Delineating Margins of Lentigo Maligna (Melanoma) Using a Hyperspectral Imaging System

Ac a De m Vene eol 95 INVESTIGATIVE REPORT Ac a De m Vene eol 2015; 95: 549–552 © 2015 The Au ho s. doi: 10.2340/00015555-2010 Jou nal Compila ion © 2015 Ac a De ma o-Vene eologica. ISSN 0001-5555 Len igo maligna (LM) is an in si u o m o melanoma which can p og ess in o in asi e len igo maligna mela- noma (LMM). Va ia ions in he pigmen a ion and hus isibili y o he umou make assessmen o lesion bo - de s challenging. We es ed hype spec al imaging sys- em (HIS) in in i o p eope a i e delinea ion o LM and LMM ma gins. We compa ed lesion ma gins delinea ed by HIS wi h hose es ima ed clinically, and con i med his ologically. A o al o 14 LMs and 5 LMMs in 19 pa- ien s we e included. HIS analysis ma ched he his o- pa hological analysis in 18/19 (94.7%) cases while in 1/19 (5.3%) cases HIS showed lesion ex ension no con i med by his opa hology ( alse posi i es). Compa ed o clinical examina ion, HIS de ined lesion bo de s mo e accu a- ely in 10/19 (52.6%) o cases (wide , n = 7 o smalle , n = 3) while in 8/19 (42.1%) cases lesion bo de s we e he same as delinea ed clinically as con i med his ologically. Thus, HIS is use ul o he de ec ion o subclinical LM/ LMM bo de s. Key wo ds: len igo maligna; len igo ma- ligna melanoma; umou ma gin assessmen ; hype spec- al imaging. Accep ed No 12, 2014; Epub ahead o p in No 14, 2014 Ac a De m Vene eol 2015; 95: 549–552. Noo a Nei aanmäki-Pe u, MD, Depa men o De - ma ology and Alle gology, Helsinki Uni e si y Cen al Hospi al, Box 160 Meilahden ie 2, FIN-00029 Helsinki, Finland. E-mail: [email p o ec ed] Len igo maligna (LM) is an in si u o m o melanoma whe e he neoplas ic cells a e con ined o he epide mis and lack de mal in asion. LM is he mos p e alen in si u sub ype (79–83%) o melanoma. I un ea ed LM may p og ess in o in asi e len igo maligna melanoma (LMM). The incidence o LM and LMM is cons an ly inc easing o e he o he melanoma sub ypes (1, 2). Ea ly and e icien su gical emo al is he me hod o choice in he ea men o LM and LMM. Assessmen o he bo de s o LM and LMM is chal- lenging bo h clinically and his ologically. Lesions can ex end se e al cm beyond he clinically es ima ed ma gins (3). In clinical p ac ice Wood’s ligh (320–400 nm) is widely used o help he delinea ion o he umou ma gins. Melanin abso bs mos o he UV adia ion skin is exposed o. Thus, Wood’s ligh inc eases he con as be ween heal hy skin and a eas p esen ing e en mino inc eases in epide mal melanin pigmen a ion (4). How- e e , he pigmen con en may no be inc eased in all a eas o LM g ow h. We ha e ea lie shown ha a no el hype spec al ima- ging sys em (HIS) e icien ly de ec s a eas o subclinical skin ield cance isa ion (5). Hype spec al imaging combines adi ional spec oscopy and imaging echni- ques by p oducing 3 dimensional da a cubes, e e ed as hype spec al images, whe e in addi ion o spa ial (x, y loca ion) in o ma ion, image con ains a spec al g aph (z in ensi y) o each pixel (6). Thus, hype spec- al image consis s o a s ack o images, o which each image is aken a di e en na ow wa eleng h and each pixel in he image has i s own spec al signa u e (Fig 1). Delinea ing Ma gins o Len igo Maligna Using a Hype spec al Imaging Sys em Noo a NEITTAANMÄKI-PERTTU1,2, Ma i GRÖNROOS2, Leila JESKANEN1, Ilkka PÖLÖNEN3, Annama i RANKI1, Olli SAKSELA1 and E na SNELLMAN2,4 Depa men o De ma ology and Alle gology, 1Helsinki Uni e si y Cen al Hospi al, Helsinki and 2Päijä -Häme Cen al Hospi al, Lah i, 3Depa men o Ma hema ical In o ma ion Technology, Uni e si y o Jy äskylä, Jy äskylä, and 4Depa men o De ma ology, Tampe e Uni e si y and Tampe e Uni e si y Hospi al, Tampe e, Finland Fig. 1. Hype spec al imaging p ocess and da a analysis. Hype spec al image (cube) consis s o o e lapping images, o which each is aken a di e en wa eleng h. Each pixel in he hype spec al image has i s own spec al signa u e. Ma hema ical algo i hms (VCA, FVA) a e used o sepa a e he spec a (endmembe s) o benign and malignan issue. The abundance images ep esen he a eas o lesional and heal hy skin. 550 N. Nei aanmäki-Pe u e al. Di e en biological issues can be iden i ied om hei unique spec al signa u es e lec ing hei biochemical cha ac e is ics (6–8). This pilo s udy aimed o es he easibili y o he HIS in delinea ion o he ma gins o LM and LMM p eope a i ely in o de o a oid he need o e-excisions. MATERIALS AND METHODS Pa ien s The s udy p o ocol ollowed he Decla a ion o Helsinki and was app o ed by he local e hics commi ee. All olun ee ing pa ien s p o ided hei w i en in o med consen . The pa ien s we e ec ui ed om hose emi ed o he Depa men o De - ma ology o hei suspec ed LMs. Nine een pa ien s, 7 women and 12 men, wi h clinically suspec LM o LMM loca ed on hei aces o scalps we e included in he s udy. The pa ien s’ mean age was 77.9 ( ange 67–97 yea s). Th ee pa ien s displayed skin pho o- ype I, 8 pho o- ype II, and 8 pho o- ype III (9). Nine pa ien s ou o 19 had ea lie been ea ed o a non-melanoma skin cance o p emalignan skin lesions (basal cell ca cinoma n = 2, ac inic ke a osis n = 2 o bo h n = 5). None o he pa ien s had ea lie been ea ed o melanoma. Clinical assessmen o he lesion bo de s using digi al imaging and Wood’s ligh examina ion P io o he imaging p ocesses lesions we e e alua ed using de ma oscopy (De mli e® DL3). Clinical assessmen o he lesion bo de s was ca ied ou using Wood’s ligh examina ion (Bu - on®) and digi al pho og aphy (Canon Ixus 130, 14.1 megapixel). Hype spec al imaging sys em and image analysis All 19 lesions we e imaged p io o su gical emo al using HIS. The used handheld HIS was de eloped o he s udy a he VTT Technical Resea ch Cen e o Finland and Uni e si y o Jy äskylä, Finland. This p o o ype imaging sys em consis ed o a hype spec- al image (500–850 nm) (10, 11), ex e nal ligh sou ce o isible and in a ed ligh , ib e op ic ing ligh and a holde o he image and he ing ligh . The ield o iew was 12 cm2. HIS acqui ed he di use e lec ance o he de ec ed skin a eas apidly in a ew seconds. The acqui ed hype spec al da a cube was analysed using an assump ion o he linea mix u e model ( e ex componen analysis, VCA and il e ec o algo i hm FVA) (12–14) o achie e pu e spec a (endmembe s) o LM/LMM and heal hy skin (Fig. S11) and o p oduce abundance maps o delinea ion o he lesion bo de s (Fig. 2, Fig. S21). The hype spec al image and he ana- lysing p ocess a e shown in Fig. 1, and de ailed in Appendix S11. His opa hological sampling The lesions o 5/19 pa ien s we e biopsied be o e ec ui ing o he s udy and excised immedia ely a e he imaging p ocesses. The edges o he specimens we e ma ked wi h o ien ing su u es, and inked o o ien a ion. To help mapping he indings o 14/19 pa ien s, we ook a ge ed 3 mm punch biopsies (2–4 pe pa ien ) om he middle and om lesions bo de s, de ined using he HIS, and a e wa ds he lesions we e comple ely excised wi h wide excision ma gins. The biopsy si es and excision ma gins we e ma ked and pho og aphed. An expe ienced de ma opa hologis (LJ) examined he samples wi hou any backg ound in o ma ion. RESULTS The compa a i e analyses included a o al o 14 LMs and 5 LMMs loca ed on he ace and scalp (nose n = 1, ea n = 4, eyelid n = 2, o ehead n = 2 and on he cheek a ea n = 10) o 19 pa ien s. The in asion dep h (B es- low hickness) o he LMMs a ied be ween 0.5 and 1.25 mm. The mean lesion a ea was 2.4 cm 2 ( ange 1.6–7.6 cm 2 ). In he delinea ion o LM o LMM ma gins, HIS analysis ma ched he his opa hological analysis in 18/19 (94.7%) 1h p://www.medicaljou nals.se/ac