Long wavelength identification of microcalcifications in breast cancer tissue using a quantum cascade laser and upconversion detection

Transcription

1 Downloded from orbit.dtu.dk on: Sep 28, 2018 Long wvelength identifiction of microclcifictions in brest cncer tissue using quntum cscde lser nd upconversion detection Tseng, Yu-Pei; Bouzy, P.; Stone, N.; Pedersen, Christin; Tidemnd-Lichtenberg, Peter Published in: Proceedings of SPIE Link to rticle, DOI: / Publiction dte: 2018 Document Version Publisher's PDF, lso known s Version of record Link bck to DTU Orbit Cittion (APA): Tseng, Y-P., Bouzy, P., Stone, N., Pedersen, C., & Tidemnd-Lichtenberg, P. (2018). Long wvelength identifiction of microclcifictions in brest cncer tissue using quntum cscde lser nd upconversion detection. In Proceedings of SPIE [104900F] SPIE - Interntionl Society for Opticl Engineering. (Proceedings of SPIE, the Interntionl Society for Opticl Engineering, Vol ). DOI: / Generl rights Copyright nd morl rights for the publictions mde ccessible in the public portl re retined by the uthors nd/or other copyright owners nd it is condition of ccessing publictions tht users recognise nd bide by the legl requirements ssocited with these rights. Users my downlod nd print one copy of ny publiction from the public portl for the purpose of privte study or reserch. You my not further distribute the mteril or use it for ny profit-mking ctivity or commercil gin You my freely distribute the URL identifying the publiction in the public portl If you believe tht this document breches copyright plese contct us providing detils, nd we will remove ccess to the work immeditely nd investigte your clim.

2 PROCEEDINGS OF SPIE SPIEDigitlLibrry.org/conference-proceedings-of-spie Long wvelength identifiction of microclcifictions in brest cncer tissue using quntum cscde lser nd upconversion detection Y. P. Tseng, P. Bouzy, N. Stone, C. Pedersen, P. Tidemnd-Lichtenberg Y. P. Tseng, P. Bouzy, N. Stone, C. Pedersen, P. Tidemnd-Lichtenberg, "Long wvelength identifiction of microclcifictions in brest cncer tissue using quntum cscde lser nd upconversion detection," Proc. SPIE 10490, Biomedicl Vibrtionl Spectroscopy 2018: Advnces in Reserch nd Industry, F (13 Februry 2018); doi: / Event: SPIE BiOS, 2018, Sn Frncisco, Cliforni, United Sttes

