bedside tools (40). This was confirmed

in a clinical study on pediatric patients scheduled for routine cardiac catheterization (41). Prior

to the intervention requiring anesthesia and endotracheal intubation, the children were

examined by EIT during intentional positioning of ETT in the right main bronchus and after its

correction. The authors concluded that right to left lung symmetry could be used to confirm

ETT position.

Breathing patterns

EIT has been used to demonstrate that breathing patterns, particularly rate and effort, strongly

influence the distribution of tidal ventilation in spontaneously breathing healthy infants (25).

The observation that deep sighs and shallow small tidal breathing patterns result in different

patterns of ventilation distribution has questioned the traditional understanding that infant and

adult ventilation distribution differs fundamentally. More importantly, the ability of EIT to make

multiple prolonged intra-subject recordings has shown that the ventilation distribution is a

highly complex phenomenon in the human lung. Since EIT measurements do not interact with

the breathing pattern of the subject, EIT is an ideal instrument to explore not only breathing

pattern but also changes in regional ventilation distribution over time, i.e. during REM and

NREM sleep.

8

Lung perfusion

The distribution of lung perfusion in the diseased pediatric lung is poorly understood, especially

in newborn infants. The presence, and clinical significance, of mis-matching of regional

ventilation and perfusion is greater during active lung disease or right heart disease. EIT may

offer potential to monitor both ventilation and perfusion (16, 42, 43). To date, studies in infants

and children on EIT-derived lung perfusion have been limited. A single study in 26 preterm

infants of varying postnatal age receiving synchronized conventional mechanical ventilation

showed that gravity-dependent differences in the EIT signal within the cardiac domain

(approximating relative perfusion) were present in the right lung (13). The non-dependent lung

regions contributed a greater component of the perfusion than the dependent regions. This

study showed that these differences were greatest in the first week of life. These findings

contributed to similar gravity-dependent differences in ventilation-perfusion matching, with

greater regional mismatching in infants less than a week old and requiring any form of oxygen

therapy.

Schibler and co-workers used the same EIT image reconstruction and filtering

methodology to evaluate 18 spontaneously breathing infants before and after ventricular

septum defect surgery, where a known and well established change in pulmonary circulation

results. A decrease in EIT-derived perfusion (with increased ventilation-to-perfusion ratios) was

seen in the middle and non-dependent regions of the lung following ventricular septum repair

(44). The role of EIT in assessing lung perfusion needs further clarification.

The following Table E8.1 provides an overview of EIT studies enhancing our knowledge of

neonatal/pediatric lung physiology.

9

Physiological principle addressed

Intention of study

Description of study subjects (age, disease, mode of support)

Type of EIT imaging

Physiological measures

Summary of results References

Lung hysteresis

Confirming existing concepts using human subjects

1. Preterm infants (RDS) receiving HFOV, surfactant and a recruitment maneuver

2. Pediatric RDS patients

n=15 (18) n=15 (45) n=9 (23)

Raw time course data

EELV 1. Confirmed regional lung hysteresis in preterm RDS. 2. Surfactant therapy rapidly improved and stabilized regional EELV. 3. Effect on EELV was most prominent in the dependent lung regions. 4. Atelectasis predominated in dependent lung regions and overdistension in non-dependent regions in pediatric RDS. 5. Response to lung recruitment depended on the degree of lung atelectasis.

(18, 23, 45)

Tidal volume

Identified new concepts

Healthy infants and children

n=10 (26) n=56 (27) n=32 (28) n=12 (25) n=24 (46)

fEIT images VT distribution 1. Tidal ventilation distribution was highly variable and influenced by many factors. 2. Distribution followed gravity dependent patterns in infants. 3. Lying on the left and right did not produce similar ventilation patterns in children.

(25-28, 46)

Identified new concepts

Preterm and term infants with RDS

n=20 (30) n=1 (31) n=27 (32)

fEIT images VT distribution Tidal ventilation was posture-dependent and demonstrated highly variable patterns on a breath-to-breath basis.

(30-32)

10


Intention of study


Type of EIT imaging



Right-to-left lung asymmetry

Identify new concepts

Observational studies in term and preterm spontaneously breathing and mechanically ventilated infants

n=12 (25) n=10 (26)

fEIT images of ventilation

VT distribution 1. Different patterns of right and left lung ventilation were seen in prone, supine and lateral positioning (especially in left lung) during spontaneous breathing. 2. Head position influenced symmetry of ventilation. 3. These patterns were lost in mechanically ventilated infants, except when positioned prone.

(25, 26)

Right-to-left lung asymmetry

Identify new concepts

18 children requiring anesthesia and endo-tracheal intubation prior to heart catheterization

fEIT images of ventilation

VT distribution Increase in right lung ventilation was found after intentional intubation of the right main bronchus. The right-to-left asymmetry was reduced after the ETT placement in trachea. Cases of un-intended left main bronchus and oesophagus intubations were identified.

(41)

Lung mechanics

Confirm existing concepts in human studies

22 preterm infants (RDS) receiving HFOV

Raw time course data

Time constant of respiratory system (actual and modelled data)

Differences in the time constant of the lung related to the volume state and position on quasi-static pressure-volume relationship of the lung.

(21)

Breathing patterns

Confirm existing concepts and identify new concepts

Term and preterm spontaneously breathing infants

n=12 (25) n=10 (26)

EIT waveforms, fEIT images of ventilation distribution

Breathing rate and regional VT

1. High variability of spontaneous neonatal breathing rate and VT was confirmed. 2. Regional ventilation distribution is instantaneously modified by changes in the breathing pattern.

(25, 26)

11


Intention of study


Type of EIT imaging



Lung perfusion

Confirming existing concepts

26 preterm infants (RDS) (13) and 18 spontaneously breathing infants before and after VSD repair (44)

fEIT images filtered to respiratory and cardiac domains

Tidal ventilation and heart beat related impedance changes

1. Cardiac signal greater in non-dependent lung regions. 2. Heterogeneous gravity dependent pattern of ventilation-perfusion mismatching. Improved ventilation-perfusion ratios with age and after VSD repair.

(13, 44)

Table E8.1. EIT to understand the physiological principles of the respiratory system in children and infants

EELV: end-expiratory lung volume; ETT: endotracheal tube; fEIT: functional EIT; HFOV: High-frequency oscillatory ventilation; IPPV:

intermittent positive pressure–controlled ventilation; PEEP: positive end-expiratory pressure; RDS: respiratory distress syndrome; VT:

tidal volume; VSD: ventricular septum defect

12

Understanding the interaction between clinical interventions and regional

volumetric behavior of the lung

Current cardiorespiratory function monitors generally consider the respiratory system as a single

compartment, assuming a uniformity of behavior and response to interventions. As shown in

the previous section this incorrectly generalizes and simplifies reality and may, potentially, lead

to inappropriate application of respiratory therapies, particularly applied airway pressure.

Ventilator-lung interaction

a. During conventional positive-pressure ventilation

In small observational studies the ability of EIT to demonstrate changes in ventilation patterns

correlating to changes in ventilator settings has been established, potentially offering the ability

of EIT to guide respiratory support at the bedside (31, 47). In 28 preterm infants, the pattern

of regional ventilation whilst receiving synchronized intermittent positive pressure ventilation

(SIPPV) with a targeted VT modality was described by Armstrong and co-workers (32). Meta-

analysis of targeted VT use in prematurity has suggested that this modality offers a modest but

clinically important reduction in chronic lung disease (48). To be optimal, it relies on accurate

feedback when the mechanical properties of the lung change. This is determined from gas flow

at the airway opening, a measure of global mechanical properties. EIT imaging of more than

3000 inflations demonstrated that the regional variability of targeted VT within the thorax was

significantly greater than that estimated at the airway opening. This variability was noted to

occur on a breath-by-breath basis, such that all lung regions frequently experienced rapidly

changing tidal ventilation. This observational study was unable to conclude whether alternative

PEEP levels would have altered this variability, but this study highlights the limitations, and

assumptions, of many common bedside respiratory function monitors that consider the

respiratory system as a single compartment, and the potential of EIT to be integrated into

modern feedback driven respiratory support modalities.

b. During non-invasive respiratory support

The main goal of non-invasive respiratory support is the preservation of adequate lung volume

during spontaneous breathing without the negative influences of ETT. This limits the ability of

accepted bedside tools of respiratory function monitoring. Hough and co-workers measured the

13

effect of body position on regional ventilation distribution in preterm infants receiving CPAP and

demonstrated that gravity had little impact (49). The same group also showed that the use of

high flow nasal cannula oxygen therapy using flow rates up to 8L/min increases EELV (50).

Miedema and co-workers measured changes in EELV and ventilation distribution during different

levels of nasal CPAP in 22 stable preterm infants (51). They showed that increasing CPAP from

2 to 6 cmH2O resulted in a homogenous increase in EELV. This optimization of EELV also

resulted in more homogeneous distribution of tidal ventilation.

