Inligting

Word Hematoxylin/Eosien-gekleurde weefsel gebruik vir kankertelling?

Word Hematoxylin/Eosien-gekleurde weefsel gebruik vir kankertelling?


We are searching data for your request:

Forums and discussions:
Manuals and reference books:
Data from registers:
Wait the end of the search in all databases.
Upon completion, a link will appear to access the found materials.

Ek is 'n ingenieurstudent wat vir die eerste keer op die gebied van borskanker werk. Ek probeer om borskanker te kry (byvoorbeeld Allred -telling). Ek het 2 soorte beelde teëgekom:

Eerste (laat wees A):

Tweede (laat wees B):

Beelde is afkomstig van die VitroVivo -webwerf.

My verwarring: Word beeld A gebruik vir borskanker?

Wat ek dink ek verstaan: In kankertelling (in bors), probeer ons 'n telling vind. Byvoorbeeld in beeld B het ons immunopositiewe (bruin) kerne en immunonegatiewe (blou) kerne, en die verhouding van bruin tot totaal is 'n aanduiding. Hierdie bruin kerne kan wees Ki67, HER2, Estrogeen, Progesteroon proteïen/antigeen. Maar in H & E -bevlekte beelde word kerne van sitoplasma geskei. Daar is geen positiewe/negatiewe kerne nie, slegs een tipe kern. Is daar 'n manier waarop H & E -gekleurde weefsels gebruik word om punte te kry?


Immunohistochemie en immunocytochemie

Teenvlekke vir ensiem/chromogeen immuunvlek

Hematoksilien is waarskynlik die algemeenste kern teenvlek wat gebruik word wanneer 'n ensiem/chromogeen opsporingstelsel gebruik word. Daar is verskillende formulerings beskikbaar, geklassifiseer volgens die tipe middel wat gebruik word en of dit progressief of regressief is. Almal van hulle gee uiteindelik selkerne 'n aangename blou kleur van verskillende kleure en intensiteit, afhangende van die tipe hematoksilien wat gebruik word.

Hematoksielien alleen (of meer akkuraat sy oksidasieproduk, hematien) is anionies en het dus nie veel affiniteit vir DNA nie. Mordante is ystersoute, naamlik dié van yster, aluminium, wolfram en lood. Mordante kombineer met hematien, wat 'n positief gelaaide kleurstof -mordantkompleks tot gevolg het, sodat dit aan anioniese chromatien kan bind. Aluin (aluminium besmeer) hematoksiliene kan progressief of regressief gebruik word. Met progressiewe hematoksiliene (soos Mayer's, Carazzi's en Gill's), word weefsel of selle in hematoksilien geïnkubeer totdat die verlangde graad van kernkleuring bereik is, voordat dit blou word. In vergelyking, in die geval van regressiewe hematoksilien (soos Harris's), word weefsels of selle geïnkubeer totdat 'n mate van oorkleuring bereik word, voordat van die oortollige hematoksilien verwyder word deur onderdompeling in 'n suuroplossing, soos 1% suur alkohol. Hierdie proses staan ​​bekend as differensiasie. Progressiewe hematoksiliene is dus geriefliker om te gebruik as regressief, as gevolg van die afwesigheid van 'n differensiasiestap en die gevolglike verenigbaarheid met alkoholoplosbare ensiem/substraat eindprodukte, soos dié wat deur HRP en AEC geproduseer word.

Of dit nou progressief of regressief is, sodra die verlangde vlak van kernkleuring bereik is, is hematoksiliene "blou". By suur pH kleur hematoksiliene die kerne rooi. Sodra dit egter aan 'n alkaliese omgewing blootgestel word, verander hematoksilien 'n aangename blou kleur. Lopende kraanwater word gereeld vir hierdie doel gebruik, aangesien dit voldoende alkaliniteit het, veral in gebiede met 'harde' water. In gebiede met 'sagte' water kan 'n geskikte alkaliese oplossing gebruik word om hematoksielien te blou, soos 0,05% (v/v) ammoniak.

Ander algemeen gebruikte tinktorale kern teenvlekke is onderskeidelik liggroen, vinnig rooi, toluidienblou en metileenblou vlekke, hetsy groen, rooi of blou.

Een belangrike oorweging by die gebruik van 'n nukleêre tinktoriese teenkleur is om nie die kleuring te intens te maak as jy 'n kernantigeen demonstreer nie, aangesien die teenkleur moontlik die positiewe sein van die opsporingstelsel kan masker.


Hematoksielien en Eosien kleuring van weefsel en selafdelings

Hematoksilien- en eosien- (H&E)-vlekke word al vir minstens 'n eeu gebruik en is steeds noodsaaklik vir die herkenning van verskeie weefseltipes en die morfologiese veranderinge wat die basis vorm van kontemporêre kankerdiagnose. Die vlek is jare lank onveranderd omdat dit goed werk met 'n verskeidenheid fixeermiddels en 'n wye verskeidenheid sitoplasmiese, kern- en ekstrasellulêre matrikskenmerke vertoon. Hematoksielien het 'n diep blou-pers kleur en vlek nukleïensure deur 'n komplekse, onvolledig verstaanbare reaksie. Eosien is pienk en vlek proteïene nie-spesifiek. In 'n tipiese weefsel word kerne blou gekleur, terwyl die sitoplasma en ekstrasellulêre matriks verskillende grade van pienk kleuring het. Goed vasgemaakte selle toon aansienlike intranukleêre besonderhede. Kerne toon verskillende seltipe- en kankersoortspesifieke patrone van kondensasie van heterochromatien (hematoksielienkleuring) wat diagnosties baie belangrik is. Nucleoli -vlek met eosien. As daar oorvloedige poliribosome teenwoordig is, sal die sitoplasma 'n duidelike blou glans hê. Die Golgi -sone kan voorlopig geïdentifiseer word deur die afwesigheid van vlekke in 'n gebied langs die kern. Die vlek onthul dus oorvloedige strukturele inligting, met spesifieke funksionele implikasies. 'N Beperking van hematoksielienkleuring is dat dit onversoenbaar is met immunofluoressensie. Dit is egter handig om een ​​seriële paraffiensnit te vlek uit 'n weefsel waarin immunofluorescentie uitgevoer sal word. Hematoksilien, gewoonlik sonder eosien, is nuttig as 'n teenkleur vir baie immunohistochemiese of hibridisasieprosedures wat kolorimetriese substrate (soos alkaliese fosfatase of peroksidase) gebruik. Hierdie protokol beskryf H & ampE -kleuring van weefsel- en selafdelings.