3 Long wvelength identifiction of microclcifictions in brest cncer tissue using quntum cscde lser nd upconversion detection Y.P. Tseng*, P. Bouzy b, N. Stone b, C. Pedersen, P. Tidemnd-Lichtenberg Technicl university of Denmrk, DTU Fotonik, Frederiksborgvej 399, 4000 Roskilde, Denmrk; b School of Physics nd Astronomy, University of Exeter, EX4 4QL, Englnd ABSTRACT Spectrl imging in the long-wve infrred regime hs gret potentil for medicl dignostics. Brest cncer is the most common cncer mongst femles in the US. The pthologicl fetures nd the occurrence of the microclcifictions re still poorly understood. However, two types of microclcifictions hve been identified s unique biomrkers: type I consisting of clcium oxlte (benign lesions) nd type II composed of hydroxyptite (benign or invsive lesions). In this study, we propose new pproch bsed on vibrtionl spectroscopy tht is non-destructive, lbel-free nd chemiclly specific for brest cncer detection. Long-wve infrred spectroscopy combining quntum cscde lsers (QCL) nd upconversion detection, offer to improve signl-to-noise rtios compred to stndrd long-wve infrred spectroscopy. We demonstrted long-wve identifiction of synthetic smples of crbonted hydroxyptite nd of microclcifiction in brest cncer tissue using upconversion detection. Absorbnce spectr nd upconverted imges of in situ brest cncer biopsy re compred with tht of Fourier-trnsform infrred (FTIR) spectroscopy. Keywords: Upconversion, Infrred imge, Medicl optics 1. INTRODUCTION Brest cncer is the most commonly dignosed cncer mongst femles in the USA nd comes in second plce s the most frequent cuse of deth from cncer fter lung cncer. In 2017, round 41,000 women re expected to die from brest cncer in the USA [1]. In order to increse the survivl rte, n erly dignosis is importnt. In this context, the microclcifictions re the unique erly mrker for brest cncer detection. They re chrcterized by bnorml clcium deposits in mmmry glnd. Mmmogrphy is commonly used to screen popultions for brest cncers. Unfortuntely, this technique is not ble to discriminte between microclcifictions from benign or mlignnt lesion [2]. Two types of microclcifictions re observed in the brest: type I consisting of clcium oxlte (benign lesions) nd type II composed of clcium hydroxyptite (prolifertive lesions nd invsive cncer). In biologicl tissues, hydroxyptite (HAP) is not pure compound but contins within its lttice round 5-6 % of crbonte (CO 2-3 ) substitution [3]. Two types of substitution hve been observed for the intercltion of crbonte in the HAP lttice: type 2- A when CO 3 replces the OH - 2- ions, nd type B when CO 3 replces the phosphte (PO 3-4 ) ions [4]. In order to understnd the pthologicl minerliztion process in biologicl smples, in prticulr in brest cncer microclcifiction, Fourier Trnsform Infrred (FTIR) spectroscopy cn be used to mesure the spectrl fetures of brest cncer tissue nd HAP synthetic crbonted smples [3]. FTIR system is equipped with stndrd infrred detectors, such s Mercury cdmium telluride (MCT) detector nd Focl Plne Arrys (FPA). However, direct infrred detectors generlly suffer from inherent drk noise nd require cooling for low-noise infrred detection [5]. In recent yers frequency upconversion hs been revitlized s n lterntive pproch to direct detection of infrred rdition [6,7]. Converting long-wve infrred emission into the visible/ner infrred regime vi nonliner frequency conversion llows for the use of silicon bsed detectors hving signl-to-noise rtio, orders of mgnitude better thn tht of direct infrred detectors. Moreover, silicon detectors re low noise, fst, sensitive, nd cost efficient compred to their mid-infrred counterprt; lthough this needs to be blnced with losses in signl during the frequency upconversion process. *yupts@fotonik.dtu.dk Biomedicl Vibrtionl Spectroscopy 2018: Advnces in Reserch nd Industry, edited by Anit Mhdevn-Jnsen, Wolfgng Petrich, Proc. of SPIE Vol , F 2018 SPIE CCC code: /18/$18 doi: / Proc. of SPIE Vol F-1

4 In this pper, we demonstrte new pproch for non-destructive, lbel-free nd chemiclly specific vibrtionl spectroscopy bsed on frequency upconversion. The combintion of high brightness infrred source i.e. quntum cscde lser (QCL), n x- nd y-xis scnning microscope, nd upconversion detection potentilly offers n improved signl-to-noise rtio compred to stndrd infrred spectroscopy imging systems in the long-wve infrred (LWIR) regime. Rster scnning t discrete wvelengths is used to form chemicl imges t selected wvelengths ccording to the spectrl fetures in brest cncer tissue. 2. EXPERIMENTAL SETUP The setup is illustrted in Figure 1. A QCL is used s widely tunble, nrrow linewidth, long-wve infrred source spnning from 9.43 to 12 µm, delivering 50 ns pulses with temporl seprtion of 1 µs. The LWIR bem is coupled into homemde scnning microscope llowing for mechniclly stble scnning of the smple. The smple is plced between two ZnSe lenses of focl lengths f 1 = f 2 = 20 mm giving LWIR bem wist (rdius) of pprox. 48 µm in the smple plne. The bem dimeter determines the sptil resolution in the rster scnning pproch. The LWIR bem is then guided nd refocused into the upconversion unit in which the bem (100 µm rdius) is mixed with ner infrred (NIR) lser bem (100 µm rdius) inside nonliner crystl optimized for sum frequency genertion. An 8 W diode lser is used to pump Nd:YVO 4 lser generting the 1064 nm single-pss mixing field, giving mximum output power of 3W in continuous-wve opertion. The sum frequency genertion (SFG) is relized in 5 x 5 x 10 mm 3 AgGS 2 crystl. The AgGS 2 crystl is cut t θ = 43.3 (φ = 0) for colliner type II phse-mtching (e LWIR + o NIR e Up ). The AgGS 2 crystl is mounted on piezo controlled rottion stge for ngle tuned birefringent phse-mtching, converting the LWIR signl to the 950 to 980 nm rnge for detection using silicon power meter. ir=20rm C -.> X-YStgefz= 20 mm b Microscopy unit if= 50 mm F ers -I- )1,5150 rm Silicon Power meter Upconverter unit Figure 1: () Schemtic lyout of LWIR light source, scnning microscopy unit, nd upconversion detection unit bsed on sum frequency genertion (b) bem profile t the focus of LWIR light source. 3. MATERIALS AND METHODS 3.1 KBr pellets with Hydroxyptite smples In order to improve the understnding of the pthologicl minerliztion process in brest cncer, we first exmine synthetic smples of crbonted hydroxyptite (HAP) using vibrtionl spectroscopy. For this, we hve compred the performnce of commercil FTIR system nd the developed upconversion system. KBr pellets of different stndrd minerls, provided by Prof Mry Tecklenburg, Centrl Michign University, USA, contining crbonte substitution of 1.24, 2.92, 4.43, 5.24, 7.52 nd 8.12% re investigted ccording to the following protocol: 99 mg of KBr powder, ws Proc. of SPIE Vol F-2