Several studies have also addressed the effect of transitioning preterm infants from either

conventional or high-frequency ventilation to nasal CPAP. Carlisle and co-workers have

described the regional patterns of EELV and tidal ventilation during and up to 20 min after

extubation from positive-pressure ventilation via ETT to CPAP of 6-8 cm H2O in 10 preterm

infants (52). Significant loss of EELV occurred immediately following extubation, and recovery

was variable and not predicted from the post-extubation oxygen requirements. EELV losses in

the dependent lung regions were more persistent than in the non-dependent, and resulted in

increased ventilation inhomogeneity. Van der Burg and co-workers measured the changes in

EELV in 20 preterm infants extubated from high-frequency ventilation without oscillations

(endotracheal CPAP) to nasal CPAP. They report that this transition did not have a clinically

relevant impact on EELV (30). VT increased after extubation but its regional distribution did not

change. EIT has also been used to assess changes in EELV and VT during biphasic positive

airway pressure (BiPAP) in stable preterm infants (51). BiPAP using a pressure amplitude of 3

cmH2O above a continuous distending pressure of 6 cmH2O, with an inspiration time of 0.5 s,

did not have a significant effect on EELV or regional VT. So far, studies investigating the impact

of other modes of non-invasive ventilation on (regional) lung volumes using EIT have been

limited.

In summary, non-invasive respiratory modes of support, such as nasal CPAP, high flow

nasal cannula, and non-invasive ventilation have become the dominant form of respiratory

support in both pediatric and neonatal respiratory critical care (53). The increased popularity of

non-invasive support in these populations has often arisen without substantive mechanistic

knowledge of the physiological interaction with the diseased lung. EIT offers great potential to

address these significant and important shortcomings in knowledge.

c. During high-frequency oscillatory ventilation

EIT has been used in several studies on HFOV in preterm infants with RDS. Initial studies

focused on EELV changes during oxygenation-guided open lung HFOV. In addition to clarifying

14

some important physiological principles of the preterm lung as described in the previous

section, these studies also showed that EIT has the potential to guide the process of lung

recruitment at the bedside by mapping the inflation and deflation limb of the pressure-volume

relationship of the lung (18). Reassuringly, the pressure-volume relationships of the lung

described by EIT were similar to those described using respiratory inductive plethysmography in

term infants (19) and piglets (54) receiving HFOV, and could be described using accepted

mathematical models (55). Additional analyses also showed that the basic assumption of a

decoupling of oxygenation and ventilation during HFOV was too simplistic (56). Ventilation is

not only impacted by the pressure amplitude set during HFV, but also by the changing position

of oscillation on the pressure-volume relationship of the lung. As a result the compliance of the

respiratory system will change during lung recruitment and thus the subsequent oscillatory

volume. This finding, that the volume state of the lung influences lung compliance (and thus

VT), has been supported by more detailed analysis in term infants (57) and preterm lambs (39).

Zannin and co-workers also observed, in preterm lambs receiving open lung HFOV, that the

regional behavior of EELV, as described by EIT was highly predictive of the point of optimum

ventilation and lung mechanics.

Surfactant therapy

EIT is able to detect regional ventilation and aeration changes occurring in the lungs after

exogenous surfactant administration. An early case report of a preterm neonate with RDS

examined by EIT before and after endotracheal surfactant instillation revealed improved

ventilation and right-to-left ventilation symmetry after surfactant (47). Later experimental data

acquired in the preterm lamb and surfactant-depletion lavage piglet models of infant RDS

showed the ability of EIT to detect not only the ventilation redistribution but also dynamic

changes in regional lung filling and emptying and respiratory system mechanics in response to

surfactant administration (8, 9, 11, 33, 58, 59). Uniformity of regional ventilation related to

effective surfactant therapy was shown to correlate with regional decreases in early molecular

markers of injury (11). More recently, Milesi and co-workers demonstrated similar patterns of

regional ventilation with EIT in preterm lambs receiving bolus surfactant and surfactant via a

novel atomisation system design to be used during non-invasive ventilation (60).

Recently, clinical studies examining the effects of exogenous surfactant administration by

EIT have been published. In a study on preterm infants with RDS treated with surfactant, a

rapid increase in EELV that was most prominent in the dependent lung regions was detected

immediately after recruitment with open lung HFOV. Surfactant also stabilized EELV at much

15

lower mean airway pressures (45). Using raw EIT recordings, it was possible to show that

surfactant increased the quasi-static time constant of the respiratory system. In another study,

preterm infants with RDS ventilated in the synchronized intermittent positive-pressure and

synchronized intermittent mandatory ventilation modes were examined by EIT before, 15 min

and 30 min after surfactant therapy (61). Increased aeration and ventilation with improved

right-to-left lung symmetry were documented.

Role of EIT in body positioning

It has been known for some time that body positioning of the (preterm) newborn infant can

affect lung function. Studies using respiratory inductive plethysmography have shown that left

lateral and prone positioning improve EELV, VT and breathing synchrony (62). Studies using EIT

have provided more insight on the distribution of lung volume changes. As discussed in the

second section of this EOS, addressing the role of EIT in improving our understanding of

neonatal/pediatric lung physiology in response to postural effects, the effect of body positioning

on VT distribution is very variable (25-28, 46).

The effect on EELV has been recently studied by van der Burg and co-workers in preterm

infants on nasal CPAP (30). They showed that switching these infants from supine to prone

positioning improved EELV and that this increase was most prominent in the dorsal, non-

dependent lung regions.

Suction of the ETT

ETT suction is a frequently performed intervention in critical care. By the nature of its purpose,

it exposes the lung to intentional derecruitment, with potential consequences on lung volume

(63) and cardiorespiratory status (64). A small series of EIT studies have shown that ETT

suction, irrespective of how performed, can cause marked, albeit transient loss of EELV in

spontaneously breathing infants (65, 66). Whilst global lung volume derecruitment during

suction had been well described previously (63, 64, 67), these EIT studies demonstrated

regional complexities of volume changes during and after ETT suctions that had previously only

been shown in animals receiving muscle-relaxants (68, 69) . The patterns of volume change

after ETT suction were highly variable and not always correlated to oxygen needs post-ETT

suction. Often EELV losses were quickly regained (65). This has potential clinical importance

with regard to how clinicians identify patients needing recruitment maneuvers after ETT

suction. Whether the patterns of ventilation change within the lung after suction seen in animal

16

studies occurs in humans requires investigation. Further research in this area is needed but is

limited by the lack of realtime pediatric EIT systems.

During anesthesia

Surprisingly, the behavior of the lung during pediatric and neonatal anesthesia has seen little

investigations. Humpreys and co-workers identified a significant drop in EELV during induction

of anesthesia and intubation in 38 infants and children undergoing elective cardiac surgery,

which normalized with commencement of PEEP (70). In contrast, the distribution of ventilation

changed from a preferentially dependent lung pattern to non-dependent after induction and

starting positive pressure support. This highlights the potential of EIT as a monitoring tool to

optimize EELV during clinical interventions.

The following Table E8.2 summarizes the studies using EIT to examine the effects of clinical

interventions on regional lung function.

17

Clinical problem Clinical intervention

Patient population

Type of lung disease

Mode of respiratory support

Type of EIT imaging

Physiological measure (EELV, VT, etc.)


Distribution of tidal ventilation during conventional ventilation

SIPPV with VTV 28 stable preterm infants

Preterm RDS and evolving CLD

SIPPV+VTV fEIT images of VT

Pattern and variability of VT

1. VT distribution was highly variable and changed significantly on a breath-to-breath basis.2. Current bedside monitoring of VT under-estimates the variability of VT within the lung.

(32)

IPPV and body positioning (supine, prone, quarter-prone)

24 preterm infants

6 sponta-neously breathing infants

RDS IPPV fEIT images of VT

VT in four image ROIs (anterior, posterior, right left); GI index

1. Ventilated infants had higher global ventilation inhomogeneity than the healthy ones. 2. Ventilation distri-bution among ROIs was not affected by posture in ventilated infants posture-dependent in spontaneous breathing.

(71)

Distribution of tidal ventilation during HFV

Oxygenation-guided open lung HFV

Preterm infants

n=15 (18) n=10 (56)

RDS HFV Raw images EELV and VT 1. Regional hysteresis was present during the open lung maneuver. 2. The P-V relationship of the lung could be mapped and optimal EELV identified. 3. VT was influenced by the volume state of the lung.