Algemene oorsig

Kleuring van weefselafdelings met chemiese en biologiese kleurstowwe word al vir meer as 'n eeu gebruik om verskeie weefseltipes en morfologiese veranderinge wat met kontemporêre kankerdiagnose geassosieer word, te visualiseer. Die vlekprosedure is egter arbeidsintensief, benodig opgeleide tegnici, duur en lei dikwels tot die verlies van onvervangbare monsters en vertraagde diagnose. In samewerking met Brigham en Women's Hospital (Boston, MA) beskryf ons 'n 'berekeningskleur' ​​-benadering om foto's van onbevlekte weefselbiopsies digitaal met Haematoxylin en Eosin (H & ampE) kleurstowwe digitaal te vlek om kanker te diagnoseer.

Ons metode gebruik neurale netwerke om vinnig foto's van nie-bevlekte weefsels te vlek, wat dokters betyds inligting gee oor die anatomie en struktuur van die weefsel. Ons rapporteer ook 'n "berekenings-ontkleuring"-algoritme wat kleurstowwe en vlekke van foto's van voorheen gekleurde weefsels kan verwyder, wat hergebruik van pasiëntmonsters moontlik maak.

Hierdie metodes en neurale netwerke help dokters en pasiënte deur nuwe berekeningsprosesse op die punt van sorg, wat naatloos in kliniese werkstrome in hospitale oor die hele wêreld kan integreer.


Resultate en bespreking

Doeltreffendheid van iHE

Ons het 'n eenvoudige maar effektiewe kleursisteem (Fig. 1A) ingestel. Die weefsels is in 'n kleurstofoplossing gedompel in 'n sentrifugebuis wat in 'n houer van vlekvrye staal vasgemaak is. 'N Ultrasoniese transducer is aan die onderkant van 'n houer van vlekvrye staal en 'n verkoeler vasgemaak, wat gebruik is om die ultrasoniese transducer af te koel. Die vlekvrye staalhouer is met water gevul om beter ultraklankoordrag te verseker. Die maksimum elektriese krag van die ultrasoniese transducer was 60 watt. Tydens die proses om die weefsel met hematoksilien te kleur of te spoel, het ons die insetspanning van die ultrasoniese kragtoevoer aangepas en die akoestiese kragdigtheid in die vlekvrye staalhouer op 1.2 beheer.

1,5 W/cm 2 . Tydens weefselkleuring met eosien was die akoestiese kragdigtheid 0,8

Oor die algemeen is dit moeilik om ongeskonde weefsel eenvormig te vlek (Fig. 1C). Hier het ons twee maniere gebruik om verf te verbeter - DCM delipidasie en ultraklank. DCM is 'n oplosmiddel wat die lipiede in die selmembraan maklik kan oplos 5,11,15,31. 'N Vergelyking van Fig. 1C-I met Fig. 1C-III toon aan dat die brein wat met DCM-delipidasie behandel word, dieper gekleur kan word, waarskynlik omdat die brein poreus word en kleurverspreiding geneig is om te verbeter na delipidasie. 'N Vergelyking van Fig. 1C-III met Fig. 1C-IV toon aan dat die brein wat met ultraklank bevlek is, in 'n korter tyd meer eenvormige kleuring gehad het. In figuur 1C-II toon die beeld dat die breinweefsel wat met DCM-delipidasie behandel is en vir 6 uur onder ultraklank gekleur is, 'n eenvormige hematoksielienkleuring bied. Met ander woorde, delipidasie en ultraklank was onontbeerlik vir iHE, en albei het bygedra tot die kleurresultate. Deur Fig. 1C-II met Fig. 1C-III te vergelyk, het ons ook gevind dat encephalocele van die brein met ultraklank en delipidasie groter was as dié van die brein sonder ultraklank en delipidasie, moontlik as gevolg van die krimping van die brein tydens dehidrasie en delipidasie.

Nog 'n effek van ultraklank was versnelde spoel (Fig. 1B). Volwasse muisbreine met delipidasie is eers vir 6 uur deur hematoksielien gekleur en daarna in die middellyn gesny. Die een helfte is onder ultraklank gespoel, terwyl die ander helfte sonder ultraklank gespoel is. Die spoeltemperatuur was 50 °C. Om die absorpsie van die ultraklank spoeloplossing en statiese spoeloplossing (kontrole) te verkry, het ons die boonste oplossing oorgedra en die absorpsie gemeet met behulp van die Lambda 950 UV/VIS-spektrometer by 'n golflengte van 445 nm. Met verloop van tyd het die absorpsie van die ultraklankspoeloplossing baie hoër geword as die van die statiese spoeloplossing. Dus het die brein onder ultraklank meer kleurstowwe vrygestel as in die statiese toestand.

Vergelyking van iHE met tradisionele H&E-kleuring

Ultraklank kan lei tot weefsel- en selontwrigting, dus is 'n nie-verwaarloosbare probleem of die selstruktuur vernietig word na iHE. Soos getoon in Fig. 2, het ons gevind dat die selstruktuur behoue ​​gebly het sonder beduidende vervorming in vergelyking met die van tradisionele H & ampE -kleuring. Die kleureffekte van iHE en tradisionele H & ampE -kleuring is soortgelyk, wat daarop dui dat iHE 'n haalbare metode is. Die perisomatiese senuweevesels wat in figuur 2D getoon word, is minder as dié in figuur 2A, waarskynlik omdat die sitomembraan gedeeltelik opgelos is na delipidasie. Om die opoffering van subsellulêre struktuur verder te evalueer, het ons 100 neuronkerne van die korteks vergelyk met behulp van die H&E- en iHE-metodes, maar ons het geen beduidende verskil tussen hierdie twee metodes gevind nie (data nie getoon nie).

Vergelyking van iHE en tradisionele H & ampE -kleuring. (A – C), Beelde van 7-mikrometer dik muisbreinskywe wat met die tradisionele H & ampE-metode gekleur is. (D–F), Beelde van ongeskonde muisbreinweefsel na iHE-kleuring, sny en 2D-beelding. Die rooi stof wat verskyn in (A

F) verteenwoordig bloedselle wat nie heeltemal skoongemaak is tydens kardiale perfusie nie. Objektieflens, 20× N.A., 0,75 werkafstand, 1 mm. Skaalstaaf: 50 μm.