5 dded to 0.5 mg of Hydroxyptite (HAP) nd then mixed nd grounded together by mortr nd pestle. The mixture ws trnsferred into hydrulic press nd compressed for 30 minutes with pressure of 10 tons. A pellet contining only 100 mg of KBr hs been used to obtin bckground spectrum. 3.2 Brest cncer smple An in situ brest cncer biopsy with microclcifictions, provided by Dr Ctherine Kendll, Gloucestershire Royl Hospitls NHS Foundtion Trust, ws nlyzed using the upconversion system. We compred the imges nd spectr of this smple with using n FTIR Agilent system nd upconversion. The smple cme from ptient who hd undergone biopsy for mmmogrphiclly suspicious brest lesions, nd it ws selected from the histopthologicl report. The biopsy ws embedded in prffin, two djcent sections of 3 µm thickness were nlyzed: one section mounted on Brium fluoride (BF 2 ) without stining ws nlyzed using the FTIR nd upconversion systems, nd one section stined with Hemtoxylin Eosin (H&E) for the histologicl nlysis s gold stndrd for dignosis. 3.3 Micro-FTIR system The brest cncer biopsies were nlyzed using Agilent 670 FTIR spectrometer coupled with n Agilent 620 FTIR microscope, equipped with 15X Cssegrin objective nd Mercure Cdmium Telluride - Focl Arry Plne (MCT- FPA) detector comprising of 128x128 pixels nd Globr source. Spectr were recorded with spectrl resolution of 4 cm -1, spectrl rnge comprised between cm -1 nd sptil resolution of 5.5µm. Ech spectrum ws obtined by verging 64 scns nd bckground spectrum ws lso recorded with 256 scns. The FTIR spectr were collected using Agilent ResolutionPro softwre. The KBr pellet contining HAP were nlyzed in single-point mode using n Agilent 670 FTIR spectrometer coupled with n Agilent 620 FTIR microscope, equipped with 15X cssegrin objective nd MCT detector. Spectr were recorded in the spectrl rnge between cm -1, with spectrl resolution of 4 cm -1. Ech spectrum ws obtined by verging 32 scns nd bckground spectrum with 64 scns. 4. RESULTS AND DISCUSSION 4.1 KBr pellets with Hydroxyptite smples Absorption spectr of crbonted hydroxyptite smples were recorded using FTIR system from 4000 to 700 cm -1 with spectrl resolution of 4 cm -1, re shown in Figure 2(). Phosphte bsorption peks re found in the rnge of 1020 to 1040 nd t 962 cm -1, nd crbonte pek is observed t 875 cm -1 using the FTIR system. Figure 2(b) shows the spectrum of HAP smples with 7.52 nd 2.92% crbonte substitution mesured using the upconversion system shown in Figure 1. The spectr were recorded from 1066 to 830 cm -1 in steps of 4 cm -1, nd verged by 5 scns for ech smple. The mesured spectr re normlized similr to FTIR bsorbnce spectr in Figure 2(). The HAP spectr shown in Figure 2(b) were mesured t single point, with bem size of 48 µm rdius, t the smple pellets. Absorption peks in phosphte nd crbonte bnds re observed in smoothed spectrum (not shown) t 1044, 964 nd 875 cm -1 in the HAP 7.52% crbonte substitution, nd t 1044, 964 nd 882 cm -1 in the HAP 2.92% crbonte substitution P u% -2.92% 4.43% -6.2/% -7.82% 8.12% co - m 0.1 N b -Hp 2.92% Hp 7.52 %_ Wvenumber (cm-1) CO,' co E 0.05 o Z 1$ Wvenumber [cm-1] Figure 2: () Spectr of KBr pellets contining HAP powders mesured using micro-ftir Agilent system in the spectrl rnge cm -1 (b) Normlized spectr of HAP with 7.52% nd 2.92% crbonte substitution mesured using upconversion system. Proc. of SPIE Vol F-3