(18, 56)

18


Patient population



Type of EIT imaging



Distribution of EELV and tidal ventilation during non-invasive respiratory support

CPAP 22 preterm infants

Prematurity Nasal CPAP Raw images EELV and VT 1. Increase in CPAP results in homogeneous increase in EELV 2. Increase in CPAP results in more homo-geneous distribution of tidal ventilation

(51)

Distribution of tidal ventilation during HFNC

HFNC 13 infants age<1 year

bronchiolitis HFNC at 8 l/min and 2 l/min

Raw images, fEIT images of VT

EELV and VT distribution

HFNC at 8 l/min increased anterior (and global) EELV

(50)

Distribution of EELV and tidal ventilation at extubation

Extubation from SIPPV+VTV

Extubation from HFV

Ten stable preterm infants (52)

20 preterm infants (30)

Resolving preterm RDS

SIPPV+VTV and CPAP

HFV

Raw data and fEIT images of VT

EELV and VT 1. Extubation resulted in significant EELV loss. 2. EELV recovery was variable globally and regionally. 3. EELV is maintained after transitioning from endotracheal to nasal CPAP 4. VT increases after extubation but its distribution does not

(30, 52)

Exogenous surfactant therapy

Exogenous surfactant administration

15 preterm infants

Acute RDS Post-recruitment with open lung HFV

Raw data EELV Surfactant quickly stabilized EELV, resulted in an optimal EELV at a lower mean pressure and increased time constants

(45)

19


Patient population



Type of EIT imaging



Effect of BiPAP on tidal volumes

Cross-over study switching from nCPAP to BiPAP and back to nCPAP

22 stable preterm infants on non-invasive respiratory support

prematurity BiPAP Raw data and fEIT ventilation images

EELV and VT BiPAP does not impact EELV or VT

(51)

Effect of body positioning on EELV

Switching from supine to prone position

20 preterm infants

Resolving RDS

nCPAP Raw data EELV Prone positioning increases EELV and favors the ventilation of anterior lung regions

(30)

ETT suction Closed ETT suction

Preterm infants

n=22 (65) n=11 (66)

RDS HFOV and conventional ventilation

Raw data EELV 1. Rapid loss of EELV in all regions during ETT suction. 2. Generally rapid resolution of EELV post suction

(65, 66)

Anaesthesia Induction, intubation and starting positive pressure ventilation

38 infants and children undergoing elective pediatric cardiac surgery

None Spontaneous-ly breathing to positive pressure support

Raw data and fEIT images of ventilation

EELV and VT distribution

1. Induction of anesthe-sia and intubation resulted in transient loss of EELV that was rectified using PEEP. 2. Induction of anesthe-sia and positive pressure ventilation resulted in significant changes in ventilation distribution

(70)

Table E8.2. EIT to understand the interaction between clinical interventions and regional volumetric behaviour of the pediatric and

newborn human lung.

20

BiPAP: biphasic positive airway pressure; CLD: chronic lung disease; CPAP: continuous positive airway pressure; EELV: end-expiratory

lung volume; fEIT: functional EIT; GI: global inhomogeneity; HFNC: high-flow nasal cannula; HFV: high-frequency ventilation; IPPV:

intermittent positive pressure–controlled ventilation; nCPAP: nasal continuous positive airway pressure; PEEP: positive end-expiratory

pressure; P-V: pressure volume; RDS: respiratory distress syndrome; ROI: region of interest; SIPPV: synchronized intermittent positive-

pressure ventilation; VT: tidal volume; VTV: volume-targeted ventilation

21

Use of EIT to guide clinical care in infants and children

For EIT to have any meaningful role in pediatric and neonatal critical and respiratory care, it is

essential that EIT be used to guide clinical care with proven evidence of benefit to outcomes.

Research aimed at demonstrating a positive clinical outcome using EIT as either a tool to direct

therapy (11, 12, 36) or identify the need to alter therapy (40, 72) have been limited to animal

model studies. The potential versatility of EIT with regards to clinical environments, therapies,

populations and applications would suggest promise as a practical clinical tool. To date,

research in this field has been limited to observational studies aimed at demonstrating the

feasibility of EIT in a particular clinical role. Generally the proposed clinical pediatric roles for

EIT can be classified as either directing applied pressure during artificial respiratory support,

monitoring the clinical state of the lung during other therapies or identifying adverse events

prior to clinical deterioration.

Set applied pressure values

Lung-protective ventilation settings may vary significantly in the context of the individual lung

disease. As RDS is a heterogeneous disease, parameters minimizing alveolar overdistension and

atelectasis might be different in each patient, thus warranting a personalized approach (73). No

human studies exist yet that have used EIT to prospectively guide the clinical identification and

setting of applied pressure values. The ability of EIT to map the relationship between applied

pressure and volume state of the lung (18), and track time to volume stability after a pressure

change (21), suggests that EIT-guided ventilator algorithms would be possible.

The pressing question as to whether EIT can be effectively utilized to prospectively guide

lung protective parameters was first described in an animal model of acute lung injury (12),

comparing an EIT-guided algorithm of PEEP-titration to conventional ARDSnet lung protective

ventilation in piglets with an induced model of ARDS. Piglets were randomized to receive either

EIT-guided ventilation using the open lung concept (74) or ventilation according to the ARDSnet

protocol. During EIT-guided ventilation, the level of PEEP was adjusted every two-minutes using

real time EIT imaging until all atelectatic areas where recruited. Once atelectasis had resolved

in all lung regions, often at the expense of overdistension in previously well-recruited non-

dependent areas, PEEP was reduced until overdistension was not evident on EIT but

recruitment had been maintained. EIT was then used to guide further intervention if atelectasis

re-occurred. Resultant final PEEP levels were higher in the EIT-guided group. Plateau pressures

did not differ, reflecting the better compliance and regional volume states (on CT imaging and

22

EIT) in the EIT-guided group. The EIT-guided group also had less histopathological evidence of

lung injury, with improved alveolar capillary vascular permeability. There was excellent

correlation between CT and EIT images with regard to volume state of the lung.

More recently, EIT has been used to guide the delivery of a sustained inflation during

resuscitation of a newly-born preterm lamb (75). This study found that the time needed to

optimally aerate the lung at birth was highly variable, and the EIT-guided sustained inflation

group had better gas exchange, lung mechanics, regional ventilation patterns and EELV 60-min

after birth than a group of lambs receiving a sustained inflation consistent with current

resuscitation guidelines. The same group then extended these findings to compare a sustained

inflation (SI) guided by the realtime volume response of the global EIT signal against a pre-

determined 30-s SI at birth and an intentional long SI (36). The EIT-guided SI provided the

best clinical response and less injury than the 30-s SI. Interestingly, there was no difference in

outcomes between the EIT-guided and long SI (still EIT-guided). Importantly, this study

demonstrated that EIT might have an interventional role in the Delivery Room, where existing

monitoring is limited.

Optimum PEEP was identified in very low birth weight preterm infants ventilated in an

assisted mode using EIT during a decremental PEEP trial prior to extubation (76). A simple EIT

measure, the ratio between the ventilation of the upper and lower chest region, was used as

the decision criterion. The PEEP at which the ventilation distribution was most homogeneous

was detected and compared with the empirically set pressure chosen by the attending staff.

The EIT-derived PEEP value was significantly higher than the routinely set pressure after

extubation to CPAP. The unique feature of EIT used during assisted ventilation is that it allows

separate analysis of ventilation distribution during spontaneous and ventilator-generated

breaths (31, 77).

These studies show that real time EIT image-guidance could produce a real difference in

outcome if properly implemented. Studies involving patients need to follow in order to confirm

those promising results.

Monitor clinical state of the lung

Although EIT has been used extensively in small observational trials to describe the clinical

state of the lung (see previous section), there have not been any human studies examining the

potential to influence outcomes in human infants and children. Such studies will be needed if

EIT is to be seen as a valuable alternative to existing bedside monitoring systems.

23

Identify adverse events

Adverse events are common in pediatric and neonatal critical care (72). The bedside recognition

of an adverse event usually requires some form of clinical deterioration, often after a prodromal

period in which current bedside monitors fail to detect meaningful change. In this environment,

the direct ability of EIT to continually (and without radiation) display the regional VT states of

the lung offers promise as an effective clinical tool. The potential role of EIT to guide ETT

suction has been discussed in the previous section. EIT has been shown, in piglets with a

controlled pneumothorax, to be able to identify as little as 10 mL of free air in the pleural

space, and significantly earlier than any deterioration in oxygen saturation or heart rate (≥100

mL free air) (72). In another study involving piglets, EIT was compared with standard bedside

tools (colorimetric capnography, oxygen saturation, heart rate and blood pressure and airway

flow) during endotracheal intubation. Whilst all techniques were able to quickly identify

oesophageal intubation, only EIT could indicate the location of the ETT within the respiratory

tree, specifically whether the ETT was located in the trachea or malpositioned in a main

bronchus (40). This was confirmed in a clinical study on children requiring general anesthesia

and endotracheal intubation (41). These children were examined by EIT during one-lung

ventilation resulting from intentional brief position of ETT in the right main bronchus. The

authors demonstrated that the non-intended placement of the tube in the left main bronchus

and oesophagus in two children was also correctly identified by EIT. These findings imply that

EIT imaging of right-to-left lung ventilation symmetry could be used to determine ETT location

without the need for chest radiography.

Data acquired in infants in a routine setting are currently limited to several case reports.

Miedema and co-workers presented a preterm infant with a unilateral pneumothorax detected

by EIT and confirmed by chest radiography (78). Van der Burg and colleagues published the

EIT recording of a preterm infant with unilateral atelectasis (79). Figure E8.2 shows a case of a

ventilated neonate in whom a highly asymmetric ventilation distribution identified by EIT led to

the detection of an ETT malposition. The ventilation distribution improved immediately after the

intubation depth had been reduced by ETT withdrawal and was confirmed during a repeated

EIT examination carried out 6 min later.