Ongeskonde muisbrein gekleur met iHE

'N Ongeskonde C57BL/6 -muisbrein is bevlek met iHE, en die beelde in Figuur 3 toon aan dat die brein eenvormig gekleur is. Die hippokampus was onderskeibaar, en die sellulêre kerne morfologie was duidelik (Fig. 3D-F). So kan ons H & ampE -kleuringinligting verkry vir 'n muisbrein op grond van die natuurlike ruimtelike konteks daarvan. Die enkefalokoele van die brein was duidelik (Fig. 3C).

Snye uit 'n ongeskonde muisbrein wat bevlek is met iHE, gevolg deur 2D -beeldvorming. (A) 3D -projeksie van 20 snye in (B). (B) Die koronale vlak sny uit die C57BL/6 -muisbrein na iHE. Objektiewe lens, 4 × en 20 × N.A., 0,2 en 0,75. Die brein is 8 μm by die kroonvlak gesny, en ons het een sny vir elke 400 μm van die reukbol tot by die epencephalon gekies. (DF) Vergroting van (C).

Ander ongeskonde muisweefsels gekleur met iHE

Om 'n lewermetastase-model te genereer, het ons 4T-1 melkkankerselle in 7-week-oue BALB/c-muise ingespuit, die muise vir 40 dae grootgemaak en die muise op 3 maande oud perfuseer. Vervolgens is die lewer gekleur, en snye is voorberei vir daaropvolgende beeldvorming met behulp van Nikon NIS-Elements-mikroskopie, die resultate word in Fig. 4A-C getoon. In die prentjie word die swart stippellyn gebruik om die gewasarea en die normale gebied te verdeel, en die resultate toon dat die kern-sitoplasmiese verhouding van die gewasarea anders was as dié van die normale gebied. Die muislong, nier, maag, voorpoot, hart en oogballe is gekleur met iHE (Figuur 4D–G en S4A–H). Die pulmonale lob en alveoli, klassieke longstrukture, kan in Fig. 4D,E waargeneem word. Nierbuisies word duidelik in figuur 4F, G voorgestel. Die muismaag is bevlek, en die villus by die pylorus word in aanvullende figuur 4G, H getoon. Die klassieke spierstruktuur in die muisvoorpoot is na iHE geopenbaar, en die velmikrostruktuur is ook waargeneem (Aanvullende Fig. S4A,B). Die hart is 'n orgaan wat uit kardiomiosiete bestaan, en die tipe kamer en miokardiale selle is duidelik waargeneem (aanvullende figuur S4C, D). Ons het ook iHE toegepas op muis oogballe, en die sellulêre stratifikasie van die retina kon waargeneem word (Aanvullende Fig. S4E, F).

Beelde van ander muisweefsels wat met iHE gevlek is. (AC) Beelde van muislewer met 'n gewas. (D,E) iHE vir muislong. (F,G) iHE vir muisnier. Objektiewe lens, 20 × N.A., 0,75 werkafstand, 1 mm.

Verenigbaarheid van iHE met bloedvlekke

Kankerselgroei word geassosieer met beduidende veranderinge in bloedvate 30 . Soos hierbo bespreek, is H & ampE -kleuring 'n klassieke manier om die besonderhede van 'n tumor aan te bied. Hier het ons H & ampE -kleuring gekombineer met bloedvatkleuring (Fig. 5) om gelyktydig inligting oor bloedvate en sellulêre kerne waar te neem. Hierdie dubbelkleuringsmetode het die potensiaal om betekenisvolle inligting oor tumorstatus te verskaf.

Die muisbrein is met koolstofink geperfeer voor iHE. (A,B) Hippocampus, (B) is die vergroting van die boks in (A). (C,D) Thalami, (D) is die vergroting van die boks in (C). Objektieflens, 20× N.A., 0,75 werkafstand, 1 mm.

Die beginsel van iHE word getoon in Fig. 6. Na DCM delipidasie word die sitomembraan gedeeltelik opgelos en word die weefsel poreus 25. Ons neem aan dat die ultraklank stabiele kavitasie veroorsaak (ultraklank energiedigtheid en lt10 Watt/cm 2). Die porieë van die weefsel kan hul deursnee in die ultraklankveld verander omdat ultraklank 'n longitudinale golf is. Net soos 'n veer wat strek en saamdruk as gevolg van 'n eksterne krag, word die weefsel op die mikroskaalvlak gestrek en saamgepers. So word die porieë in die golfkam van die ultraklank gestrek, terwyl die porieë in die golftrog 26,32,33 saamgepers word. Intussen is daar met die ultrasoniese golf periodieke kompressie en skaarsheid in die vloeistof. Massa -oordrag in en uit die weefsel word dus vergemaklik 34. Daarom word die diffusie wat deur die konsentrasieverskil in die deeltjies aangedryf word in die ultrasoniese veld versnel, en die kleurstowwe diffundeer vinniger in weefsel in.

Moontlike beginsel van iHE. 'n Vaste weefsel word as die oorspronklike weefsel gebruik. Na dehidrasie word die weefsel aan delipidasie onderwerp, gevolg deur die H&E-kleuringsprosedure is die H&E-kleuringsproses vereenvoudig sodat die leser die beginsel vir hoe ultraklank by die proses betrokke is beter kan verstaan. Dichlorometaan kan 'n poreuse toestand van die selmembraan veroorsaak, gelykstaande aan die poreuse toestand van die weefsel. Die porieë van die weefsel verander in die ultrasoniese veld omdat ultraklank 'n longitudinale golf is. So word die porieë in die golfkam van die ultraklank gestrek, terwyl die porieë in die golftrog saamgepers word. Intussen word die ewekansige beweging van deeltjies in die ultraklankveld verbeter. Die verspreidingsproses word verbeter en verkry vinnige en eenvormige kleuring.

