Controlling Legged Robots with Coupled Artificial Biochemical Networks
Artificial intelligence and robots in the OR - Still room ... 19_Neumuth_Künstliche...
35
Artificial intelligence and robots in the OR - Still room for ebM? Thomas Neumuth
Transcript of Artificial intelligence and robots in the OR - Still room ... 19_Neumuth_Künstliche...
Artificial intelligence and robots in the OR- Still room for ebM?
Thomas Neumuth
Innovation Center Computer Assisted Surgery (ICCAS)
• Affiliated to Medical School of Universität Leipzig • ~85 engineering and computer science researchers
Medicine Biomedical and informationtechnology
PersonalizedPredictivePreventiveParticatory
Hood and Friend 2011
4P m
edic
ine
Technology is integral part of medicine
Technology has a different evolution speed
• Information retrieval from distant monitors and devices• Vendor dependent device operation• Technical obstacles for advanced workflow management • Limited flexibility in the selection and combination of systems
4
[Quelle: Bulitta, 2006]
5
acquire
analyze
act
6
acquire
analyze
act
Robotic systems for surgery
• Master-slave systems
7
• (Semi-) autonomous systems
© Intuitive, Transenterix, Avatera, Stryker, Smith & Nephews
Current evidence for robotic systems
• Operation times• Operative costs
• Blood loss• Complication rates• Hospital stay length• Transfusion rate
8
FDA, Feb 28, 2019
Imaging and cooperative robots
9
© Siemens Artis Zeego
Imaging and cooperative robots
10
Bieck et al. 2017
11
acquire
analyze
act
Current research areas
• Anaesthesia management• Risk for postoperative pain, nausea, and respiratory depression• Predict hypotension, hypoxemia
• Surgery• Prediction of procedure times and anticipation of risk situations• Decrease human-technology interaction, Automatic documentation
• Administration• Decrease non-value-adding activities• Provide KPIs
13
Situation aware biomedical technologies
14Neumuth et al. 2009, Franke et al. 2015
Situation aware biomedical technologies
15Neumuth 2012
SDC – Service oriented device connectivity
Navigation
© Aesculap
HIS, PACS
© Ch. Toulon
Dokumentation
© Oehm & Rehbein
3D-Rekonstruktion
© Medstation
© OrthoMit
US-Regis-trierung
Planung
© Aesculap © SurgiTAIX
Tools Displays MMI
© Philips© Maquet
OP-Tisch
© Trumpf
Tracking
© NDI
Roboter
© meditec HIA
SDCOne Language
ISO/IEEE 11073-10207IEEE 11073-20701
ISO/IEEE 11073-20702Medical device interoperability
17
acquire
analyze
act
Hyperspectral Imaging
18
Hyperspectral imagung for tissue perfusion
19
Pathological bowel
Perfusion state ?
anastomosis
Perfusion state ?
a) Patient with postop. perfusion insufficiency
b) and c) Patients without postop. perfusion insufficiency
Jansen-Winkeln et al. Chirurg 2018
© Universitätsklinikum Leipzig AöR (2009): Thema, Autor
Total number of spectra: nthyroid = 3128, nparathyroid = 383
HSI-based discrimination of thyroid and parathyroid
Barberio et al. 2018
But surgery is just one option …
21
acquire
analyze
act
Digital patient models for decision-making
22
ENT-Tumorboard at the University Hospital Leipzig, Germany Cypko et al. 2017
Prediction and clinical decision support
Artificial intelligence and robots in the OR- Still room for ebM?
The digital footprint
25
acquire
analyze
act
The digital footprint
26
acquire
analyze
act
Numeric vs. symbolic process models
27
SurgicalWorkflow
Numericalmodels
(digital footprint)
Symbolicmodels
Model generalization
T. Neumuth: Process modeling strategies
…
A B A C
A B C
A B C DD
START
END
A
B
C
Single cases Merged model
Model analysis: workflow assessment
29
END
paracentesisparacentesis knife
cornea
injecthealon
chamber ant
capsulorhexisrhexis cannulacapsula lentis
cutlancet clear cut
cornea
excision materialutrata‘s tweezers
capsula lentis
100% 93% 81%
92%
94%
10%
86%
START
Preparation Capsulorhexis Lens Removal LensImplantation
Removal of Healon Completion
(model filtered with threshold 5%)
Neumuth et al. 2011
Model analysis: workflow assessment
30
END
paracentesisparacentesis knife
cornea
injecthealon
chamber ant
capsulorhexisrhexis cannulacapsula lentis
cutlancet clear cut
cornea
excision materialutrata‘s tweezers
capsula lentis
injectvision blue
chamber ant
irrigatesauter cannulachamber ant
100%
100%
19%
93%
51%
81%
83%
5%
84%
99%
94%
12%
18%
34%
6%
92%
94%
10%
86%
START
Phase Capsulorhexis or Cataract interventions: Generic process models
for outpatient (n=51) and inpatient cases (n=54)
(model filtered with threshold 5%)
Preparation Capsulorhexis Lens Removal LensImplantation
Removal of Healon Completion
Neumuth et al. 2011
Model analysis: resource impact estimation
Nissen fundoplication, animal studylaparoscopic cases vs. telemanipulator-assisted cases (each n=12)
31Krauss et al. 2012
Model analysis: technology requirements
Opening ofthe skin
Preparation of muscle tissue
Bone removal Removal of intervertebral disc
Completion
Neumuth et al. 2009 32
n=14 Unit AVG Min Max SDEV
Time Frame hh:mm:ss 00:45:08 00:22:23 01:13:43 00:13:08
Iterations activity 15.2 7 27 6.3
Cumulated Duration hh:mm:ss 00:19:55 00:12:40 00:32:11 00:06:52
Objective: Development of an eco system for data acquisition and AI-based analysis for post market surveillance of medical devices
33
Summary
• Robots have arrived in ORs• AI in the OR has not yet arrived
• Technologies have digital footprints that enable continuous ebMwithout cognitive biases
If you can't measure it, you can't improve it.(Peter F. Drucker)
34
35
Prof. Dr. Thomas NeumuthInnovation Center Computer Assisted Surgery (ICCAS)[email protected], Twitter: @iccasLeipzig
CCL Advisory Board Meeting, Mar 12 2019
Thank you.