24

Figure E8

newborn w

distribution

(bottom).

left lungs.

right side

Goe-MF II

Com

without co

adverse ev

may be th

to determi

meaningfu

should inc

mortality, m

The

EIT in ther

8.2. EIT e

with a body

n (top) wh

The wavefo

The orienta

of the ches

EIT device

mon to all t

omplex post

vent, and a

e simplest

ne whether

ul outcome m

clude radiat

morbidity a

following T

rapy guidan

examination

y weight of

hich immed

orms in the

ation of the

st is on the

(CareFusio

these studi

t-hoc recons

always befo

method of

r EIT can id

measures w

tion need,

nd time to d

Table E8.3 g

nce in neona

of a 3-w

f 3340 g. E

diately disap

e panels hig

functional

left side o

n, Höchberg

es was the

struction, w

re cardiores

introducing

dentify adve

will need to

cost effec

discharge.

gives a sum

atal/pediatr

k old intub

EIT detecte

ppeared af

ghlight the

EIT ventilat

of the image

g, Germany

observatio

were useful

spiratory co

g EIT to clin

erse events

be conside

ctiveness as

mmary of stu

ic lung dise

bated and

ed an asym

fter the co

magnitude

tion images

e. (The raw

y).)

n that very

in rapidly in

ompromise.

nical setting

s earlier tha

red prior to

s well as

udies releva

ease.

mechanica

mmetric righ

rrection of

of ventilati

s: anterior is

w data were

simple rea

ndicating th

Using EIT

gs. Further

an current m

o conducting

long-term

ant for esta

ally ventilat

ht-to-left ve

f the tube

on in the r

s at the top

e recorded

al-time fEIT

he occurren

within this

studies are

methods. T

g these stud

outcomes

ablishing the

ed term

entilation

position

right and

p and the

with the

images,

nce of an

s context

e needed

The most

dies, but

such as

e role of

25

Clinical problem

Clinical inter-vention

Patient popu-lation



Type of EIT imaging

Physio-logical measure

Summary of results Outcomes References

Setting applied pressures

Open lung PEEP maneuver guided by EIT

12 piglets Induced ARDS

Conventional ventilation

Raw data and fEIT images

EELV and regional VT

EIT-guided recruitment strategy produced better gas exchange, lung mechanics and less lung injury than current ARDS-Net strategy

Suggests that EIT may be able to guide setting of the optimal applied pressure.

(12)

Decremental PEEP trial, extubation

14 preterm infants

RDS Conventional ventilation

fEIT ventilation images

U/L ventilation ratio

EIT-guided PEEP selection aimed at optimizing anteroposterior ventilation distribution and resulted in higher pressure values than the empirically chosen

EIT may be able to guide setting of the applied pressure after extubation during initiation of non-invasive ventilation support.

(76)

Identifi-cation of pneumo-thorax

Sequential introduction of gas into right pleural space

6 piglets

Induced ARDS



EELV, aeration maps and VT

EIT was able to detect 10 mL of pleural gas, and significantly earlier than deoxygenation or bradycardia

EIT may identify pneumothoraces before clinical deterioration

(72)

Case report 1 preterm infant

RDS HFOV fEIT image VT distribu-tion

fEIT image showed a decrease in tidal ventilation at the affected side

EIT was able to detect pneumothorax in preterm infants

(78)

Identifi-cation of atelectasis

Case report 1 preterm infant

RDS nCPAP fEIT ventilation images

VT distribution

fEIT showed a decrease in tidal ventilation on the affected side

EIT was able to detect atelectasis in preterm infants

(79)

26

Clinical problem


Patient popu-lation



Type of EIT imaging



Identifi-cation of recruitability and over-distension

Sustained inflation and stepwise recruitment

10 children ARDS Conventional ventilation

fEIT images of regional VT and Crs

Change in regional Crs

compared with best Crs (80)

1. EIT identified responders and non-responders to recruitment maneuver. 2. Collapsed regions were better recruited by stepwise recruitment maneuver than sustained inflation. 3. Regional overdistension was more pronounced in non-responders but occurred in both groups.

EIT identified recruitable atelectasis and overdistension

(22)

Location of ETT position

Comparison of different clinical tools for detecting correct tracheal, single bronchus and oesophageal ETT location with EIT.

6 piglets Induced ARDS



EELV, aeration maps and VT.

Real time fEIT images were the only method that could identify oesophageal intubation, correctly located ETT and single bronchus intubation.

fEIT ventilation images may hold potential as a radiation-free method of determining the correct ETT position.

(40)

Intentional brief single lung ventilation

18 intu-bated children during anaesthesia

Healthy Conventional ventilation

fEIT images

Right and left lung VT patterns

EIT identified: 1. Single lung ventilation due to ETT malposition.

2. Non-intended left main bronchus and oesophageal ETT placements in 2 children

fEIT may have potential to aid intubation during anesthesia

(41)

27

Clinical problem


Patient popu-lation



Type of EIT imaging



Guiding resuscitation at birth

Delivery of a sustained lung inflation

35 preterm lambs

RDS Neopuff™ and conventional ventilation

Raw data and fEIT

EELV and regional VT

EIT-guided SI at birth resulted in better gas exchange, lung mechanics and less lung injury than a 30-s SI.

Suggest that EIT may be able to guide optimising respiratory support at birth.

(36)

Table E8.3. Summary of animal and human studies using of EIT to guide clinical care in pediatric and newborn lung disease.

ARDS: acute respiratory distress syndrome; Crs: respiratory system compliance; EELV: end-expiratory lung volume; ETT: endotracheal

tube; fEIT: functional EIT; HFOV: high-frequency oscillatory ventilation; nCPAP: nasal positive airway pressure; PEEP: positive end-

expiratory pressure; RDS: respiratory distress syndrome; U/L: upper-to-lower; VT: tidal volume

28

Practical limitations unique to infants and children

This section will focus on the limitations specific to the pediatric and infant population. In

general these can be summarized as limitations related to the population itself, engineering and

design and defining the clinical role. To date there are no suitable EIT systems available for

clinical use in children and infants, this is despite the availability of a number of systems

approved for adult use. The intrinsic factors that make EIT more attractive as a clinical tool in

children and infants also create specific practical limitations that, to a certain extent, account for

this discrepancy. The general limitations of EIT have been discussed in detail elsewhere, and

most are as pertinent in children as adults, particularly the need for realtime image

reconstruction and display.

The pediatric population has a greater variation in size, with any pediatric EIT system

needing to adapt to patients from <500 g to >70 kg in weight. This offers unique problems

relating to electrode design and also image reconstruction algorithms. In younger patients the

small torso limits space for any monitoring, and the addition of EIT electrodes compete with

other monitoring systems, such as transcutaneous gas monitoring. The close proximity with

existing monitoring systems increases the risk of electrical interference and noise, especially in

the already electrically crowded critical care environment (81). Newer systems, with better

shielding should be able to negate these concerns. Although electrode interfaces exist for

adults, it cannot be assumed that simply miniaturising adult designs will be appropriate for the

pediatric population, who often have impaired skin integrity and need higher humidity

environments. A single EIT patient interface, such as a flexible belt or band, containing an array

of electrodes appears to be the most suitable solution in this population. Such an interface will

need to be designed specifically for the shape, and movement, of the pediatric and infant

chests, which varies considerably during the different developmental stages of childhood. At the

same time, it must be sufficiently compliant to avoid constricting the chest. Image

reconstruction will also need to consider these differences and use algorithms specifically

designed for the patient population if clinicians are to be provided with accurate tomographic

information, and thus maximize the potential for meaningful outcomes. To date, existing

commercial EIT systems use a single algorithm designed for an adult chest shape and would

not be suitable for children and infants. Movement artefact and electrode contact failure are

likely to be higher in the non-compliant child and infant.

Which EIT parameters to display at the bedside, and in what format, have yet to be

agreed upon. Fundamental questions such as whether to use absolute or relative images and

29

which ones (for example ventilation, aeration and perfusion), whether EIT will simply be a

diagnostic or a monitoring tool or specifically direct clinical care, and how to train clinicians to

interpret EIT images need to be addressed before commercial systems, with broad and

meaningful utility, will be possible.

Recommendations for future direction

The exact role of EIT in the pediatric and neonatal population, especially the critical care

environment has yet to be systematically established and consensus statements are clearly

needed. The goal of future research and device development should be to create practical and

functional EIT systems in which the role of EIT in different clinical settings can be determined

and then evaluated to ascertain whether benefits to outcome are possible. Through such an

approach EIT is likely to become widely adopted.

Pediatric-specific electrode interfaces and image reconstruction algorithms are urgently

needed. The growing market in pediatric and neonatal critical care, and particularly non-

invasive respiratory support, justifies the commercial potential. These systems should be

developed through collaboration between clinicians and medical equipment manufacturers so

acceptable pediatric solutions are found. Electrode interfaces and display of information will

need to be individualized to this diverse population and needs.

In the first instance, EIT is likely to be more easily adopted as a monitoring tool,

particularly for early identification of simple adverse events and/or in environments where

existing monitoring tools are impractical, particularly during non-invasive respiratory support.