Oor die algemeen skep die iHE -metode poreuse weefsel vir vlekke en verhoog dit ewekansige beweging. Daarom word die verspreidingsproses verbeter en word 'n vinnige en eenvormige H & ampE -kleuring verkry. Ons het 'n stel metodes (iHE) vir die verkryging van H & ampE -kleuringinligting van ongeskonde weefsel vir volume beeldvorming opgestel. Ultraklank is gebruik om die kleuring van weefsel in die hele berg te vergemaklik, en die vermoë om spoel te verbeter, is kwantitatief geëvalueer. As gevolg hiervan kan hierdie metode die invallingsfase van die gewas klassifiseer op grond van die natuurlike 3D -grens daarvan en die tumormetastase beter toelig op groot skaal op grond van 'n muismodel. iHE was ook versoenbaar met bloedvatkleuring, wat impliseer dat ons H&E-kleuring en bloedvatinligting vir 'n enkele weefsel gelyktydig kan verkry. Ons resultate bied dus 'n nuwe hulpmiddel om tumormetastase en infiltrasie in 'n muismodel te bestudeer. Verder toon ultraklank belofte in die fasilitering van ander tipes biologiese weefselverwerking, soos optiese skoonmaak, immunolabeling en chemiese kleuring van ongeskonde weefsels.


3 RESULTATE

3.1 Wndchrm-gebaseerde ontleding van morfologie in nie -kankeragtige en maagkankerweefsels

Om die biologiese morfologie van sel- en weefseltoestande kwantitatief te evalueer, het ons 'n masjienleeranalise uitgevoer deur die wndchrm algoritme, en spesifieke beeldmetings (Figuur 1A). Beelddatastelle is opgestel in ooreenstemming met patologiese diagnose, met behulp van weefselmikroarrays wat afkomstig is van adenokarsinoompasiënte by die mens. Vier-en-vyftig H & E-gekleurde weefselbeelde van 1360 × 1024 pixels is vir elke klas versamel: nie-kanker, graad 1 (goed gedifferensieerd), graad 2 (matig gedifferensieerd) en graad 3 (swak gedifferensieerd) (figuur 1B, figuur S1A en tabel S1). Kortliks, wndchrm beeldkenmerke uit alle beelde van elke gedefinieerde klas onttrek, en 'n klassifiseerder opgelei om tussen die klasse te onderskei deur opleidingsdatastelle te gebruik. Die klassifikasieprestasie is dan bekragtig met toetsbeelde wat lukraak gekies is, waar hierdie stappe outomaties uitgevoer is. Ons het 20 kruisvalidasie-ontledings onder die nie-kanker en graad 1-3 van maagkanker uitgevoer (Figuur 1A, links). As 'n eerste stap van die analise het ons die optimale aantal beelde ondersoek wat nodig is vir 'n doeltreffende klassifikasie. Die resultate het getoon dat die waarde van klassifikasie-akkuraatheid (CA) verbeter het met toenemende aantal opleidingsbeelde (Figuur S1B), terwyl dié van standaardfoute kleiner geword het soos dikwels gesien in masjienleerontledings. 31 Die beste klassifikasie is gevind met 54 opleidingsbeelde by CA 0.78 (die maksimum CA is moontlik 1.0), en hierdie CA waarde was merkbaar hoër as ewekansige klassifikasie by CA 0.25. Verder is die relatiewe ooreenkomste tussen die klasse gevisualiseer met dendrogramme (Figuur 1C). Daarbenewens, met behulp van 20 beelde in elke klasse, het ons die klassifikasie-ooreenkoms tussen niekanker, chroniese gastritis en grade 1-3 (Figuur S1C) bevestig. Prestasie van die klassifikasietoets was voldoende, aangesien die spesifisiteit en sensitiwiteit daarvan om kankergrade van nie-kankeragtige weefsels te onderskei onderskeidelik 100% en 92% was (Tabel S2, boonste). Vir addisionele assessering van morfologiese kenmerke, het ons elke klas in twee subklasse verdeel, en die mate van verskille van grade 1-3 van nie-kankeragtige weefsels gemeet, soos aangedui deur morfologiese afstand (MD) (Figuur 1D). Die MD van Non-cancer_1 toon ooreenkoms met Noncancer l_2 en verskil van kankerweefsels van drie grade. Verder, toe die beelde digitaal geteël is (sien Metodes), het die getalle opleidingsbeelde toegeneem, maar die oorsig van die weefsels het verlore gegaan. CA-waardes was egter grootliks onveranderd op 0.79-0.69 met behulp van hierdie geteëlde beelde (Figuur S1D), wat daarop dui dat plaaslike morfologie sowel as histologiese oorsig aanwysers is om tussen niekanker- en kankerweefsels te onderskei.

Ons het toe gedetailleerde binêre vergelykings tussen nie-kankermonsters en elke graad van maagkanker uitgevoer om die doeltreffendheid van wndchrm gebruik elkeen van die 54 beelde wat voldoende CA -waardes getoon het (Figuur S1E). ROC -krommeanalises het die akkuraatheid van klassifikasies geverifieer, want AUC's was onderskeidelik 0.99, 0.98 en 0.99 vir Noncancer versus Grade 1, 2 en 3 (die maksimum AUC is 1.0, in teenstelling met ewekansige toewysing van 0.5) (Figuur 1E). Verteenwoordigende lyste van insiggewende beeldkenmerke in elke klassifikasietoets is aangedui volgens die relatiewe Fisher -diskriminasie tellings (Figuur 1F). Baie stelle beeldkenmerke is algemeen gebruik om graad 1, 2 en 3 van nie-kanker te onderskei (r > 0.7), alhoewel 'n paar duidelike kenmerke ook betrokke was (data nie getoon nie). Ons resultate het dit getoon wndchrm ontledings het die mensgebaseerde patologiese ondersoeke van H&E-beelde van kankerweefsels hoogs opgesom.

3.2 Wndchrm-gebaseerde analise onthul informatiewe kenmerke van H & E-bevlekte beelde

Om te verstaan ​​watter morfologiese kenmerke bydra tot die klassifikasie van nie -kanker- en kankergrade, het ons die H&E RGB (rooi, groen, blou) beelde digitaal in grys skale in hematoxylin- en eosinkanale ontbind (Figuur 2A). 30 Sellulêre kerne en sitoplasmiese komponente word gewoonlik met onderskeidelik hematoksilien en eosien gekleur. 4, 10 Met behulp van die ontbindde beelde vir nie-kanker en grade 1-3, soos getoon in figuur 2B en figuur S2, het ons CA gemeet onder nie-kanker en grade 1-3 (figuur 2C). Kruisvalideringstoetse van hematoksilien- en eosienbeelde het ekwivalente CA-waardes (onderskeidelik 0.72 en 0.69) aangedui. Sensitiwiteit en spesifisiteit was ewe hoog op 82%-98% (Tabel S2, tweede van bo), wat daarop dui dat hematoksielien- en eosienbeelde morfologiese kenmerke bevat wat onderskei tussen kanker en nie -kankeragtige weefsels.