6 It is noted tht the bsorption spectr mesured by upconversion in Figure 2(b), ppers noisier thn tht of the FTIR spectr, see Figure 2(). In order to investigte the origin of the noise in the upconversion spectr, the bsorption profile of the HAP 7.52% pellet ws investigted t 881 nd 920 cm -1 using the upconversion system with rster scnning. Figure 3 shows the trnsmitted intensity in the upconverted imges of the HAP smple with 7.52% crbonte substitution in 2D t the two different LWIR wvelengths. The sme re ws scnned t the two wvelengths, showing n interference-like pttern, however, shifted sptilly. A similr pttern ws observed in nother pellet (not shown). The interference fringe re moving with wvelength, hence, the noise presented in the upconverted spectr is expectedly due to interference between the front nd bck surfce of the pellet, when using coherent illumintion source like the QCL. 2D view 881 cm1(crbonte bnd) 920 cm-1 (No spectrl feture) 2D view o 0.3 7e0 0.1 o. o.s ]mm] 0.1 0b Mm] m so 3D view Figure 3: 2D/3D Intensity mps of HAP with 7.52% crbonte substitution mesured t 881 nd 920 cm -1 respectively mesured using upconversion system. Note the different scles t the two wvelengths. The HAP spectr shown in Figure 2(b) were mesured t single point in KBr pellets. The noise observed in upconversion spectr ppers to be minly due to interference effects rther thn heterogeneity of the HAP powder in the pellet. The trnsmitted intensity distributions seen in Figure 3 re consistent with the use of coherent light source to mesure on slightly wedged pellet. 4.2 Brest cncer tissue Figure 4 shows the FTIR nd the upconverted imges of microclcifictions found in in situ brest cncer tissue. Figure 4() shows histologicl section of in situ brest cncer tissue with H&E stining. Figure 4(b) is obtined t 1032 cm -1 using the micro-ftir imging system with MCT-FPA detector including 128x128 pixel rrys. Figure 4(c) shows the sme tissue smple imged by upconversion specificlly selecting the wvenumber of 1020 cm -1 corresponding to the phosphte bnd. The upconverted imge is mesured by rster scnning with step size of 20 µm. The LWIR signl through the smple ws upconverted (from 1020 cm -1 ) into NIR (959 nm) which ws detected using silicon bsed power meter. The upconverted imge is seen to show good greement with the FTIR imge t similr wvenumber. Proc. of SPIE Vol F-4

7 c Figure 4: () Histologicl section with H&E stining using u trnsmisssion microscopee with mgnificction x4 (b) FTIR imge t phosphte pek distributioon t 1032 cm m-1 with mgniffiction x15 (piixel size: 5.5 µ µm2) nd (c) up pconverted refers to the -1 imge refe fers to the phospphte pek distrribution t cm with scn nned step size of o 20 µm. Only few disccrete wvelenngths re requiired to identify fy the microclcifictions foor phosphte nd crbonte bnds t nd 875 cm-1. Figure 5 shhows the upcoonverted imgges obtined t the wvenum mbers of 1020, 975, 964, 920, 9 881, 8755, nd 842 cm-1, respectively.. These imgees re cquiredd with rsterr scnning stepp size of 20 µ µm. It cn be seen, s tht theree is significnnt decrese inn signl (incresed bsorbnnce) t the pho osphte bsorrption bnd inn the microclcifiction re (blck squree) compred to t the surrounnding. Acquiriing imges tt other differeent wvelengthhs llows the estimtion of the reltive coompositionl contributions to the spectrl bsorption, thus enblingg nlyzed furrther nlysis of the spectrl fetures of microclcificti m ions. Figure 5: Upconverted imges of in situ s brest cnccer tissue obtiined t 1020, 975, 9 964, 920, 881, 875, nd d 842 cm-1 o 959 to 976 nm. n corresponnding to the upcconverted signls in the rnge of Proc. of SPIE Vol F-5

