THE ANATOMICAL RECORD 292:2013–2022 (2009)

Laser-Induced AutofluorescenceMeasurements on Brain Tissues

ALEXANDRU PASCU,1 MIHAELA OANA ROMANITAN,2

JOSE-MARIA. DELGADO,3 LEON DANAILA,4 AND MIHAIL–LUCIAN PASCU1*1Department of Laser, National Institute for Laser, Plasma and Radiation Physics,

Bucharest, Romania2Neurology Clinics, Emergency University Hospital, Bucharest, Romania3Division de Neurociencias, Universidad Pablo de Olavide, Sevilla, Spain

4Neurosurgery Clinics, National Institute of Neurology and Neurovascular Diseases,Bucharest, Romania

ABSTRACTIt was demonstrated that comparison of the autofluorescence spectra

induced with laser radiation in ultraviolet and visible allows the identifica-tion of brain tumor tissues and normal tissues as well as the differencebetween them. The measurements were performed on homogenates toensure an optimal reproducibility of the results. We conclude that the auto-fluorescence spectra of the tumor samples are close to those measured fornormal tissues, but there are differences between them that allow distin-guishing the tumor from the normal tissue. One difference is that for eachpair of tumor/normal tissue samples, the peak autofluorescence for the nor-mal tissue is shifted with respect to that for the tumor–typically between10 and 20 nm; overall autofluorescence intensity is also different for thecomponents of the same pair, the difference being in the range 15%–30%. Aparameter that can also be used is the variation of the ratio of some fluo-rescence intensity peaks between normal and tumor tissue samples. Meas-urements of this parameter yielded variations ranging between 10% and40%. Another conclusion of the study is that in vitro experiments showthat it is mandatory to use pairs of samples (normal/tumor tissue) takenfrom the same patient. The results show that, after further experimentalin vitro tests, the method may be adapted to real-time intraoperative condi-tions by measuring the autofluorescence of the tumor and of the adjacentnormal tissue. Anat Rec, 292:2013–2022, 2009. VVC 2009 Wiley-Liss, Inc.

Keywords: autofluorescence; LIF; biomolecular spectroscopy;neurosurgery

One of the main neurosurgical problem lies in accu-rately identify the margins of brain tumors to allow atumor’s precise excision without destruction of the sur-rounding healthy tissue. Excision is optimal if thetumoral mass is removed from the brain without affect-ing the surrounding healthy tissue or with minimuminjury to it. The result of resection and the histopatho-logical diagnosis of the extracted tumor tissue impose atherapeutic strategy to further treat the disease. Giventhese clinical requirements, it is highly recommended toaccurately identify tumor tissue borders during the sur-gical operation proper, by means of specific methods andtechniques. Identification is done in two steps: (a) a pre-operative imaging of the tumor tissue and borders using

CT, MRI, and/or ultrasound-based equipment; and (b) anintraoperative stage, in which the tumor borders areidentified by direct visual inspection; the operating

Grant sponsor: The EC/ESF/COST network B30: NeuralRegeneration and Plasticity: NEREPLAS.

*Correspondence to: Mihail–Lucian Pascu, National Institutefor Laser, Plasma and Radiation Physics, Laser Department,Str. Atomistilor nr. 409, Magurele, Ilfov, 077125 Bucharest,Romania. Fax: þ40-21-457-5739. E-mail: mihai.pascu@inflpr.ro

Received 18 February 2009; Accepted 28 July 2009

DOI 10.1002/ar.21034Published online in Wiley InterScience (www.interscience.wiley.com).

VVC 2009 WILEY-LISS, INC.

microscope, endoscopic techniques, and autofluorescencemeasurements of the tumor and healthy tissues aremethods also available.

Laser-induced autofluorescence is one of the maincandidates for use in the operative field to identify tu-mor borders in both benign and malignant cases. It mayallow, in principle, accurate identification of the inter-face between normal brain tissue and tumor tissue bymeasuring the optical fluorescence spectra emitted bythese tissues after excitation with laser optical beamshaving suitable characteristics. Literature reports showthat the autofluorescence method is promising for delin-eating brain tumor resection margins (Bottirolli et al.,1998; Lin et al., 2000; Lin et al., 2001; Croce et al.,2003; Cubillos et al., 2006). At the same time, moreresearch needs to be devoted to developing the instru-ments, procedures, and clinical recommendations forintraoperative use of autofluorescence, at the level ofboth in vivo and in vitro measurements (Kremer et al.,2009).

This paper is conceived as a synthesis of character-istics of the laser-induced autofluorescence methodand technique applied in neurosurgery for diagnosticsand/or intraoperative procedures. Relevant researchcontributions of the authors in this field are alsodescribed.