Such an approach will allow evaluation of the ability of EIT to alter existing high-risk practices

(for example intubation) and investigations (chest radiography) without compromising safety

and outcomes. From these experiences it is likely that the development of EIT systems that can

be used to guide clinical practices, especially in the more meaningful but harder setting of

applied pressures, can occur. The ultimate goal of such systems must be to demonstrate

improvements in outcomes if EIT is to progress beyond a research tool in pediatric and

neonatal practice.

Conclusions

Although EIT is a promising technique that provides the clinician valuable information on

changes in lung aeration and perfusion, studies exploring its use to guide clinical practice are

30

still lacking. Improvement in hardware, interface and software are urgently needed before EIT

can take up this challenge. With these solutions it is likely that EIT will find a role in monitoring

clinical care, especially during non-invasive ventilation.


The first draft of this online document was prepared by D. Tingay with collaboration of A. H.

van Kaam, I. Frerichs, G. K. Wolf, A. Schibler, T. Riedel and P. C. Rimensberger. It was

reviewed and approved by all other authors and collaborators.

31

References

1. Schibler A, Frey U. Role of lung function testing in the management of mechanically

ventilated infants. Arch Dis Child Fetal Neonatal Ed 2002;87:F7-F10.

2. Hammer J, Newth CJ. Infant lung function testing in the intensive care unit. Intensive Care

Med 1995;21:744-752.

3. Heulitt M, Donn SM. Monitoring interactions between spontaneous respiration and

mechanical inflations. In: Rimensberger PC, editor. Pediatric and neonatal mechanical

ventilation Berlin Heidelberg: Springer-Verlag; 2015. p. 473-489.

4. Salyer J. Monitoring oxygenation of ventilated infants and children. In: Rimensberger PC,

editor. Pediatric and neonatal mechanical ventilation. Berlin Heidelberg: Springer-Verlag; 2015.

p. 375-387.

5. Schmalisch G, Cheifetz IM. Ventilation: Adequacy of breathing assessment. In:

Rimensberger PC, editor. Pediatric and neonatal mechanical ventilation. Berlin Heidelberg:

Springer-Verlag; 2015. p. 388-404.

6. Tobias JD. Transcutaneous carbon dioxide monitoring in infants and children. In:



7. Vogt B, Falkenberg C, Weiler N, Frerichs I. Pulmonary function testing in children and

infants. Physiol Meas 2014;35:R59-90.

8. Dargaville PA, Rimensberger PC, Frerichs I. Regional tidal ventilation and compliance during

a stepwise vital capacity manoeuvre. Intensive Care Med 2010;36:1953-1961.

9. Frerichs I, Dargaville PA, van Genderingen H, Morel DR, Rimensberger PC. Lung volume

recruitment after surfactant administration modifies spatial distribution of ventilation. Am J


10. Tingay DG, Bhatia R, Schmolzer GM, Wallace MJ, Zahra VA, Davis PG. Effect of

sustained inflation vs. stepwise PEEP strategy at birth on gas exchange and lung mechanics in

preterm lambs. Pediatr Res 2014;75:288-294.


Davis PG. Surfactant before the first inflation at birth improves spatial distribution of ventilation

and reduces lung injury in preterm lambs. J Appl Physiol 2014;116:251-258.

12. Wolf GK, Gomez-Laberge C, Rettig JS, Vargas SO, Smallwood CD, Prabhu SP, Vitali SH,

Zurakowski D, Arnold JH. Mechanical ventilation guided by electrical impedance tomography in

experimental acute lung injury. Crit Care Med 2013;41:1296-1304.

32

13. Carlisle HR, Armstrong RK, Davis PG, Schibler A, Frerichs I, Tingay DG. Regional

distribution of blood volume within the preterm infant thorax during synchronised mechanical

ventilation. Intensive Care Med 2010;36:2101-2108.

14. Fagerberg A, Sondergaard S, Karason S, Aneman A. Electrical impedence tomography

and heterogeneity of pulmonary perfusion and ventilation in porcine acute lung injury. Acta


15. Frerichs I, Pulletz S, Elke G, Reifferscheid F, Schadler D, Scholz J, Weiler N. Assessment

of changes in distribution of lung perfusion by electrical impedance tomography. Respiration

2009;77:282-291.

16. Grant CA, Pham T, Hough J, Riedel T, Stocker C, Schibler A. Measurement of ventilation

and cardiac related impedance changes with electrical impedance tomography. Crit Care

2011;15:R37.

17. Pelosi P, D'Andrea L, Vitale G, Pesenti A, Gattinoni L. Vertical gradient of regional lung

inflation in adult respiratory distress syndrome. Am J Respir Crit Care Med 1994;149:8-13.


lung volume and ventilation during lung recruitment in high-frequency ventilated preterm infants

with respiratory distress syndrome. J Pediatr 2011;159:199-205 e192.

19. Tingay DG, Mills JF, Morley CJ, Pellicano A, Dargaville PA. The deflation limb of the

pressure-volume relationship in infants during high-frequency ventilation. Am J Respir Crit Care

Med 2006;173:414-420.

20. van der Burg PS, Miedema M, de Jongh FH, Frerichs I, van Kaam AH. Cross-sectional

changes in lung volume measured by electrical impedance tomography are representative for

the whole lung in ventilated preterm infants. Crit Care Med 2014;42:1524-1530.




22. Wolf GK, Gomez-Laberge C, Kheir JN, Zurakowski D, Walsh BK, Adler A, Arnold JH.

Reversal of dependent lung collapse predicts response to lung recruitment in children with early

acute lung injury. Pediatr Crit Care Med 2012;13:509-515.

23. Gomez-Laberge C, Arnold JH, Wolf GK. A unified approach for EIT imaging of regional

overdistension and atelectasis in acute lung injury. IEEE Trans Med Imaging 2012;31:834-842.

24. Davies H, Kitchman R, Gordon I, Helms P. Regional ventilation in infancy. Reversal of

adult pattern. N Engl J Med 1985;313:1626-1628.

25. Frerichs I, Schiffmann H, Oehler R, Dudykevych T, Hahn G, Hinz J, Hellige G. Distribution

of lung ventilation in spontaneously breathing neonates lying in different body positions.


33

26. Heinrich S, Schiffmann H, Frerichs A, Klockgether-Radke A, Frerichs I. Body and head

position effects on regional lung ventilation in infants: An electrical impedance tomography

study. Intensive Care Med 2006;32:1392-1398.

27. Lupton-Smith AR, Argent AC, Rimensberger PC, Morrow BM. Challenging a paradigm:

Positional changes in ventilation distribution are highly variable in healthy infants and children.


28. Pham TM, Yuill M, Dakin C, Schibler A. Regional ventilation distribution in the first 6

months of life. Eur Respir J 2011;37:919-924.

29. Riedel T, Kyburz M, Latzin P, Thamrin C, Frey U. Regional and overall ventilation

inhomogeneities in preterm and term-born infants. Intensive Care Med 2009;35:144-151.

30. van der Burg PS, Miedema M, de Jongh FH, Frerichs I, van Kaam AH. Changes in lung

volume and ventilation following transition from invasive to noninvasive respiratory support and

prone positioning in preterm infants. Pediatr Res 2015;77:484-488.

31. Frerichs I, Hahn G, Schiffmann H, Berger C, Hellige G. Monitoring regional lung

ventilation by functional electrical impedance tomography during assisted ventilation. Ann N Y

Acad Sci 1999;873:493-505.

32. Armstrong RK, Carlisle HR, Davis PG, Schibler A, Tingay DG. Distribution of tidal

ventilation during volume-targeted ventilation is variable and influenced by age in the preterm

lung. Intensive Care Med 2011;37:839-846.

33. Frerichs I, Dargaville PA, Rimensberger PC. Regional respiratory inflation and deflation

pressure-volume curves determined by electrical impedance tomography. Physiol Meas

2013;34:567-577.




35. Tingay DG, Lavizzari A, Zonneveld CE, Rajapaksa A, Zannin E, Perkins E, Black D,

Sourial M, Dellaca RL, Mosca F, Adler A, Grychtol B, Frerichs I, Davis PG. An individualized

approach to sustained inflation duration at birth improves outcomes in newborn preterm lambs.

Am J Physiol Lung Cell Mol Physiol 2015;309:L1138-1149.

36. Tingay DG, Rajapaksa A, Zonneveld CE, Black D, Perkins EJ, Adler A, Grychtol B,

Lavizzari A, Frerichs I, Zahra VA, Davis PG. Spatiotemporal aeration and lung injury patterns

are influenced by the first inflation strategy at birth. Am J Respir Cell Mol Biol 2016;54:263-272.

37. Grant CA, Fraser JF, Dunster KR, Schibler A. The assessment of regional lung mechanics

with electrical impedance tomography: A pilot study during recruitment manoeuvres. Intensive

Care Med 2009;35:166-170.

38. Rimensberger PC, Tingay DG. Respiratory system compliance (i.e. the pressure-volume

curve relationship) during ongoing mechanical ventilation. In: Rimensberger PC, editor.

34

Pediatric and neonatal mechanical ventilation. Berlin Heidelberg: Springer-Verlag; 2015. p. 421-

434.