8 The bsorption spectrum is nlyzed in the re where phosphte bnd is highly bsorbing in the Figure 6(). Figure 6(b) shows the spectrum extrcted from microclcifiction re (dshed squre), wheres Figure 6(c) shows spectrum extrcted from surrounding tissue. The MCT-FPA detector becomes noisy in the LWIR regime ( cm -1 ), therefore the signtures below 900 cm -1 cnnot be considered relible. The upconverted spectr ws retrieved from series of imges (shown in Figure 5) mesured t 1020, 975, 964, 920, 881, 875, nd 842 cm -1, respectively. The FTIR nd upconverted spectr mesured from surrounding tissue demonstrte tht there is only smll tissue phosphte bsorption feture in this re. A single pixel (green circle) nd n verging of 6 pixels (red circle) using upconversion re compred in Figure 6 (b) nd (c). A spectrum hs been lso extrcted from FTIR imge (Figure 4 b) recorded using micro-ftir Agilent imging system. b Wvenumber = 1020cm c 1 Á á 0.5 -Agilent FTIR Single < Averge Ê C Wvenumber [cm-1] Agilent FTIR Single Averge [Jim] o Wvenumber Figure 6: () Upconverted imge t phosphte pek t 1020 cm -1 nd spectr mesured using micro-ftir imging nd upconversion systems from (b) microclcifiction nd (c) tissue surrounding in in situ brest smple. 4. CONCLUSION It is demonstrted in the study tht brest cncer tissues cn be nlyzed using n upconversion detector with rster scnning microscopy system t discrete wvenumbers, potentilly improving the signl-to noise rtio using silicon bsed detectors when compred to direct detection in the LWIR rnge. It ws possible to detect the crbonte bnds t 875 cm -1 tht cn be chllenging for the Agilent micro-ftir imging system to mesure due to combintion of the limited number of IR photons longer thn 10 μm imged on the MCT-FPA detector nd its inherent sensitivity in this wvelength rnge. Only few discrete wvelengths re required to identify the microclcifiction from phosphte nd crbonte bnds t 1020 nd 875 cm -1, respectively. Therefore, the combintion of tunble QCL, rster scnning microscope, nd the upconversion system shows potentil for chemicl imging of tissue t selected wvelengths. The system cn potentilly be developed into brest biopsy dignostic tool for brest cncer in clinicl setting. ACKNOWLEDGMENTS The Mid-TECH project hs received founding from the Europen Union Horizon 2020 reserch innovtion progrmme under Grnt Agreement No We thnk Prof Mry Tecklenburg for providing us crbonted hydroxyptite powers. Proc. of SPIE Vol F-6

9 REFERENCES [1] DeSntis C. E., M J., Suer A.G., Newmn A.L., Jeml A. Brest Cncer Sttistics, 2017, Rcil Disprity in Mortlity by Stte, CA Cncer J Clin 67, (2017). [2] Bker R., Rogers K. D., Prker A.W., Stone N. New reltionships between brest microclcifictions nd cncer, British Journl of Cncer. 103, (2010). [3] Pleshko N., Boskey A., Mendelsohn R. Novel infrred spectroscopic method for the determintion of crystllinity of hydroxyptite minerls, Biophysicl journl. 60, (1991). [4] Kerssens M.M., Mtousek P., Rogers K., Stone N. Towrds sfe non-invsive method for evluting the crbonte substitution levels of hydroxyptite (HAP) in microclcifictions found in brest tissue, Anlyst. 135, (2010). [5] Roglski A., Infrred Detectors, CRC Press, [6] Dm J. S., Pedersen C., Tidemnd-Lichtenberg P., Room temperture mid-ir single photon spectrl imging Nture Photonics, 6, (2012). [7] Tidemnd-Lichtenberg P., Dm J. S., Andersen H. V., Høgstedt L., nd Pedersen C., Mid-infrred upconversion spectroscopy J. Opt. Soc. Am. B 33(11), D28 -D35 (2016). Proc. of SPIE Vol F-7