MATERIALS AND METHODSAutofluorescence Definition: Basic Data

The autofluorescence of a tissue, particularly brain tis-sue, is the fluorescence, which emits without having anyexogenous fluorescent substance added to it; autofluores-cence radiation is emitted by the tissue when an opticalradiation of suitable wavelength falls on it. Conse-quently, autofluorescence spectra give accurate informa-tion about the content and molecular structure of theemitting tissue; particular care has to be taken to avoidinterferences from other exogenous substances. Theautofluorescence is emitted by molecules that exist nor-mally in the tissues, following their resonant interactionwith optical radiation.

The scheme that describes the fluorescence emissionand the fluorescence lifetime of a molecule (usuallynamed fluorophore) is shown elsewhere (Danaila andPascu, 2001; Valeur, 2001). As a rule, autofluorescence isemitted after one photon of radiation is absorbed in thenear-ultraviolet or in the shorter-wavelengths range ofthe visible. Because the fluorophores are surrounded byother tissue components, which normally (but not only)act as a liquid environment of the molecules, the absorp-tion and fluorescence emission spectra are broad, andtheir structures are also broadened. Consequently, theautofluorescence spectra show a partial superimpositionof the individual emitted bands, although the spectralcharacteristics remain specific to the emitting molecules.The pH of the tissue in which the emitting moleculesare embedded may also affect autofluorescence charac-teristics, such as the spectral width, shape, etc. Somemolecules found in the tissue along with the emittingfluorophores may reabsorb the emitted fluorescence,which leads to quenching of the autofluorescence, andthereby to errors in estimating the emitter positions andconcentrations.

Autofluorescence Experimental Monitoring

The experimental system for measuring autofluores-cence signals should be conceived taking into accountthat fluorescence excitation is always produced at shorterwavelengths than is the fluorescence radiation. At longerwavelengths, autophosphorescence is emitted, which hasa lifetime longer than that of autofluorescence. The spec-tral distribution of the autofluorescence emission is muchbroader and more structured than the pumping radiation,its duration is usually longer, and its polarization state isdifferent; the coherence state of the excitation beam is dif-ferent from that of the autofluorescence radiation. Exper-imentally, it is possible to excite autofluorescence using alaser beam that is monochromatic, coherent, and highlypolarized. The autofluorescence radiation obtained isincoherent, broad–band, and nonpolarized. Autofluores-cence measures exclusively the properties of the tissues,because it is produced only by their endogenous mole-cules. It may provide two main types of information(Sokolov et al., 2002): (a) image–it allows obtaining amore or less extended image of a part of the studied tis-sue; and (b) fluorescence radiation–it is excited by focus-ing the pumping beam on the tissue. The fluorescenceradiation has a spectral distribution and a specific inten-sity; it also has a polarization state, and is characterizedby specific ratios between the spectrum peaks.

Autofluorescence is highly sensitive to endogenousemitters of the tissue, if properly excited. It allows iden-tification of the molecular components of a tissue andtheir combinations, and investigation of the interactionbetween them and the surrounding media (usuallydescribed by the pH of the tissue); it also allows monitor-ing of the modifications produced in the molecular con-stitution of a tissue by the effect of natural and/orartificial factors. The autofluorescence emission of a fluo-rophore may be characterized by the following proper-ties: overall intensity of the emitted spectrum; emissionspectral range; spectral structure of the radiationdescribed by the intensity peaks and their relative ratiosand by the wavelength of the main intensity peak (ifsuch peak exists); fluorescence lifetime, measured eitherby its full time width (FTW) or by the full time width atthe half maximum of the fluorescence peak (FTWHM), ifonly one peak is involved (Yong et al., 2006).

The correct measurement of the lifetime requiresexciting the autofluorescence with a laser pulse that isshort enough; it should have an FTWHM of 5 ns, atmost. This parameter may be measured for each fluores-cence peak if the autofluorescence spectrum exhibitsmore peaks.

Human brain tissues contain endogenous fluorophoresas part of the following classes of biomolecules: aminoa-cids (tryptophan, tyrosine, phenylalanine); structuralproteins (collagen, elastin); enzymes and co-enzymes (fla-vin adenine dinucleotide—FAD, the reduced form of thenicotinamide adenine dinucleotiode-NADH, flavins); lip-ids (phospholipids, lipofuscin, ceroids); porphyrins; andvitamins (A, K, and D). The autofluorescence emitted bythe tissue contains signals originating from the activefluorophores and depends on their concentration andspatial distribution, as well as the properties of the sur-rounding environment, such as pH, optical properties,structure and organization, homogeneity, (an)isotropy,and turbidity.

2014 PASCU ET AL.

Autofluorescence is, therefore, sensitive to the above-mentioned properties, and may signal small modifica-tions of any of them. Moreover, tissue histology and bio-chemistry leave their signatures on the autofluorescencespectra. Consequently, variations in the health state ofthe tissue that modifies its histology and/or histochemi-cal properties can be evidenced by autofluorescencemeasurements. Such is the case of tumor/malignant tu-mor tissues that have properties that differ from thoseof a normal tissue; autofluorescence may become amethod to detect and investigate malignant tumor tis-sues. In neurosurgery, it may assist the neurosurgeon tofind, in real time, the borders of the malignant tumors.To do this, specialized equipment should be developed todetect the malignant tumors and/or their limits in thebrain for early stages or more-developed tumor tissues.This article brings together the autofluorescence meas-urements made by the authors in brain tissues.