39. Zannin E, Ventura ML, Dellaca RL, Natile M, Tagliabue P, Perkins EJ, Sourial M, Bhatia

R, Dargaville PA, Tingay DG. Optimal mean airway pressure during high-frequency oscillatory

ventilation determined by measurement of respiratory system reactance. Pediatr Res

2014;75:493-499.

40. Schmolzer GM, Bhatia R, Davis PG, Tingay DG. A comparison of different bedside

techniques to determine endotracheal tube position in a neonatal piglet model. Pediatr

Pulmonol 2013;48:138-145.

41. Steinmann D, Engehausen M, Stiller B, Guttmann J. Electrical impedance tomography for

verification of correct endotracheal tube placement in paediatric patients: A feasibility study.

Acta Anaesthesiol Scand 2013;57:881-887.

42. Frerichs I, Hinz J, Herrmann P, Weisser G, Hahn G, Quintel M, Hellige G. Regional lung

perfusion as determined by electrical impedance tomography in comparison with electron beam

CT imaging. IEEE Trans Med Imaging 2002;21:646-652.

43. Borges JB, Suarez-Sipmann F, Bohm SH, Tusman G, Melo A, Maripuu E, Sandstrom M,

Park M, Costa EL, Hedenstierna G, Amato M. Regional lung perfusion estimated by electrical

impedance tomography in a piglet model of lung collapse. J Appl Physiol 2012;112:225-236.

44. Schibler A, Pham TM, Moray AA, Stocker C. Ventilation and cardiac related impedance

changes in children undergoing corrective open heart surgery. Physiol Meas 2013;34:1319-

1327.


lung volume and ventilation during surfactant treatment in ventilated preterm infants. Am J


46. Schibler A, Yuill M, Parsley C, Pham T, Gilshenan K, Dakin C. Regional ventilation

distribution in non-sedated spontaneously breathing newborns and adults is not different.


47. Frerichs I, Schiffmann H, Hahn G, Hellige G. Non-invasive radiation-free monitoring of

regional lung ventilation in critically ill infants. Intensive Care Med 2001;27:1385-1394.

48. Wheeler K, Klingenberg C, McCallion N, Morley CJ, Davis PG. Volume-targeted versus

pressure-limited ventilation in the neonate. Cochrane Database Syst Rev 2010:CD003666.

49. Hough JL, Johnston L, Brauer SG, Woodgate PG, Pham TM, Schibler A. Effect of body

position on ventilation distribution in preterm infants on continuous positive airway pressure.

Pediatr Crit Care Med 2012;13:446-451.

50. Hough JL, Pham TM, Schibler A. Physiologic effect of high-flow nasal cannula in infants

with bronchiolitis. Pediatr Crit Care Med 2014;15:e214-219.

35

51. Miedema M, van der Burg PS, Beuger S, de Jongh FH, Frerichs I, van Kaam AH. Effect of

nasal continuous and biphasic positive airway pressure on lung volume in preterm infants. J

Pediatr 2013;162:691-697.

52. Carlisle HR, Armstrong RK, Davis PG, Kamlin O, Argus B, Schibler A, Tingay DG.

Ventilation is transiently re-distributed towards the non-dependent lung during elective

extubation in preterm infants. Meeting of the Pediatric Academic Societies. Baltimore; 2009. p.

3450.

53. Chow SSW. Report of the Australian and New Zealand Neonatal Network 2012. Sydney:

ANZZN; 2014.

54. Brazelton TB, 3rd, Watson KF, Murphy M, Al-Khadra E, Thompson JE, Arnold JH.

Identification of optimal lung volume during high-frequency oscillatory ventilation using

respiratory inductive plethysmography. Crit Care Med 2001;29:2349-2359.

55. Venegas JG, Harris RS, Simon BA. A comprehensive equation for the pulmonary

pressure-volume curve. J Appl Physiol 1998;84:389-395.

56. Miedema M, de Jongh FH, Frerichs I, van Veenendaal MB, van Kaam AH. The effect of

airway pressure and oscillation amplitude on ventilation in pre-term infants. Eur Respir J

2012;40:479-484.

57. Tingay DG, Mills JF, Morley CJ, Pellicano A, Dargaville PA. Indicators of optimal lung

volume during high-frequency oscillatory ventilation in infants. Crit Care Med 2013;41:237-244.




59. Frerichs I, Dargaville PA, Dudykevych T, Rimensberger PC. Electrical impedance

tomography: A method for monitoring regional lung aeration and tidal volume distribution?


60. Milesi I, Tingay DG, Zannin E, Bianco F, Tagliabue P, Mosca F, Lavizzari A, Ventura ML,

Zonneveld CE, Perkins EJ, Black D, Sourial M, Dellaca RL. Intratracheal atomized surfactant

provides similar outcomes as bolus surfactant in preterm lambs with respiratory distress

syndrome. Pediatr Res 2016.

61. Chatziioannidis I, Samaras T, Mitsiakos G, Karagianni P, Nikolaidis N. Assessment of

lung ventilation in infants with respiratory distress syndrome using electrical impedance

tomography. Hippokratia 2013;17:115-119.

62. Gouna G, Rakza T, Kuissi E, Pennaforte T, Mur S, Storme L. Positioning effects on lung

function and breathing pattern in premature newborns. J Pediatr 2013;162:1133-1137, 1137

e1131.

63. Tingay DG, Copnell B, Mills JF, Morley CJ, Dargaville PA. Effects of open endotracheal

suction on lung volume in infants receiving HFOV. Intensive Care Med 2007;33:689-693.

36

64. Hoellering AB, Copnell B, Dargaville PA, Mills JF, Morley CJ, Tingay DG. Lung volume

and cardiorespiratory changes during open and closed endotracheal suction in ventilated

newborn infants. Arch Dis Child Fetal Neonatal Ed 2008;93:F436-441.

65. Hough JL, Shearman AD, Liley H, Grant CA, Schibler A. Lung recruitment and

endotracheal suction in ventilated preterm infants measured with electrical impedance

tomography. J Paediatr Child Health 2014;50:884-889.

66. van Veenendaal MB, Miedema M, de Jongh FH, van der Lee JH, Frerichs I, van Kaam

AH. Effect of closed endotracheal suction in high-frequency ventilated premature infants

measured with electrical impedance tomography. Intensive Care Med 2009;35:2130-2134.

67. Tan AM, Gomez JM, Mathews J, Williams M, Paratz J, Rajadurai VS. Closed versus

partially ventilated endotracheal suction in extremely preterm neonates: Physiologic

consequences. Intensive Crit Care Nurs 2005;21:234-242.

68. Tingay DG, Copnell B, Grant CA, Dargaville PA, Dunster KR, Schibler A. The effect of

endotracheal suction on regional tidal ventilation and end-expiratory lung volume. Intensive

Care Med 2010;36:888-896.

69. Hepponstall JM, Tingay DG, Bhatia R, Loughnan PM, Copnell B. Effect of closed

endotracheal tube suction method, catheter size, and post-suction recruitment during high-

frequency jet ventilation in an animal model. Pediatr Pulmonol 2012;47:749-756.

70. Humphreys S, Pham TM, Stocker C, Schibler A. The effect of induction of anesthesia and

intubation on end-expiratory lung level and regional ventilation distribution in cardiac children.

Paediatr Anaesth 2011;21:887-893.

71. Hough JL, Johnston L, Brauer S, Woodgate P, Schibler A. Effect of body position on

ventilation distribution in ventilated preterm infants. Pediatr Crit Care Med 2013;14:171-177.

72. Bhatia R, Schmolzer GM, Davis PG, Tingay DG. Electrical impedance tomography can

rapidly detect small pneumothoraces in surfactant-depleted piglets. Intensive Care Med

2012;38:308-315.

73. Gomez-Laberge C, Wolf GK. Toward individualized treatment of acute respiratory distress

syndrome: Response-guided mechanical ventilation. Pediatr Crit Care Med 2011;12:690-692.

74. Lachmann B. Open up the lung and keep the lung open. Intensive Care Med

1992;18:319-321.

75. Tingay DG, Polglase GR, Bhatia R, Berry CA, Kopotic RJ, Kopotic CP, Song Y, Szyld E,

Jobe AH, Pillow JJ. Pressure-limited sustained inflation vs. gradual tidal inflations for

resuscitation in preterm lambs. J Appl Physiol 2015;118:890-897.

76. Rossi Fde S, Yagui AC, Haddad LB, Deutsch AD, Rebello CM. Electrical impedance

tomography to evaluate air distribution prior to extubation in very-low-birth-weight infants: A

feasibility study. Clinics (Sao Paulo) 2013;68:345-350.

37

77. Frerichs I. Bedside lung imaging methods (electrical impedance tomography). In:



78. Miedema M, Frerichs I, de Jongh FH, van Veenendaal MB, van Kaam AH. Pneumothorax

in a preterm infant monitored by electrical impedance tomography: A case report. Neonatology

2011;99:10-13.

79. van der Burg PS, Miedema M, de Jongh FH, van Kaam AH. Unilateral atelectasis in a

preterm infant monitored with electrical impedance tomography: A case report. Eur J Pediatr

2014;173:1715-1717.