a/con en /?doi=10.2340/00015555-2010 Fig. 2. Len igo maligna melanoma on lowe eyelid (pa ien 19). (a) Lesion in Wood’s ligh , (b) Clinical wide excision ma gins, (c) Hype spec al abundance map showing subclinical lesion ex ension (a ows), (d) His ological image (HE-s aining magni ica ion × 20) o he squa ed a e om Fig 2 c. A ypical melanocy ic nes s in de mo-epide mal junc ion and sola elas osis. The wide excision e i ied he HIS esul s o subclinical lesion ex ension. Ac a De m Vene eol 95 551 Hype spec al imaging o len igo malignas cases while in 1/19 (5.3%) cases HIS showed lesion ex ension no con i med by his opa hology ( alse posi i- es). In 10/19 (52.6%) o he cases lesion ma gins we e delinea ed mo e accu a ely (wide , n = 7 o smalle , n = 3) by HIS han by clinical examina ion wi h Wood’s ligh , as con i med by his opa hological analysis (Fig 2, Fig S2 1 ). In 8/19 (42.1%) cases he lesion ma gins we e equally delinea ed by HIS and clinical examina ion, as con i med by his opa hology. No alse nega i es we e de ec ed when compa ing HIS de ec ion wi h his opa hology. In he alse posi i e case an ac inic ke a osis and benign len igo was p esen his ologically on he LM bo de s which complica ed he in e p e a ion o he HIS image. The lesions we e ope a ed in acco dance wi h cu en s anda ds by emo ing LMs wi h 5 mm clinical ma gins and LMMs wi h 10 mm ma gins i ana omically pos- sible (15). As his was a pilo s udy, he ma gins gi en by HIS we e no used in he excisions. A e ecei ing he esul s om he his opa hological analyses, 3/14 o he LM lesions and 2/5 LMM lesions needed e-excision because o he subclinical ex ension o he lesion bo - de s. I he excision bo de s had been selec ed on he basis o he HIS analysis, 5 e-excisions could ha e been a oided (Table I). DISCUSSION This s udy indica es ha he HIS is capable o de ec ing he subclinical bo de s o LM and LMM. Impo an ly, in o e 50% o he cases, he lesion ma gins we e as- sessed mo e accu a ely by using he HIS han wi h he clinical me hods. The ad an ages o HIS include a handheld image wi h a la ge ield o iew (12 cm 2 ) and a quick imaging p ocess. All s udied skin a eas including he nose and ea s we e sui able o imaging wi h he HIS and i ed he ield o iew. The lesions we e emo ed wi h ma gins de e mined a e delinea ing he umou isually using Wood’s ligh . Th ee LM pa ien s and 2 LMM pa ien s needed e-ex- cision because o he his ologically e i ied subclinical ex ension o he lesions. In e es ingly, i he pa ien s had ini ially been ope a ed on using ma gins gi en by HIS, he e-excisions could ha e been a oided. HIS also seems o be in e sely use ul, since clinical assessmen s delinea ed 3 LM lesions inco ec ly la ge han depic ed using HIS and con i med by his opa hology. Disc imina ion o LM om sun-damaged skin a he pe iphe y o lesions may also be challenging his ologi- cally (3). The cy ological a ypia in LM may a y and be sub le. The di use melanocy ic o e g ow h o sun- damaged skin and he p esence o benign melanocy es along he hai ollicles make i challenging o assess he pe iphe al ma gins o LM. Immunohis ochemis y, o e.g. MART-1, is some imes used o iden i y LM. Howe e , also no mal, ch onically sun-exposed skin has a high numbe o MART-1 posi i e melanocy es (16, 17). In one case in ou s udy, benign len igines and subclinical AK su ounding he LM made i di - icul o co ec ly in e p e he HIS image and led o a alse posi i e in e p e a ion. These lesions migh ha e complica ed he his opa hological analysis as well. Wood’s ligh u ned ou o be o only ma ginal help in assessing he ma gins o LMs. Especially in cases whe e also benign len igines we e p esen , he delinea- ion o LM wi h Wood’s ligh was complica ed. Since bo h melanin and haemoglobin s ongly abso b isible and UV ligh (4), a u