'N Tipiese lys van insiggewende beeldkenmerke in die klassifikasietoets is opgestel volgens die relatiewe Fisher -diskriminasie tellings, en het algehele ooreenkomste getoon (Figuur 2D). Pearson-korrelasiekoëffisiëntwaarde was swak tussen hematoksilien en eosienbeelde (r = 0.55), wat die teenwoordigheid van unieke morfologiese kenmerke in enige beeld voorstel. Konsekwent het dokters van Non-cancer_1 in beide hematoksielien- en eosienbeelde in dieselfde mate ongelykheid getoon tussen nie-kankeragtige en kankerweefsels (Figuur 2E, F). wndchrm ontledings het die teenwoordigheid van insiggewende kenmerke in hematoksilien en eosien-gekleurde beelde van kankerweefsel geïmpliseer.

3.3 Karakterisering van kernmorfologie in maagkankerweefsels

Ons klassifikasie-analise van hematoksielienbevlekte beelde dui aan dat kernmorfologieë nie-kanker- en maagkanker (grade 1-3) onderskei word, soos getoon deur CA-waardes (Figuur 2C). Om kernmorfologie te bepaal, het ons twee kenmerke van die kerngebied en totale intensiteit gemeet (Figuur 1A, reg). Met behulp van 'n beeldanalise sagteware (Cellomics CellInsight) is elke meetgebied met 'n vaste grootte van 1024 × 1024 pixels van oorspronklike weefselbeelde opgespoor (Figuur 3A). Deur >12 000 kerne te tel, het ons gevind dat kernarea aansienlik groter was in kankerweefsels, in vergelyking met nie-kankeragtige weefsels (Figuur 3B), en dat seinintensiteit ook hoër was in kankerselle (Figuur 3C). Omdat kerne dig versprei is en soms in kankerweefsels oorvleuel, waarskynlik as gevolg van aktiwiteite met 'n hoë groei, het ons daarna probeer om hierdie kenmerk te meet deur gebruik te maak van die kerngebied wat deurlopend bevlek is met hematoksielien. Ons stel die sagteware in om die hematoksielien-positiewe oppervlakte wat groter was as die gedefinieerde drempel (13 200 pixels) te herken, soos getoon in Figuur 3D. Die gebied met gegroepeerde kerne was prominent in graad 1 en 2 van maagkanker, maar skaars in graad 3 (figuur 3E, F). Opsommende statistieke vir die area en totale intensiteit word in Tabel S3 getoon, wat aandui dat kernmorfologie 'n voordelige parameter vir kankerklassifikasie is.

3.4 Uitdrukkingsvlakke van kern ATF7IP/MCAF1 is gekorreleer met H & E -beelde

Daar is berig dat verskillende kernfaktore 37, 38 en membraan/oplosbare faktore 39, 40 by morfologie van selle en weefsels betrokke is. Ons ondersoek vervolgens biologiese skakels tussen molekulêre uitdrukking en morfologiese kenmerke in maagkankerweefsels, met behulp van molekulêre merkergebaseerde analise of feitgedrewe analise.

Om te ondersoek hoe H & E-beelde geklassifiseer kan word op grond van molekulêre uitdrukking, het ons twee kankerverwante proteïene gekies: kern ATF7IP/MCAF1 en membraan PD-L1 (Figuur 4 en 5). ATF7IP/MCAF1 is 'n epigenetiese faktor wat betrokke is by die vorming van heterochromatien en die regulering van geen, wat gereeld ooruitgedruk word in verskillende soorte gewasse, insluitend maagkanker. ATF7IP/MCAF1 funksioneer vir óf DNA-metilering-gebaseerde geenonderdrukking óf die transkripsiefaktor Sp1-gemedieerde geenaktivering. 36 Aan die ander kant word PD-L1 oor die algemeen deur kankerselle geproduseer om immuuntoesig te ontsnap, en is dit 'n molekulêre teiken vir kanker-immuunterapie. 41-43 Vorige verslag het getoon dat die PD-L1geenpromotor word gereguleer deur DNA -metilering of Sp1 -binding in kankerselle. 44, 45 Daar is die moontlikheid dat ATF7IP/MCAF1 kan beheer PD-L1 uitdrukking via Sp1, soos aangedui deur gepubliseerde ChIP-seq data van kolonkanker (Figuur S3A).

Ons het beide H&E-kleuring en IHC uitgevoer met behulp van reeksafdelings van weefsel (Tabel S4). Nadat die snysnitte uit 'n paraffienblok gemaak en gekleur is, het ons die H&E versigtig met IHC-beelde met die hand in lyn gebring (Figuur S3B-D). Ons het 32 ​​werwe gekies uit H&E-beelde (elk 1360 × 1024 piksels) en die ooreenstemmende IHC-beelde vir ATF7IP/MCAF1-uitdrukking. Maagkankerweefsel en aangrensende nie -kankeragtige streke toon onderskeidelik hoë en lae uitdrukking van ATF7IP/MCAF1 (Figuur 4A, Figuur S3E, F). Die vlakke van IHC -seine is bevestig deur die kwantifisering van hul seine (Figuur 4B). Gebaseer op die uitdrukkingsvlakke van ATF7IP/MCAF1, het ons dan H&E-beelde geklassifiseer deur gebruik te maak van wndchrm tot lae en hoë uitdrukking van hierdie proteïen (CA 0,95-1,00, wat hoë akkuraatheid toon, ongeag beeldnommers) (Figuur S3G), wat daarop dui dat maagkankerweefsels soos getoets duidelik in hierdie twee klasse verdeel kan word. Daarbenewens was sensitiwiteit en spesifisiteit van ATF7IP/MCAF1-seine onderskeidelik 100% en 98% (Tabel S2,tweede van laer). Om die CA tussen lae en hoë klasse ATF7IP/MCAF1 te evalueer, het ons subklasse in H & E -beelde gerangskik (Laag 1, Laag 2, Hoog 1 en Hoog 2). Lae 2 het ooreenkoms met laag 1, maar beduidende verskil met hoë 1 (figuur 4C). Boonop dui die belyning van relatiewe Fisher -tellings op 'n swak korrelasie tussen die twee vergelykings (Figuur 4D, r = 0.44), wat dui op die teenwoordigheid van funksieverskille. Boonop het elke MD van Low 1 in die funksie -ruimte en die dendrogram morfologiese verskille getoon tussen lae en hoë klasse ATF7IP/MCAF1 (Figuur 4E, F).