Sample Preparation

First type of sample preparation. This approachwas exploratory; it was chosen initially to find the bestprocedures for sample processing and mounting in theoptical system. The conclusions from the measurementswere used to identify the conditions for reproducibilityin both sample preparation and optical arrangement forautofluorescence excitation and collection.

Accordingly, a protocol to conserve, transport, andtreat tumor tissues was established, to ensure unalteredsamples. Once extracted using standard operative proce-dures, the brain tissue samples were kept immersed innatural saline at 20�C in a Dewar container. The timeinterval between tumor extraction and measurementswas, at most, 60 min. The samples were irrigated fromtime to time with natural saline to prevent uncontrolleddrying and to remove blood traces. The working temper-ature was 20�C. To ensure that the conditions for repro-ducibility of the measurements were the best possible,and prevent errors in the fluorescence excitation and col-lection, the samples were placed in direct contact withan optical quartz plate of 0.5 mm in thickness; eachsample surface that was exposed to the pumping radia-tion and from which the fluorescence was emitted wasvery well defined. The exposed surface was 1–2 mm indiameter (the diameter of the excitation beam), and foreach sample 5 such spots/discs were used for measure-ments; the geometrical arrangement was reproducible,and the results obtained for each excitation wavelengthwere averaged.

Homogenate preparation. The results from thestudies performed on samples processed as in point 1above showed that there are variations in the reproduci-bility of the autofluorescence curves. To avoid this, sam-ples were processed using a different method–namely, ashomogenates: each sample was washed abundantly innatural saline, then ground and further exposed toultra-sound to obtain a homogeneous mixture. The pre-paring of the homogenates did not affect the molecularstructures or concentrations of constituent fluorophoresin the samples. It was done to prevent errors due to theunhomogeneities occurring naturally in the collectedsamples and to the geometry at the interface betweenthe tissue and the optical excitation/collection system.

The optical arrangement for autofluorescence excitation/collection was the same as described in paragraph 1above. For each sample (either tumor or normal tissue)measured, three measuring points/spots were used, andthe three signals were averaged. The autofluorescenceexcitation and detection system was for all sampleswithin the same spectral range (excitation between 337and 500 nm, and emission measurement between 400and 650 nm), so that no corrections related to variationin sensitivity with wavelength were needed from onesample to another.

Experimental Set-Up and Sequence

The main goal of the measurements was to identifythe spectral properties of the autofluorescence of cere-bral tissues for two cases: normal tissues and malignanttumor; by studying these spectra, we aimed to establishcriteria to differentiate between the fluorescence radia-tions emitted in the two cases and to quantify the differ-ences between them. The experiments were made in twosteps: (a) measurement of the spectral characteristics ofthe autofluorescence emitted by the fluorophores exist-ing in the brain tissues; and (b) comparative in vitromeasurements of the spectral properties of the autofluor-escence emitted by pairs of normal/malignant tumortissue.

The experimental set-up (shown in Fig. 1) consisted ofa system adapted to measure laser-induced fluorescence(LIF). In choosing the radiation source used to excite theautofluorescence, several variants previously reported inthe literature were evaluated (Ramanujam, 2000; Linet al., 2002; Wu et al., 2003). For the purpose of this ar-ticle, a pulsed laser with emission in the near-ultravioletand visible spectral ranges was developed. This has twoadvantages: (a) it generates short time-width laserpulses of relatively low energy, which produce a weakperturbation of the studied tissues and a higher overallefficiency of the fluorescence emission; and (b) it enablesmeasurement not only of the spectral distribution of theautofluorescence radiation but also of the fluorescenceemission lifetime.

The laser source is a frequency-doubled tunable dyelaser pumped by a nitrogen pulsed laser (NPL),described elsewhere (Danaila and Pascu, 1999; Pascu,2000; Danaila and Pascu, 2001). The main characteris-tics of the laser system unit used to pump the fluores-cence are (a) NPL: emission at 337.1 nm, laser beambandwidth <0.1 nm, FTW of the pulse 1 ns, energy/pulse50 lJ, peak power per pulse 500 kW, pulse repetitionrate continuously adjustable between 1 and 10 pulsesper second (pps), beam divergence 5 mrad � 10 mrad;and (b) tunable dye laser: NPL pumped, emits tunableradiation between 350 and 700 nm by using a successionof dyes, beam spectral bandwidth 0.2 nm maximum,pulse FTW 1 ns, energy per pulse typically 5 lJ, pulserepetition rate driven by the NPL, and laser beam diver-gence 5 mrad. The dye laser beam focusing produces anirradiance of 108 W/m2 on the sample. The frequency-doubling crystal used to cover the wavelength gap in thenear-ultraviolet up to 337.1 nm is an ADP crystal emit-ting a beam with the following characteristics: tunabilityrange 260–300 nm, spectral bandwidth 0.05 nm whenthe pumping dye laser beam has 0.1 nm full spectralbandwidth, pulse FTW � 1 ns, pulse repetition rate as

LASER-INDUCED BRAIN TISSUES AUTOFLUORESCENCE 2015

for the NPL, peak energy per pulse typically 0.5 lJ,which enables obtaining a peak irradiance of 5 � 106 W/m2 at the focus.