u e de elopmen al aspec could be o in eg a e a UV ligh sou ce o HIS o imp o e isualisa ion o he pigmen a ion. The e a e se e al comme cially a ailable de ices o skin cance de ec ion and a magni ude o esea ch in he ield (18, 19). As a as we know he e a e no comme cial applica ions o a HIS. P e iously, a ew echniques ha e been used o assess su gical ma gins includings: de ma oscopy, con ocal mic oscopy and op ical cohe ence omog aphy. De ma oscopy (epiluminescence mic oscopy) helps in de ining he LM bo de s, bu equi es an expe ienced de ma oscopis (20). LM on he ace do no show he classical de moscopic ea u es ound on he o he pa s o he skin, which makes hei obse a ion mo e chal- lenging (21). In his s udy de ma oscopy was used as a diagnos ic aid, bu no o he delinea ion o he lesions. Con ocal mic oscopy has shown po en ial in e alua- ing pigmen ed skin lesions and hei ma gins (22). The de ice de ec s o dep hs o 300 µm, i.e. o he papilla y Table I. Lesion cha ac e is ics in hype spec al images (HIS) compa ed wi h clinical e alua ion Pa ien B eslow hickness (mm) HIS use ula HIS p o ided no addi ional in o ma ionb False posi i ec Re-excision a oidable 1In si u Wide – – + 2In si u Wide – – + 7In si u Wide – – + 16 1.25 mm Wide – – + 19 0.50 mm Wide – – + 6In si u Wide d– – – 17 0.75 mm Wide d– – – 4In si u Smalle – – – 8In si u Smalle – – – 13 In si u Smalle – – – 3In si u – Iden ical – – 5In si u – Iden ical – – 9In si u – Iden ical – – 11 In si u – Iden ical – – 14 In si u – Iden ical – – 15 0.70 mm – Iden ical – – 18 0.70 mm – Iden ical – – 10 In si u – Iden ical – – 12 In si u – +Wide – aHIS shows lesion ma gins mo e accu a ely compa ed o clinical e alua ion as con i med his ologically. bLesion bo de s simila ly de ec ed clinically, by HIS and his ology. cHIS shows lesion wide han his ologically con i med. dOnly mino subclinical ex ension, no need o e-excision. Ac a De m Vene eol 95 552 N. Nei aanmäki-Pe u e al. de mis. The limi a ion in con ocal mic oscopy is a small ield o iew (FOV 8 × 8 mm mosaic composi e ima- ges) leading o se e al slow imaging sessions o each lesion. The me hod lacks any objec i e analysis and he assessmen o one FOV a ea equi es a leas 5 min o an expe which makes he me hod slow compa ed o HIS (23, 24). Op ical cohe ence omog aphy (OCT), has shown po en ial in delinea ing he bo de s o non-melanocy ic skin malignancies (25). As a as we know he e a e no s udies o he delinea ion o pigmen ed lesions using OCT. The de ice p o ides high- esolu ion c oss-sec io- nal images a g ea e dep hs (1.5–2 mm) han con ocal mic oscopy. The FOV is small (6 × 6 mm), hus making he imaging p ocess slowe han in HIS. The analysis emains subjec i e. We ha e ea lie shown ha he HIS is use ul in he de ec ion o skin ield cance isa ion (5). In he p esen s udy HIS showed i s po en ial in he de ec ion o he subclinical bo de s o LM and LMM. By de ec ing ac- cu a e ma gins o he lesions, cumbe some e-excisions could be a oided. In addi ion, HIS could also be used o spa e acial issue in cases whe e lesion bo de s a e smalle han shown by clinical assessmen s. HIS can o e clinicians a p ac ical ool o a non-in asi e de- linea ion o umou bo de s. As his was a pilo s udy wi h limi ed cases u he s udies a e wa an ed o alida e he esul s. ACKNOWLEDGEMENTS This s udy was unded by he No o No disk Founda ion’s No o P eSeed G an . The au ho s decla e no con lic s o in e es . REFERENCES 1. Reed JA, Shea CR. Len igo maligna melanoma in si u on ch onically sun-damaged skin. A ch Pa hol Lab Med 2011; 135: 838–841. 2. Swe e SM, Bold ick JC, Jung SY, Egbe BM, Ha ell JD. Inc easing incidence o len igo maligna melanoma sub y- pes: no he n Cali o nia and na ional ends 1990–2000. J In es De ma ol 2005; 125: 685–691. 