3.5 Uitdrukkingsvlakke van sitoplasmiese PD-L1 is gekorreleer met H&E beelde

Ons het verder ondersoek of die uitdrukking van die membraanagtige proteïen PD-L1 in kanker gekoppel is aan weefselmorfologie. Ons het weer H&E-kleuring en IHC met anti-PD-L1-teenliggaampies uitgevoer, deur gebruik te maak van reeksafdelings van weefselmikroskikkings in maagkankermonsters (Figuur 5A en Tabel S5). Ons het IHC seinvlakke van PD-L1-kleuring gekwantifiseer, gegroepeer in lae en hoë uitdrukking van hierdie proteïen, en datastelle van die ooreenstemmende H & E-beeld (1360 × 1024 pixels) (figuur 5B, figuur S4A, B) verder geskep. Verder het ons die tumor proporsie telling (TPS) geëvalueer deur positief gekleurde selle in 100 selle per beeld te tel en gevind dat PD-L1 High aansienlik hoër TPS getoon het, terwyl PD-L1 Low baie lae TPS gehad het (Figuur S4C). Die H&E beelde is geklassifiseer as Laag en Hoog PD-L1, by CA 0,86 met behulp van 60 beelde (Figuur S4D). Gevoeligheid en spesifisiteit van PD-L1 seine was onderskeidelik 88% en 84% (Tabel S2, laer). Ons het die morfologiese verskil tussen PD-L1 Lae en Hoë subklasse bevestig soos getoon deur die CA (Figuur 5C). Die relatiewe Fisher-diskriminante tellings van beeldkenmerke het die teenwoordigheid van kenmerke wat vir die ongelykheid verantwoordelik is, voorgestel (Figuur 5D,r = 0.59). Elke MD van laag 1 in die funksie ruimte en die dendrogram het morfologiese verskille getoon tussen lae en hoë klasse PD-L1 (figuur 5E, F).

Gesamentlik het hierdie resultate aangedui dat die uitdrukking van ATF7IP/MCAF1 en PD-L1 met weefselkenmerke gekorreleer is, wat daarop dui dat die ruimtelike voorkoms van die kankergeassosieerde proteïene morfologiese inligting van die patologiese weefsels weerspieël.


Klassifikasie van hematoksielien en met eosien bevlekte borskanker Histologie mikroskopiese beelde met behulp van oordragleer met EfficientNets

Borskanker is 'n dodelike siekte en is 'n hoofoorsaak van sterftes by vroue wêreldwyd. Die proses van diagnose gebaseer op biopsieweefsel is nie-privaat, tydrowend en vatbaar vir menslike foute, en daar kan konflik wees oor die finale diagnose as gevolg van wisselvalligheid tussen die waarnemers. Rekenaargesteunde diagnosestelsels is ontwerp en geïmplementeer om hierdie probleme te bekamp. Hierdie stelsels dra aansienlik by tot die verhoging van die doeltreffendheid en akkuraatheid en die vermindering van die koste van diagnose. Boonop moet hierdie stelsels beter presteer, sodat hul vasgestelde diagnose meer betroubaar kan wees. Hierdie navorsing ondersoek die toepassing van die EfficientNet-argitektuur vir die klassifikasie van hematoksielien- en eosienbevlekte borskankerhistologiebeelde wat deur die ICIAR2018-datastel verskaf word. Spesifiek, sewe EfficientNets is verfyn en geëvalueer op hul vermoë om beelde in vier klasse te klassifiseer: normale, benigne, in situ karsinoom, en indringende karsinoom. Boonop is twee standaard vleknormaliseringstegnieke, Reinhard en Macenko, waargeneem om die impak van vleknormalisering op prestasie te meet. Die uitkoms van hierdie benadering toon dat die EfficientNet-B2-model 'n akkuraatheid en sensitiwiteit van 98.33% opgelewer het deur Reinhard-vleknormaliseringsmetode op die oefenbeelde en 'n akkuraatheid en sensitiwiteit van 96.67% met die Macenko-vleknormaliseringsmetode opgelewer het. Hierdie bevredigende resultate dui aan dat die oordrag van generiese kenmerke van natuurlike beelde na mediese beelde deur middel van fyn afstelling op EfficientNets bevredigende resultate kan behaal.

1. Inleiding en agtergrond

Een van die grootste oorsake van sterftes by vroue regoor die wêreld is borskanker [1]. Dit word gedefinieer as 'n groep siektes waarin selle in die borsweefsel op 'n onbeheerde manier verander en verdeel, wat gewoonlik knoppe of groeisels tot gevolg het. Hierdie tipe kanker begin dikwels in die melkkliere of buise wat hierdie kliere met die tepel verbind. In die beginfases van die siekte is die klein gewas wat verskyn baie makliker om effektief te behandel, wat die vordering van die siekte voorkom en die morbiditeitsyfer verlaag, daarom is sifting van kardinale belang vir vroeë opsporing [2].