The system to excite and collect the autofluorescencesignal (Fig. 1) consisted of a pair of optical fibers, one ofwhich is used to excite the autofluorescence; it transmitsthe excitation beam with losses lower than 1%. Theexcitation optical fiber is a quartz fiber with a core of1.5 mm in diameter; it has a transmission spectral rangebetween 180 and 700 nm, and a numerical apertureNA ¼ 0.22. The fiber used for the collection of theautofluorescence signal is a fiber bundle of 6 mm indiameter, transmitting between 400 and 700 nm.

The sample is either a solution or a piece of braintissue prepared for in vitro measurements. The auto-fluorescence radiation transmitted by the collectionoptical fiber is focused on the input slit of a monochro-mator; this is part of the spectral analyzing unit,and works between 220 and 800 nm at a linear disper-sion of 0.4 nm/mm; signal detection is made by using afast photomultiplier, sensitive between 180 and 800 nmand able to measure optical signals with rise timesof up to 1 ns. With this system, the spectral distribu-tion and the lifetime of the autofluorescence weremeasured.

The second variant for signal detection was a com-puter-controlled, 2048 pixel CCD camera workingbetween 300 and 800 nm, used only for spectral distribu-tion measurements.

RESULTSFluorophore Measurements

The measured samples were primary endogenous fluo-rophores, such as aminoacids, structural proteins,enzymes and coenzymes, porphyrins, and riboflavins.Literature reports (Zuluaga et al., 1999; Drezek et al.,2001a,b) on the fluorescence properties of such fluoro-phores show sets of data measured under experimentalconditions that differ slightly from one case to another;this prevents the construction of a reference systembased on which the interpretation of the autofluores-cence data measured in brain tissues could be made. Inthis article, we submit data measured in primary fluoro-phores that are potentially present in the brain tissues,aiming to have controllable comparison conditions forthe autofluorescence measurements made in componentsand in tissues samples. Figure 2 shows the fluorescencedistribution for phenylalanine excited at 266 nm. The

Fig. 1. The fluorescence/autofluorescence excitation set-up.

2016 PASCU ET AL.

sample consisted of phosphate-buffered aqueous solu-tions at pH ¼ 7 and a concentration of 10�5 M. The fluo-rescence peaks are broad and have different intensities;they are unstructured and constitute a superimpositionof narrower peaks originating in transitions betweenseveral specific pairs of electronic states of the phenylal-anine molecule.

To obtain an objective basis for the fluorescence spec-tra, we measured LIF for various fluorophores of inter-est. The results are given in Table 1, which shows thefluorescence peak positions, together with the corre-sponding excitation wavelengths.

The fluorescence lifetime of NADH is very short; a p-i-n (PIN), fast photodiode was used (Fig. 1), which wasable to indicate only that the fluorescence lifetime isshorter than 0.5 ns.

The autofluorescence measurements performed onsome of the main molecular components of the brain tis-sues allow the following conclusions: (a) the autofluores-cence spectra of the fluorophores strongly depend on theabsorption characteristics of the fluorophore molecules,which–in turn–depend on the solvent, solution pH, andconcentrations in the samples; this is a rather general

conclusion in LIF but, in this case, it becomes particu-larly critical; (b) the fluorescence excitation and emissionspectra are mainly obtained by excitation with laserradiation in the near-ultraviolet and visible (200–500nm) and are emitted in the visible; (c) each of the fluoro-phore molecules has a specific fluorescence pattern/tracewhich depends on the excitation wavelength. The spec-tral distribution of the fluorescence and the number, rel-ative intensity, and shape of the peaks are stronglydependent on the excitation wavelength. For qualitativeanalysis, the use is recommended of successively morewavelengths to better demonstrate the specific peaks foreach fluorescence source; and (d) if fluorescence peaksare emitted by different fluorophores that are very closeto each other in the spectrum, one possibility to revealthe emitting molecules and to differentiate betweenthem is to measure the fluorescence lifetime, which maydiffer from one fluorophore to another.

Brain Tissue Measurements

The data reported above can be used to reveal altera-tions of normal brain tissues towards tumor tissues (ei-ther benign or malignant). For such cases, the firstconsideration is the interaction between the laser beamand the tissue on which it is falling (Danaila and Pascu,2001). Figure 3 describes the processes of laser beaminteraction with brain tissues and that with tissues ingeneral. At the point of contact with the tissue border,the laser beam (of intensity I0) is reflected in part (IR)and backscattered (Ib.s.); a non-negligible part of it issent back to the source medium without interacting withthe tissue.