3. B euninge H, Schlagenhau B, S oebel W, Schaumbu g- Le e G, Rassne G. Pa e ns o local ho izon al sp ead o melanomas: consequences o su ge y and his opa hologic in es iga ion. Am J Su g Pa hol 1999; 23: 1493–1498. 4. Pa ake as L-R, Halpe n AC, Ma ghoob AA. U ili y o Wood´s ligh : i e cases om pigmen ed lesion clinic. B J De ma ol 2005; 152: 1039–1044. 5. Nei aanmäki-Pe u N, G ön oos M, Tani T, Pölönen I, Ranki A, Saksela O, e al. De ec ing ield cance iza ion using a hype spec al imaging sys em. Lase s Su g Med 2013; 45: 410–417. 6. Ma in ME, Wabuyele MB, Chen K, Kasili P, Panjehpou M, Phan M, e al. De elopmen o an ad anced hype spec al imaging (HSI) sys em wi h applica ions o cance de ec- ion. Ann Biomed Eng 2006; 34: 1061–1068. 7. Siddiqi AM, Li H, Fa uque F, Williams W, Lai K, Hugh- son M, e al. Use o hype spec al imaging o dis inguish no mal, p ecance ous, and cance ous cells. Cance 2008; 114: 13–21. 8. Panasyuk SV, Yang S, Falle DV, Ngo D, Lew RA, F e- eman J, e al. Medical hype spec al imaging o acili a e esidual umo iden i ica ion du ing su ge y. Cance Biol The 2007; 6: 439–446. 9. Fi zpa ick TB. The alidi y and p ac icali y o sun- eac i e skin ypes I h ough VI. A ch De ma ol 1988; 124: 869–871. 10. Saa i H, Aallos V-V, Holmlund C, Malinen J, Mäkynen J. Handheld hype spec al image . P oc SPIE 2010; 7680. 11. Saa i H, Pölönen I, Salo H, Honka aa a E, Hakala T, Holm- lund C, e al. Minia u ized hype spec al image calib a ion and ua ligh campaigns. P oc Spie 2013; 8889: 10–12. 12. Nascimen o JMP, Dias JMB. Ve ex componen analysis: A Fas algo i hm o unmix hype spec al da a. IEEE T ans Geosci Remo e Sensing 2005; 43: 898–910. 13. Bowles J, Palmadesso P, An oniades J, BaumbackM, Rick- a d LJ. Uses o il e ec o s in hype spec al da a analysis. P oc SPIE 1995; 2553: 148–157. 14. B o R, De Jong S. A as non-nega i i y-cons ained leas squa es algo i hm. J Chemom 1997; 11: 393–401. 15. Bichakjian CK, Halpe n AC, Johnson TM, Foo e Hood A, G ichnik JM, Swe e SM, e al. Ame ican Academy o De ma ology Guidelines o ca e o he managemen o p ima y cu aneous melanoma. J Am Acad De ma ol 2011; 65: 1032–1047. 16. McLeod M, Choudha y S, Giannakakis K, Nou i K. Su - gical ea men s o len igo maligna: a e iew. De ma ol Su g 2011; 37: 1210–1228. 17. Hendi A, B odland DG, Zi elli JA. Melanocy es in long- s anding sun-exposed skin: quan i a i e analysis using he MART-1 immunos ain. A ch De ma ol 2006; 142: 871–876. 18. Calin MA, Pa asca SV, Sa as u R, Calin MR, Don u S. Op ical echniques o he nonin asi e diagnosis o skin cance . J Cance Res Clin Oncol 2013; 139: 1083–1104. 19. Fe is LK, Ha is RJ. New diagnos ic aids o melanoma. De ma ol Clin 2012; 30: 535–545. 20. Robinsson JK. Use o digi al epiluminescence mic oscopy o help de ine he edge o len igo maligna. A ch De ma ol 2004; 140: 1095–1100. 21. P along P, Ba helie E, Dalle S, Poulalhon N, Deba bieux S, Thomas L. De moscopy o len igo maligna: epo o 125 cases. B J De ma ol 2012; 167: 280–287. 22. Ca e a C, Palou J, Mal ehy J, Segu a S, Aguile a P, Sale ni G, e al. Ea ly s ages o melanoma on he limbs o high- isk pa ien s: clinical, de moscopic, e lec ance con ocal mic o- scopy and his opa hological cha ac e iza ion o imp o ed ecogni ion. Ac a De m Vene eol 2011; 91: 137–146. 23. Gui e a P, Moloney FJ, Menzies SW, S e ch JR, Quinn MJ, Hong A, e al. Imp o ing managemen and pa ien ca e in len igo maligna by mapping wi h in i o con ocal mic oscopy. JAMA De ma ol 2013; 149: 692–698. 24. Ca e a C, Puig S, Malhe y J. In i o con ocal e lec- ance mic oscopy in melanoma. De ma ol The 2012; 25: 410–422. 25. Alawi SA, Kuck M, Wa lich C, Ba z S, McKenzie G, Fluh JW, e al. Op ical cohe ence omog aphy o p esu gical ma gin assessmen o non-melanoma skin cance – a p ac- ical app oach. Exp De ma ol 2013; 22: 547–551. Ac a De m Vene eol 95