80. Costa EL, Borges JB, Melo A, Suarez-Sipmann F, Toufen C, Jr., Bohm SH, Amato MB.

Bedside estimation of recruitable alveolar collapse and hyperdistension by electrical impedance

tomography. Intensive Care Med 2009;35:1132-1137.








Bartłomiej Grychtol, Marc Bodenstein, Hervé Gagnon, Stephan H. Böhm, Eckhard Teschner, Ola

Stenqvist, Tommaso Mauri, Vinicius Torsani, Luigi Camporota, Andreas Schibler, Gerhard K.


ONLINE SUPPLEMENT 9

Recommendations for future direction in EIT development,

evaluation and use

2

Recommendations for future direction of EIT development, evaluation and use

Introduction

The role of EIT in clinical practice, especially the critical care environment, has yet to be

systematically established. In the process of clarifying EIT´s role, consensus statements are

clearly needed. This section will summarize the goals and priorities of future research and

device development required to create practical and clinically useful EIT systems. Through this

the role of EIT in different clinical settings can be determined and then evaluated to ascertain

whether benefits to outcome are possible. This initially requires consideration of the most

important unmet clinical needs and gaps in scientific and clinical knowledge. Through such an

approach EIT is likely to become widely adopted.

Introducing new technologies into clinical practice brings new challenges for the clinician,

even when evidence of benefit exists. The success of EIT as a clinical tool will require

educational resources to support clinicians. In the experience of the author group, the use of

functional imaging is conceptually novel for clinicians. As evident by this review, there is now a

broad clinical and engineering expertise in EIT, especially in critical care settings. A series of

internet-accessible training tools, such as videos, demonstrating the correct application of EIT,

problem-solving common problems (such as electrode contact issues), common clinical

scenarios and how to interpret the meaning of different EIT images would greatly enhance

initial clinical uptake.

Prior to clinical trial evaluation, patient-specific electrode interfaces and image

reconstruction algorithms are urgently needed (1, 2). These systems should be developed

through collaboration between clinicians and medical equipment manufacturers so acceptable

age-specific solutions are found. The importance of chest shape in accurate image

reconstruction has been detailed earlier (1-3). Collaborations aiming to share data sets of chest

shapes by gender, age and developmental status will aid in this. Ideally appropriate chest

algorithms should be automatically chosen for the patient´s examination. The growing market

in pediatric and neonatal critical care, and the increasing obesity in adults, justifies the

commercial potential in these populations, particularly as non-invasive respiratory support

becomes more popular. Electrode interfaces and display of information will also need to be

individualized to diverse populations and needs.

3

Potential unmet clinical needs

Recent shifts in the direction of modern respiratory care offer unique opportunities for new

validated and practical tools. There is increasing clinician-driven demand for advanced point-of-

care monitoring of physiological status. Unfortunately, many of the accepted existing bedside

lung function tools, such as lung mechanics derived from airway flow recordings, require the

use of an endotracheal tube. The increased shift towards non-invasive respiratory support in

critical care, such as CPAP and high-flow nasal cannulae, has come at the expense of fewer

reliable options for direct monitoring of lung function (4). In this environment of dichotomous

clinical needs EIT is an attractive new technology. Similarly concerns regarding radiation

exposure mainly in childhood limit the use of many intermittent imaging tools that have found

practical value in the adult population, such as computed tomography (CT).

Clinical EIT: Towards a monitoring, diagnostic or therapeutic tool?

Improved computing and engineering capacity offers the potential for EIT to be more than a

simple monitor of clinical status, either as a diagnostic tool or as part of a therapeutic

intervention via feedback control of clinical care, such as delivery of mechanical respiratory

support. In the first instance, EIT is likely to be more easily adopted as a monitoring tool,

particularly for early identification of simple adverse events and/or in environments where

existing monitoring tools are impractical, such as non-invasive respiratory support. This

approach will allow evaluation of the ability of EIT to alter existing high-risk practices and/or

need for static investigations (e.g. chest radiography) without compromising safety and

outcomes. The ability of EIT to correctly identify endotracheal tube position (5-7), and then

ultimately eliminate the routine use of chest radiography in this indication, or the use of EIT to

guide need for and response to endotracheal tube suction (8-10) are two areas in which EIT

could be easily assessed.

More pressing, and clinically meaningful, is the use of EIT as a means of identification

and localization of air leaks. Air leaks are a relatively frequent complication of intensive care,

especially in preterm infants and post surgery, and associated with important short and long-

term morbidity, despite being simple to rectify if diagnosed. Currently, chest radiography or

ultrasound is usually used to diagnose air leaks but clinical suspicion rarely occurs before clinical

deterioration. Animal models have identified the potential for EIT to diagnose, and localize, air

leaks before clinical deterioration (11-13). Clinical use has been documented in case reports

(14). Simple prospective observational studies of EIT to identify air leak, in conjunction with

current gold standard clinical care, would be highly informative, as well as determining the

4

diagnostic potential of EIT. Such studies would require large sample sizes and would only be

possible through multicentred collaboration. Studies assessing the role of EIT as a monitoring

tool are relatively easy to design, can be powered to short term outcomes and can be

performed across different age groups. Focusing on the simpler role of EIT as a monitor would

also allow clinicians to become broadly familiar with the new technology. From these

experiences it is likely that the development of EIT systems that can be used to guide clinical

practices, especially in the more meaningful but harder therapeutic setting, can occur.

Research investigating the role of EIT as a diagnostic tool needs to consider the context

of existing tools: Will EIT be used to identify conditions or clinical states in which diagnostic

tools are lacking or not wide-spread, for example bedside assessment of ventilation-perfusion

mismatch, or as an alternative to existing tools, for example chest radiography for air leaks? It

should be remembered that EIT simply measures changes in relative volume states within the

chest, and thus clinical interpretation will be essential. For example a sudden decrease in tidal

ventilation in one lung region may indicate localized collapse, secretion plugging, overinflation

or evolving air leak, all diagnoses with different clinical meaning. New EIT algorithms, such as

regional time constant and atelectasis parameters (15, 16), may allow more direct diagnostic

potential. At this stage the use of EIT as a diagnostic tool is unconfirmed, future research must

first determine the conditions in which EIT may offer diagnostic potential and then

systematically evaluate the validity, reliability, reproducibility, sensitivity and specificity relative

to the risk, limitations and cost of current ‘gold’ standard tools. We would recommend that this

be initially addressed through clinical studies limited to consortia of expert EIT users, allowing

evidence and experience-based consensus guidelines.

The ultimate goal of any new physiological tool must be to demonstrate improvements

in outcomes if progress beyond a research tool can be expected. Ideally EIT would find a role

as a tool that drives direct clinical care. The promising observational adult and preclinical data in

the use of EIT to guide positive end-expiratory pressure (PEEP) settings during adult and

pediatric ARDS (17, 18), and newborn resuscitation at birth (19), provides an initial focus for

clinical research. These could be addressed using a two-step clinical trial approach. In the first

instance, the potential of EIT to guide assisted respiratory support could be assessed with

short-term practical outcomes targeting utility, feasibility and safety. Once these aspects have

been determined the role of using EIT to guide clinical care requires assessment in large

randomized trials that target meaningful long-term outcomes, such as disability-free survival.

The use of EIT in adult ARDS offers considerable appeal. The disease is characterized by

gravitational heterogeneities that can change rapidly, and respiratory support that aims to

5

create uniformity of ventilation, through positioning, ventilator settings and minimizing

interstitial edema. The existing data suggests that EIT may provide a solution to all of these

needs and the functional and dynamic nature of its imaging overcomes some of the limitations

of static measures (18). The well-established international ARDS consortia offer a network for

the multicenter randomized control trial (RCT) evaluation of EIT as a tool for guiding the

effectiveness of specific ARDS therapies and, also, monitoring disease progression compared to

existing tools, such as oxygenation, lung compliance and chest radiography/CT. It is likely these

trials could be performed quicker and with more meaningful power than in the pediatric

population.

Severe pediatric ARDS, although not a common condition would be a reasonable clinical

entity to initially evaluate the role of EIT in guiding advanced ventilator settings in children,

specifically PEEP (18). At present there is no practical realtime feedback system for determining

optimal PEEP in regular clinical use. Pediatric ARDS is a lung condition in which the PEEP setting

is important, being associated with regional volume heterogeneity and often responsive to open

lung recruitment. Patients are also often managed with a cuffed endotracheal tube and muscle

relaxants (unlike neonatal acute lung disease), thus offering less variables during PEEP titration.

To achieve important clinical practice change, ideally, EIT-guided PEEP titration during active

disease and weaning should demonstrate improvements in long-term disease free survival as

well as short-term outcomes such as reduced ventilator days. This would require a large and

lengthy multicenter RCT. Such RCT are not uncommon in pediatric/neonatal critical care, but

until EIT has been adopted more broadly as a clinical tool, initially to monitor lung states, it will

not be feasible or likely to achieve intended goals.