Die proses van borskankerdiagnose begin met palpasie, periodieke mammografie en ultrasoniese beeldinginspeksie. Die resultate van hierdie prosedures dui aan of verdere toetsing nodig is. Indien kanker by 'n pasiënt vermoed word, word 'n biopsie uitgevoer en weefsel vir mikroskopiese ontleding verkry sodat 'n patoloog 'n histologiese ondersoek van die onttrekte weefsel kan doen om die diagnose te bevestig [2, 3]. Sodra die biopsie voltooi is, word die weefsel in 'n laboratorium ontleed. Die weefselvoorbereidingsproses moet begin met formalienfiksasie en daarna in paraffienafdelings in te sluit. Die paraffienblokke word dan in skywe gesny en op glasskyfies vasgemaak. Ongelukkig is interessante strukture soos die sitoplasma en kerne in die weefsel nog nie duidelik nie. Die gebrek aan duidelikheid in die weefsel vereis dat die weefsel gekleur word sodat die strukture meer sigbaar kan word. Typically, a standard and well-known staining protocol, using hematoxylin and eosin, is applied. When added to the tissue, the hematoxylin can bind itself to deoxyribonucleic acid, which results in the nuclei in the tissue being dyed a blue/purple color. On the other hand, the eosin can bind itself to proteins, and, as a result, other relevant structures such as the stroma and cytoplasm are dyed a pink color. Traditionally, after staining, the glass slide is coverslipped and forwarded to a pathologist for examination [4]. Routinely, the expert gathers information on the texture, size, shape, organization, interactions, and spatial arrangements of the nuclei. Additionally, the variability within, density of, and overall structure of the tissue is analyzed. In particular, the information concerning the nuclei features is relevant for distinguishing between noncarcinoma and carcinoma cells. In contrast, the information concerning the tissue structure is relevant for distinguishing between in situ and invasive carcinoma cells [5].

The noncarcinoma class consists of normal tissue and benign lesions these tissues are nonmalignant and do not require immediate medical attention. In situ and invasive carcinoma, on the other hand, are malignant and become continuously more lethal without treatment. Spesifiek, in situ carcinoma refers to the presence of atypical cells that are confined to the layer of tissue in the breast from which it stemmed. Invasive carcinoma refers to the presence of atypical cells that invades the surrounding normal tissue, beyond the glands or ducts from where the cells originated [2]. Invasive carcinoma is complicated to treat, as it poses a risk to the entire body [3]. This threat means that the odds of surviving this level of cancer decreases as the progression stages increase. Moreover, without proper and adequate treatment, a patient’s in situ carcinoma tissue can develop into invasive carcinoma tissue. Therefore, it is of paramount importance that biopsy tissue is examined correctly and efficiently so that a diagnosis can be confirmed and, subsequently, treatment can begin. Examples of histology images belonging to each of these classes are shown in Figure 1.

The task of performing a practical examination on the tissue is not simple and straightforward. On the contrary, it is rather time-consuming and, above all, prone to human error. The average diagnostic accuracy between professionals is around 75% [6]. These issues can result in severe and fatal consequences for patients who are incorrectly diagnosed [7].

The advancement of image acquisition devices that create whole slide images (WSI) from scanning conventional glass slides has promoted digital pathology [8]. The field of digital pathology focuses on bringing improvement in accuracy and efficiency to the pathology practice [9] by associating histopathological analysis with the study of WSI [8].

An excellent solution to address the limitations of human diagnosis is computer-aided diagnosis (CAD) systems, which are developed to automatically analyze the WSI and provide a potential diagnosis based on the image. These systems currently contribute to improving efficiency and reducing both the cost of diagnosis and interobserver variability [5, 10]. Even though current CAD systems that operate at high sensitivity provide relatively good performance, they will remain a second-opinion clinical procedure until the performance is significantly improved [10].

Recently, deep learning approaches to the development of CAD systems have produced promising results. Previous attempts to classify breast cancer histology images using a combination of handcrafted feature extraction methods and traditional machine learning algorithms required additional knowledge and were time-consuming to develop. Conversely, deep learning methods automate this process. These systems allow pathologists to focus on difficult diagnosis cases [11].

Hence, to ensure early diagnosis in breast cancer candidates, increase treatment success, and lower mortality rates, early detection is imperative. Although the advent of and advancements in computer-aided systems have benefited the medical field, there is plenty of room for improvement.

1.1. Research Problem

In general, the shortage of available medical experts [12], the time-consuming quest to reach a final decision on a diagnosis, and the issue of interobserver variability justify the need for a system that can automatically and accurately classify breast cancer histopathology images. Previous approaches to this problem have been relatively successful considering the available data and return adequate classification accuracies but tend to be computationally expensive. Thus, this work will explore the use of seven lightweight architectures within the EfficientNet family [13]. Since the EfficientNet models were designed to optimize available resources, while maintaining high accuracies, a CAD system that performs at the level of the current state-of-the-art deep learning approaches, while consuming less space and training time, is desirable. Transfer learning techniques have become a popular addition to deep learning solutions for classification tasks. In particular, many state-of-the-art approaches utilize fine-tuning to enhance performance [14]. Therefore, this research explores the application of seven pretrained EfficientNets for the classification of breast cancer histology images. Furthermore, the addition of stain normalization to the preprocessing step will be evaluated. Hence, the primary question that this research will answer is, “Can fine-tuned EfficientNets achieve similar results to current state-of-the-art approaches for the application of classifying breast cancer histology images?”

1.2. Research Contributions

In this research, the application of seven versions of EfficientNets with transfer learning for breast cancer histology image classification is investigated. The proposed architecture was able to effectively extract and learn the global features in an image, such as the tissue and nuclei organization. Of the seven models tested, the EfficientNet-B2 architecture produced superior results with an accuracy of 98.33% and sensitivity of 98.44%.

The key takeaway from this investigation is that the simple and straightforward approach to using EfficientNets for the classification of breast cancer histology images reduces training time while maintaining similar accuracies to previously proposed computationally expensive approaches.

1.3. Paper Structure

The remainder of the paper is structured as follows: Section 2, the literature review, provides details on previous successful approaches. Section 3, the methods and techniques, provides insight into the framework followed in this study. Section 4, the results, provides details of the results that were obtained during the research. Finally, section 5 elaborates on the insights of this work and concludes the paper.

2. Literature Review

Currently, computer-aided diagnosis (CAD) systems occupy the position of aiding physicians during the process of diagnosis, by easing their workload and reducing the disagreement that stems from the subjective interpretation of pathologists. However, the performance of these systems must be enhanced before they can be considered more dependable than a second-opinion system [10].

2.1. Traditional Approaches

In the traditional approach, expert domain knowledge is required so that the correct features may be handcrafted this is a time-consuming endeavor. Nevertheless, the approach yields acceptable results on the datasets used. For instance, Kowal [15] used multiple clustering algorithms to achieve nuclei segmentation on microscopic images. Segmentation made it possible to extract microscopic, textural, and topological features so that classifiers could be trained and images could be classified as either benign or malignant. The accuracy of patient-wise classification was in the range of 96–100%. It is worth noting that this method performs poorly when an image contains overlapping nuclei or a small number of nuclei. In this case, either the approach fails to identify the nuclei or the clustering algorithms could return unreliable results. Therefore, in order to attain an acceptable detection accuracy, a large number of sample images are required. Hence, it is evident that accurate nuclei segmentation is not a straightforward task this can also be attributed to the variability in tissue appearance or the presence of clustered or tightly clumped nuclei [5].