The beam penetrating the tissue is refracted (Ir),depending on the tissue’s optical properties, changing itspropagation direction with respect to the incident beam.A part of it is scattered forward (If.s.) the light spreadingaround the incidence point inside the tissue in all direc-tions, depending on the unhomogeneities of the tissue.Another part (Ia) is absorbed by resonant interaction.The rest (Irem) propagates further into the tissue. Theenergy balance is given by I0 ¼ IR þ Ib.s þ Ir þ If.s. þ Iaþ Irem. Of the beams mentioned, Ia is responsible for theexcitation of autofluorescence radiation. At the wave-lengths used in these studies, water and hemoglobin(the two main components of brain tissues) do notFig. 2. The phenyalanine fluorescence emission spectrum.

TABLE 1. Fluorescence peaks of several fluorophores

Compound Solvent/concentrationExcitation wavelength

(nm)/laser usedaFluorescence

wavelength peak (nm)

Collagen 10�5 M phosphate buffer aqueoussolutions at pH ¼ 7

270/SHG of dye laser at 540 nm 395

Collagen 10�5 M phosphate buffer aqueoussolutions at pH ¼ 7

285/SHG of dye laser at 590 nm 310

Collagen 10�5M phosphate buffer aqueoussolutions at pH ¼ 7

337.1/NPL 395

NADH 10�5M phosphate buffer aqueoussolutions at pH ¼ 7

290/SHG of dye3 laser at 580 nm 510

NADH 10�5M phosphate buffer aqueoussolutions at pH ¼ 7

337.1/NPL 455

NADH 10�5M phosphate buffer aqueoussolutions at pH ¼ 7

400/tunable dye laser (POPOP) 510

aSHG, Second Harmonic Generation; POPOP (dye used in the laser optical cavity), 1,4-Di-2-(5-phenyloxazolyl)benzene;Phenyl-Oxazolyl-Phenyl-Oxazolyl-Phenyl; C24H16N2O2.

LASER-INDUCED BRAIN TISSUES AUTOFLUORESCENCE 2017

interfere with the fluorescence emission process. Wateris practically nonabsorbing in the visible; the near-ultra-violet absorption does not lead to fluorescence emissionby water molecules. Hemoglobin does not absorb above600 nm, and the strong absorption below 600 nm doesnot lead to significant fluorescence emission (hemoglobinfluoresces very weakly when excited below 300 nm, andthe fluorescence lifetime is shorter than 25 ps). Eventhough these characteristics were known, brain tissuesamples not having blood within the structure were cho-sen for the study so that absorption of the laser beamdue to hemoglobin in the samples was negligible. Auto-fluorescence of the tissue samples was excited at 337.1nm and at longer wavelengths, up to 500 nm. Thesewavelengths are also recommended for in vitro measure-ments for other reasons: the depth of laser beam pene-tration in the tissue is much lower (one-two orders ofmagnitude) below 337 nm, the radiation at 337.1 nmand above this wavelength is more intense (typically oneorder of magnitude) than in the case of beams emittedbelow 300 nm; in vivo interaction of the brain tissueswith radiation below 300 nm may lead to undesirableeffects on the cells, such as cleavage of molecular bonds(Lin et al., 2001).

Excitation at the mentioned wavelengths includes con-tributions to the autofluorescence spectrum of all the flu-orophores contained in the brain tissues (Drezek et al.,2001a,b; Trujillo et al., 1998).

Since in vitro measurements were made in brain tis-sue samples extracted under anesthesia, the fluores-cence properties of the anesthetics used were considered.The substances were sodium thiopental, pancuronium,fentanyl, suxamethonium, and sevoflurane. Followingexcitation, sodium thiopental, pancuronium, and fen-tanyl do not show fluorescence emission. Suxametho-nium affects the tryptophan side-chain chromophores,and sevoflurane quenches tryptophan fluorescence andreversibly increases NADH fluorescence (Ramanujamet al., 1994; Utzinger et al., 1999). The anesthetics aredistributed in both normal brain tissue and the tumor,so that their contribution to the autofluorescence spec-trum may be detected in both cases. It might be thattheir concentration in the tumor is higher, in which caseLIF signals would be stronger on the tumor side, so thatthe edge between the normal tissue and the tumor couldbe more precisely identified.

The tumor samples used for in vitro measurementswere extracted by neurosurgical operations performedaccording to current medical procedures. The normal tis-sue samples were taken (using standard procedures)from zones around nonruptured aneurisms that hadbeen operated on to prevent accidental ruptures. Themeasurements were made in tissue samples extractedfrom different brain tumors. A set of tumors and theirlocation in the brain as shown in the CT images aregiven in Figs. 4 and 5, which show the brain statusbefore and after tumor extraction.