Neonatal resuscitation in the delivery room is an area in which EIT may have therapeutic

utility. Increasingly, neonatologists are aware that lung protective ventilation needs to be

applied from as soon as possible after birth, but respiratory support is often applied with non-

invasive methods (for example bag and mask ventilation) and feasible and reliable monitoring is

lacking (20). Also, complicated solutions (for example PEEP titration) are not needed to achieve

meaningful clinical change, as the biggest limitation to respiratory support at birth is simply

adequately and repeatedly achieving tidal ventilation. In this regard EIT is ideal and could easily

be implemented without advanced end-user training. The most recent neonatal International

Liaison Committee for Resuscitation consensus statement has called for clinical trials of lung

function monitoring tools at birth (21). A simple belt-style EIT interface offering realtime direct

visualization of low resolution breath-to-breath ventilation could guide the clinician in important

aspects of care, such as mask seal and pressure settings (22). It is arguable that focusing on

improving these basic aspects of neonatal resuscitation is more meaningful than targeting

6

advanced aspects of respiratory care at birth, such as tidal volume targets and PEEP levels, all

of which assume that former is optimal.

The rapid through-put and general stability of the patient population compared to critical

care results in less investigation of respiratory function during anaesthesia for surgery. The

need for EIT use in these patients is probably less relevant than in intensive care. For example,

ETT position is rarely confirmed during a routine anesthesia. Nonetheless, simple to apply EIT

systems that do not interfere with other monitoring and therapeutic systems especially

diathermy would be useful in guiding respiratory care in theatre. In the first instance large

observational trials evaluating the utility as a monitoring tool for ETT position and atelectasis

and ventilation inhomogeneity would identify whether a potential exists, and help guide

appropriate clinical questions for outcome and safety based evaluation.

To date the majority of EIT research has focused on the inpatient acute care population.

This has excluded an important unmet potential of EIT as a pulmonary function monitor in

ambulatory settings. The ability of EIT to measure regional ventilation and aeration without an

interface at the airway opening, suggests that it could be used outside of pulmonary function

laboratories without expertise training. If simple electrode belts were available EIT could even

be used in home or outreach environments. In the first instance EIT could be used in

conjunction with existing pulmonary function testing to determine reliability and utility as an

early detection and temporal disease progression monitoring tool, especially for chronic

interstitial and obstructive lung disease in adults (23-25), congenital diseases like cystic fibrosis

in children and adults (26) and developmental disorders related to complications of birth, such

as bronchopulmonary dysplasia.

Conclusions

EIT offers the potential to find utility within a diverse range of clinical settings. To achieve

meaningful clinical use a structured multidisciplinary approach is needed. Initially improved

patient interfaces and population-specific image reconstruction algorithms are needed. Then

integration into the clinical settings should initially aim to use EIT as an adverse events monitor.

This should be paired with well-designed online training tools highlighting both good application

and also pertinent EIT image patterns and examples of different clinical states, uses and

diseases. This will familiarize clinicians to the technology, enhance decision-making and

empower clinicians with the skills needed to generate clinically relevant questions for

interventional therapies. Ultimately large trials of EIT-guided intervention powered to long-term

outcomes will be needed.

7


The first draft of this online document was prepared by D. Tingay with collaboration of

I. Frerichs and A. Adler. It was reviewed and approved by all other authors.

8

References

1. Adler A, Arnold JH, Bayford R, Borsic A, Brown B, Dixon P, Faes TJ, Frerichs I, Gagnon H,

Garber Y, Grychtol B, Hahn G, Lionheart WR, Malik A, Patterson RP, Stocks J, Tizzard A,

Weiler N, Wolf GK. GREIT: A unified approach to 2D linear EIT reconstruction of lung images.

Physiol Meas 2009;30:S35-55.

2. Grychtol B, Lionheart WR, Bodenstein M, Wolf GK, Adler A. Impact of model shape

mismatch on reconstruction quality in electrical impedance tomography. IEEE Trans Med

Imaging 2012;31:1754-1760.

3. Leonhardt S, Lachmann B. Electrical impedance tomography: The holy grail of ventilation

and perfusion monitoring? Intensive Care Med 2012;38:1917-1929.

4. Schibler A, Frey U. Role of lung function testing in the management of mechanically

ventilated infants. Arch Dis Child Fetal Neonatal Ed 2002;87:F7-F10.

5. Schmolzer GM, Bhatia R, Davis PG, Tingay DG. A comparison of different bedside

techniques to determine endotracheal tube position in a neonatal piglet model. Pediatr

Pulmonol 2013;48:138-145.


electrical impedance tomography in detecting regional tidal volumes during one-lung ventilation.

Acta Anaesthesiol Scand 2008;52:1131-1139.




8. Tingay DG, Copnell B, Grant CA, Dargaville PA, Dunster KR, Schibler A. The effect of

endotracheal suction on regional tidal ventilation and end-expiratory lung volume. Intensive

Care Med 2010;36:888-896.

9. Hough JL, Shearman AD, Liley H, Grant CA, Schibler A. Lung recruitment and endotracheal

suction in ventilated preterm infants measured with electrical impedance tomography. J

Paediatr Child Health 2014;50:884-889.

10. Heinze H, Eichler W, Karsten J, Sedemund-Adib B, Heringlake M, Meier T. Functional

residual capacity-guided alveolar recruitment strategy after endotracheal suctioning in cardiac

surgery patients. Crit Care Med 2011;39:1042-1049.

11. Bhatia R, Schmolzer GM, Davis PG, Tingay DG. Electrical impedance tomography can

rapidly detect small pneumothoraces in surfactant-depleted piglets. Intensive Care Med

2012;38:308-315.

12. Costa EL, Chaves CN, Gomes S, Beraldo MA, Volpe MS, Tucci MR, Schettino IA, Bohm

SH, Carvalho CR, Tanaka H, Lima RG, Amato MB. Real-time detection of pneumothorax using

electrical impedance tomography. Crit Care Med 2008;36:1230-1238.

9

13. Preis C, Luepschen H, Leonhardt S, Gommers D. Experimental case report: Development

of a pneumothorax monitored by electrical impedance tomography. Clin Physiol Funct Imaging

2009;29:159-162.

14. Miedema M, Frerichs I, de Jongh FH, van Veenendaal MB, van Kaam AH. Pneumothorax

in a preterm infant monitored by electrical impedance tomography: A case report. Neonatology

2011;99:10-13.

15. Dargaville PA, Rimensberger PC, Frerichs I. Regional tidal ventilation and compliance

during a stepwise vital capacity manoeuvre. Intensive Care Med 2010;36:1953-1961.




17. Gomez-Laberge C, Rettig JS, Smallwood CD, Boyd TK, Arnold JH, Wolf GK. Interaction

of dependent and non-dependent regions of the acutely injured lung during a stepwise

recruitment manoeuvre. Physiol Meas 2013;34:163-177.

18. Wolf GK, Gomez-Laberge C, Rettig JS, Vargas SO, Smallwood CD, Prabhu SP, Vitali SH,

Zurakowski D, Arnold JH. Mechanical ventilation guided by electrical impedance tomography in

experimental acute lung injury. Crit Care Med 2013;41:1296-1304.

19. Tingay DG, Lavizzari A, Zonneveld CE, Rajapaksa A, Zannin E, Perkins E, Black D,

Sourial M, Dellaca RL, Mosca F, Adler A, Grychtol B, Frerichs I, Davis PG. An individualized

approach to sustained inflation duration at birth improves outcomes in newborn preterm lambs.

Am J Physiol Lung Cell Mol Physiol 2015;309:L1138-1149.

20. Schmolzer GM, Kamlin OC, Dawson JA, te Pas AB, Morley CJ, Davis PG. Respiratory

monitoring of neonatal resuscitation. Arch Dis Child Fetal Neonatal Ed 2010;95:F295-303.

21. Perlman JM, Wyllie J, Kattwinkel J, Wyckoff MH, Aziz K, Guinsburg R, Kim HS, Liley HG,

Mildenhall L, Simon WM, Szyld E, Tamura M, Velaphi S, Neonatal Resuscitation Chapter C.

Part 7: Neonatal resuscitation: 2015 international consensus on cardiopulmonary resuscitation

and emergency cardiovascular care science with treatment recommendations. Circulation

2015;132:S204-241.

22. Schmolzer GM, Kamlin OC, O'Donnell CP, Dawson JA, Morley CJ, Davis PG.

Assessment of tidal volume and gas leak during mask ventilation of preterm infants in the

delivery room. Arch Dis Child Fetal Neonatal Ed 2010;95:F393-397.

23. Mauri T, Bellani G, Salerno D, Mantegazza F, Pesenti A. Regional distribution of air

trapping in chronic obstructive pulmonary disease. Am J Respir Crit Care Med 2013;188:1466-

1467.

24. Vogt B, Zhao Z, Zabel P, Weiler N, Frerichs I. Regional lung response to bronchodilator

reversibility testing determined by electrical impedance tomography in chronic obstructive

10

pulmonary disease. Am J Physiol Lung Cell Mol Physiol 2016; DOI: 10.1152/

ajplung.00463.2015.




26. Zhao Z, Muller-Lisse U, Frerichs I, Fischer R, Moller K. Regional airway obstruction in

cystic fibrosis determined by electrical impedance tomography in comparison with high


State of the art review Chest electrical impedance ...

Documents

Transcript of State of the art review Chest electrical impedance ...