A typical autofluorescence curve for a normal tissueexcited at 337.1 nm is shown in Fig. 6, measured underthe conditions specified in paragraph 1 of the SamplePreparation section; the emission peak is located at 460–470 nm, and the curve shape shows a possible contribu-tion of NADH, flavin, and porphyrin. Similar autofluor-escence spectra were measured in different samples ofnormal brain tissue, excited at 337.1 nm. The curves areslightly different from each other, which show a relativevariability and a low reproducibility of the autofluores-cence signals from one patient to another. The differen-ces may be due to the different optical properties of thetissues (the relative concentrations of the fluorophoresin them, the pH, and the different homogeneity). Thedetailed studies of a large number of autofluorescencespectra of normal brain tissues excited in the visible andultraviolet allow the conclusion that the autofluores-cence of a brain tissue is not a mathematical sum of thecontributions of the fluorophores existing in its struc-ture, because (a) the autofluorescence peaks are quitebroad, which makes it difficult to accurately estimatetheir wavelengths; (b) of the different wavelengthsto use for autofluorescence excitation, values around340 nm, particularly 337.1 nm, are most recommended

TABLE 2. Fluorescence lifetimes for several brain tissue compounds

Compound

Fluorescencelifetime

(FTWHM) ns Compound

Fluorescencelifetime

(FTWHM) ns

Tryptophan 2.9 Collagen 5.5Tyrosine 4.1 Flavin mononucleotide

(FMN)45

Phenylalanine 6.0 Porphyrin 12NADH <0.5

Fig. 3. The laser beam-tissue coupling processes.

2018 PASCU ET AL.

Fig. 4. Large left intraventricular astrocytoma. A: Preoperative contrast-enhanced coronal magneticresonance imaging scan. B: Postoperative contrast-enhanced magnetic resonance imaging scan.

Fig. 5. Intraventricular astrocytoma. A: Preoperative contrast-enhanced axial CT scan demonstrates aheterogeneously enhancing tumor within the lateral ventricle. The patient underwent craniotomy andresection of this tumor through parieto-occipital approach. B: In the recovery room, the patient was notedto have moderate left hemiparesis and the postoperative CT scan demonstrates resection of the tumor.

LASER-INDUCED BRAIN TISSUES AUTOFLUORESCENCE 2019

due to deeper penetration of the radiation in the tissuesand because at these wavelengths most fluorophores yieldmarked fluorescence signals; and (c) autofluorescencespectra differ, although not dramatically, from one brainzone to another and from one patient to another, evenunder strictly reproducible experimental conditions.

These conclusions suggest that in order to use LIF fordistinguishing between normal brain and (malignant)tumor tissues in real-time neurosurgical operations, ‘‘insitu’’ measurements for normal/tumor tissue pairsshould be made. This could enable a correct evaluationof the tumor boundaries with respect to the normal tis-sues for each specific case.

Autofluorescence of Normal/Tumor BrainTissue Pairs

Measurements were performed in several sample pairs(prepared as mentioned above in paragraph 2 of theSample Preparation section), each pair consisting of tu-mor tissue and normal tissue extracted from the samepatient. The normal tissue samples were prelevated toavoid zones of brain compression by the tumor. In col-lecting the samples, different kinds of tumor were cho-sen, so that the nature of the tumor was not a criterionin selecting the tissues to analyze.

Given the fluorophores present in brain tissues, andtheir autofluorescence properties, three wavelengthswere selected to excite autofluorescence: 337.1, 370, and410 nm.

The measurements performed on the homogenatesshowed that (a) the spectral distribution of the autofluor-escence signals for a given homogenate is very close tothat of the untreated sample and it is reproducible fromone measurement to another; and (b) the autofluores-cence spectra of the tumor samples are close to those ofnormal tissues, but the differences that do exist betweenthem enable identification of the tumor and the normaltissue. The main difference is that for each sample pairthe peak autofluorescence spectrum for the normal tis-sue is shifted with respect to the tumor case. The overall

intensity of the autofluorescence is also different for thecomponents of the pair.

As an example, Fig. 7 shows the autofluorescence of anormal sample excited at 337.1 nm and the spectrumobtained under the same conditions for the tumor sample.Comparison of the two curves illustrates that theautofluorescence peaks are shifted by 15 nm. The twoautofluorescence peaks for the normal tissue are emittedat 390 and 450 nm, the intensity ratio being: R337normal ¼I450/I390 ¼ 1.44; for the tumor sample, R337tumor ¼I450/I390 ¼ 1.31. Another parameter measured to individu-alize the autofluorescence curves is the overall intensity,defined as Ioverall ¼ $I(k). It is specific for each curve, andis a function of the pumping wavelength. For Fig. 7,I337normal ¼ 18.298, respectively, I337tumor ¼ 12.852. Theabove-mentioned quantities (the peak intensity walk-off,the ratio between the main fluorescence peaks, and theoverall autofluorescence intensity) may be used asobjective parameters to differentiate between normal andtumor brain tissues. For excitation of the samples at337.1 nm, these parameters differ from the normal to thetumor tissue by 15 nm peak intensity walk-off, 10%variation of the ratio of the peak intensities, and about40% difference in the overall autofluorescence intensity.

The same kind of measurement is made for samplesexcited at 370 nm. Figure 8 shows the autofluorescencespectra for the normal and tumor samples. The peak in-tensity ratios for 470 and 525 nm are R370normal ¼ I470/I525 ¼ 1.51; R370tumor ¼ I470/I525 ¼ 2.09; the difference isabout 40%. Thus, I370normal ¼ 20.2 and I370tumor ¼ 16.4, adifference of 23%. For the excitation at 410 nm, theobtained autofluorescence spectra are shown in Fig. 9;the respective parameters are R410normal ¼ 1.45 andR410tumor ¼ 2.15, the difference being 40%; I410normal ¼10.52 and I410tumor ¼ 16.125, the difference being 27%.

DISCUSSION

The reported data, correlated with literature reports(Hogan, 2008; Kremer et al., 2009) show that autofluor-escence measurements may constitute a promising

Fig. 6. The autofluorescence emission of a normal brain tissueexcited at 337.1 nm (sample no.1). Fig. 7. The autofluorescence emission of a healthy brain tissue and

of brain malignant tumor tissue excited at 337.1 nm.

2020 PASCU ET AL.

method to differentiate between brain tumor tissue andnormal brain tissue in the same patient. We demon-strated that by comparing the autofluorescence spectrainduced by laser radiation in the ultraviolet and visible,brain tumor, and normal tissues can be identified anddifferentiated. The recommended measurements werenot performed in tissue samples as they were extractedfrom the brain but on homogenates to ensure the bestpossible reproducibility of the results.

The conclusions were that the autofluorescence spec-tra of the tumor samples are close to those obtained fornormal tissues, but there are differences between themthat allow distinguishing the tumor from normal tissue.One difference is that for each tumor/normal tissue sam-ple pair the peak autofluorescence for the normal tissueis translated with respect to that for the tumor, typicallybetween 10 and 20 nm; the overall autofluorescence in-

tensity is different for the components of the same pair,the difference being 15–30%.

Another parameter that can be used is the variation ofthe ratio of some fluorescence intensity peaks betweenthe normal and the tumor tissue samples. When this pa-rameter can be measured, the variation range between10% and 40%.

Another conclusion is that in vitro experiments showthat for the measurements, it is mandatory to use nor-mal/tumor tissue sample pairs taken from the samepatient. In our case, for ethical considerations, the sam-ples used were extracted from the brain during medi-cally justified neurosurgical operations, in accord withoperative standards. The normal tissue samples wereextracted from normal tissues affected by nonrupturedaneurisms in patients who were different from thepatients exhibiting malignant brain tumors. The resultsshow that the method may be adapted, after further ex-perimental in vitro tests, to real-time intraoperative con-ditions by measuring the autofluorescence of the tumorand of the adjacent normal tissue; to do that, it is firstnecessary to consider the effect of tumor pressure on themolecular content and structure of the adjacent normaltissues. Because real-time investigation does not requiretissue extraction, the method could be acceptable to theethics bodies supervising neurosurgical operations;intraoperative fluorescence guidance for resection of ma-lignant tumors is currently receiving more clinical inter-est both for imaging purposes (Kantelhardt et al., 2007)and for fast intraoperative diagnostics (Croce et al.,2003; Kremer et al., 2009).

The results we report in this paper demonstrate thatfor neurosurgical clinical use of autofluorescence meas-urements, the following precautions should be taken: (a)the autofluorescence measurements, either in vitro or invivo, should be made using tissue pairs: normal/malig-nant tumor; (b) the contribution of any anesthetic to themeasured autofluorescence should always be evaluated,to avoid unexpected interference with the natural auto-fluorescence of the tissues and errors in data interpreta-tion; the fluorescence emitted by exogenous fluorophoreswhich might be present in the brain area submitted toautofluorescence screening should also be carefully eval-uated; (c) in obtaining the autofluorescence spectra, theoptimal intensity of the laser excitation beam should bechosen to prevent both damage to the brain tissues andthe quadratic Stark broadening of the absorption andfluorescence bands, the latter might induce errors in theaccurate measurement of the peak wavelengths and ofthe overall fluorescence bands.

LITERATURE CITED

Bottirolli G, Croce A, Locatelli D, Nano R, Giombelli E, Messina A,Benericetti E. 1998. Brain tissue autofluorescence: an aid forintraoperative delineation of tumor resection margins. CancerDetect Prev 22:330–339.

Croce AC, Fiorani S, Locatelli D, Nano R, Ceroni M, Tancioni F,Giombelli E. 2003. Diagnostic potential of autofluorescence for anassisted intraoperative delineation of glioblastoma resectionmargins. Photochem Photobiol 77:309–318.

Cubillos S, Obregon F, Vargas MF, Salazar LA, Lima L. 2006.Taurine concentration in human gliomas and meningiomas:tumoral, peritumoral and extratumoral tissue. Adv Exp Med Biol583:419–422.

Fig. 9. The autofluorescence spectra for healthy and malignant tu-mor bran tissues samples (excitation at 410 nm).

Fig. 8. The autofluorescence spectra for healthy and malignant tu-mor bran tissues samples (excitation at 370 nm).

LASER-INDUCED BRAIN TISSUES AUTOFLUORESCENCE 2021

Danaila L, Pascu ML. 1999. Photodynamic therapy of the cerebralmalignant tumours. Rom J Biophys 10:145–152.

Danaila L, Pascu ML. 2001. Lasers in neurosurgery. Bucharest:Academia Romana.

Drezek R, Brookner C, Pavlova I, Boiko I, Malpica A, Lotan R, Fol-len M, Richards-Kortum R. 2001b. Autofluorescence microscopy offresh cervical-tissue sections reveals alterations in tissue bio-chemistry with dysplasia. Photochem Photobiol 73:636–641.

Drezek R, Sokolov K, Utzinger U, Boiko I, Malpica A, Follen M,Richards-Kortum R. 2001a. Understanding the contributions ofNADH and collagen to cervical tissue fluorescence spectra: model-ing, measurements, and implications. J Biomed Opt 6:385–396.

Hogan H. 2008. Mapping molecular highways: second harmonicgeneration reveals polarity in neurons. Biophoton Int 36–37.

Kantelhardt SR, Leppert J, Krajewski J, Petkus N, Reusche E,Tronnier VM, Huttmann G, Giese A. 2007. Imaging of brain andbrain tumor specimens by time-resolved multiphoton excitationmicroscopy ex-vivo. Neuro Oncol 9:103–112.

Kremer P, Mahmoudreza F, Ding R, Pritsch M, Zouba S, Frei E.2009. Intraoperative fluorescence staining of malignant braintumors using 5-aminofluorescein-labelled albumin. Oper Neuro-surg 64:53–61.Lin W-C, Mahadevan-Jansen A, Jansen ED, Toms SA, Steven A.2002. Tumor demarcation using optical spectroscopy. US Pat.6,377,841.

Lin W-C, Toms SA, Johnson M, Jansen ED, Mahadevan-Jansen A.2001. In vivo brain tumor demarcation using optical spectroscopy.Photochem Photobiol 73:396–402.

Lin W-C, Toms SA, Motamedi M, Jansen ED, Mahadevan-Jansen A.2000. Brain tumor demarcation using spectroscopy: an in vitrostudy. J Biomed Opt 5:214–220.

Pascu ML. 2000. Laser physics. In: Simunovic Z, editor. Lasers inmedicine and dentistry, Rijeka, Croatia: Vitagraaf d. o. o.

Ramanujam N. 2000. Fluorescence spectroscopy in vivo. In: MeyersRA, editor. Encyclopedia of analytical chemistry. Chichester:Wiley. p 20–56.

Ramanujam N, Mitchell MF, Mahadevan A, Warren S, Thomsen S,Silva E, Richards-Kortum R. 1994. In vivo diagnosis of cervicalintraepithelial neoplasia using 337-nm-excited laser-induced fluo-rescence. Proc Natl Acad Sci USA 91:10193–10197.

Sokolov K, Follen MM, Richards-Kortum R. 2002. Opticalspectroscopy for detection of neoplasia. Curr Opin Chem Biol6:651–658.

Trujillo EV, Sandison DR, Utzinger U, Ramanujam N, Follen-Mitch-ell M, Richards-Kortum R. 1998. Method to determine tissuefluorescence efficiency in vivo and predict signal-to-noise ratio forspectrometers. Appl Spectrosc 52:943–951.

Utzinger U, Trujillo EV, Atkinson EN, Mitchell MF, Cantor SB,Richards-Kortum R. 1999. Performance estimation of diagnostictests for cervical precancer based on fluorescence spectroscopy:effects of tissue type, sample size, population, and signal-to-noiseratio. IEEE Trans Biomed Eng 46:1293–1303.

Valeur B. 2001. Molecular fluorescence: principles and applications.Weinheim: Wiley.

Wu T, Qu JY, Cheung T-H, Lo KW-K. 2003. Preliminary study ofdetecting neoplastic growths in vivo with real-time calibratedautofluorescence imaging. Opt Exp 11:291–298.

Yong W, Butte PV, Pikul BK, Jo JA, Fang Q, Papaioannou T, BlackKL, Marcu L. 2006. Distinction of brain tissue, low-grade andhigh-grade glioma with time-resolved fluorescence spectroscopy.Front Biosci 11:1255–1263.

Zuluaga AF, Utzinger U, Durkin A, Fuchs H, Gillenwater A,Jacob R, Kemp B, Fan J, Richards-Kortum R. 1999. Fluorescenceexcitation emission matrices of human tissue: a system for in vivomeasurement and method of data analysis. Appl Spectrosc53:302–311.

2022 PASCU ET AL.