An alternative approach is utilizing information on tissue organization as in the work by Belsare et al. [16], which presents a framework to classify images into malignant and nonmalignant. Firstly, segmentation was done using spatio-color-texture graphs. After that, statistical feature analysis was employed, and classification was achieved with a linear discriminant classifier. The choice of this classifier considerably impacted the outcome of this approach, as the result outperformed the use of k-nearest-neighbor and state vector machine classifiers, especially for the detection of nonmalignant tissue. Accuracies of 100% and 80% were achieved for nonmalignant and malignant images, respectively.

2.2. Deep Learning Approaches

The increase in the availability of computing power has led to the emergence of advanced architectures called convolutional neural networks (CNNs). Contrary to the conventional approach, no expert domain knowledge is required to define algorithms for segmentation, feature extraction, and classification, but instead expert knowledge is needed to annotate the dataset for a CNN to achieve superior results. Instead, these networks can automatically determine and extract discriminative features in an image that contribute to the classification of the image. Generally, a CNN will use a training set of images to learn features that are unique to each class so that when a similar feature is detected in an unseen image, the network will be able to assign the image to a class with confidence.

2.2.1. Convolutional Neural Network Approaches

The success of convolutional neural networks (CNNs) with general computer vision tasks motivated researchers to employ these models for classifying histopathology images. For the classification of hematoxylin and eosin-stained breast cancer histology images, both Araújo et al. [5] and Vo et al. [17] used the Bioimaging 2015 dataset [18] and classified the images into four classes (normal, benign, in situ, and invasive) and two groups (carcinoma and noncarcinoma). The former work [5] proposes a CNN that can integrate information from multiple histological scales. The process begins with stain normalization via the method proposed by Macenko et al. [19] in a bid to correct color discrepancies. After that, 12

overlapping patches were extracted from each image. The chosen size of the patches ensures that no relevant information is lost during extraction and, therefore, every patch can be appropriately labeled. Then, data augmentation was used to increase the number of images in the dataset. Finally, a patch-wise trained CNN and a fusion of a CNN and support vector machine classifier (CNN + SVM) were used to determine the patch class probability. Image-wise classification was attained through a patch probability fusion method. The evaluation showed that using majority voting strategy as the fusion method produced the best results. Considering all four classes, patch-wise classification with the CNN achieved an accuracy rate of 66.7%, while the CNN + SVM achieved an accuracy rate of 65%. The image-wise classification achieved higher results at 77.8% accuracy for both classifiers. With only two classes, patch-wise accuracy for the CNN was 77.6%, and for the CNN + SVM, the approach yielded 76.9% accuracy. The image-wise classification for the 2-class task produced the best results at 80.6% for the CNN and 83.3% for the CNN + SVM. The reason for the lower patch-wise classification is that images may contain sections of normal-looking tissue. Since during patch generation the extracted patches inherit the image’s label, this may confuse the CNN. The increase in image-wise classification accuracy is due to the fusion method that is applied. The authors also recorded the sensitivity rates for each of the classes. It is worth noting that overall, for image-wise classification, the approach was more sensitive to the carcinoma class than the noncarcinoma class. This outcome, although not ideal, is preferable since the architecture that was proposed focuses on correctly classifying the carcinoma (malignant) instances [5].

The approach taken by Vo and Nguyen [17] proposed a combination of an ensemble of deep CNNs and gradient boosting tree classifiers (GBTCs). Stain normalization via Macenko et al. [19] and data augmentation were the initial steps of the process. Unlike the standard data augmentation method of rotating and flipping images, the proposed method [17] incorporates reflection, translation, and random cropping of the images. The normalized and augmented data was then used to train the proposed architecture. Specifically, three deep CNNs (Inception-ResNet-v2) were trained using three different input sizes:

. Then, visual features were extracted and fed into GBTCs, which increased classification performance. The majority voting strategy was used to merge the outputs of the GBTCs, resulting in a much more robust solution. Recognition rates of 96.4% for the 4-class classification and 99.5% for the 2-class classification were reported. This result surpasses state-of-the-art achievements. An interesting note is that the authors added global average pooling layers in place of dense (fully connected) layers, and this did not negatively impact the accuracy of the ensemble. Similar to Araújo et al. [5], the authors of this work recorded the sensitivities of their approach. The results indicate that, for the 4-class task, the proposed method struggles with the classification of the in situ instances, while the other three classes have incredibly high sensitivities. For the 2-class task, the approach yields a 100% sensitivity on carcinoma instances and 98.9% on noncarcinoma instances. These results indicate that the approach was able to successfully learn both local and global features for the multiclass and binary classification. However, the downfall of this approach is the computational expense.

2.2.2. Convolutional Neural Network with Transfer Learning Approaches

For the TK-AlexNet proposed by Nawaz et al. [3] to classify breast cancer histology images, the classification layers of the AlexNet architecture were replaced with a single convolutional layer, and a max-pooling layer was added before the three fully connected layers with 256, 100, and 4 neurons, respectively. The input size of the proposed network [3] was increased to 512 512. The transfer learning technique used in this application was to fine-tune the last three layers on the ICIAR2018 dataset after having the entire network trained on the ImageNet dataset. The images were stain-normalized with the method proposed in Macenko et al. [19]. An interesting fact is that the authors compared the performance of the model with both non-stain-normalized and stain-normalized images and concluded that using the latter resulted in a gain in performance. After that, data augmentation techniques such as mirroring and rotation were applied, and overlapping patches of size were extracted from each image. Hence, there was a total of 38400 images generated. Evaluation of the model was done using a train-test split of 80%–20%.

The image-wise accuracy reported in [3] was 81.25%, and the patch-wise accuracy was 75.73%. A noteworthy observation is that the normal and benign classes were classified with 85% sensitivity however, the in situ and invasive carcinoma classes were classified with 75% sensitivity. For a model to be practical as a second-opinion system, it should ideally have a higher sensitivity to the carcinoma class given the dangers of misdiagnosis.

For this classification task, the Inception-ResNet-v2 was used by Ferreria et al. [20]. The classification layers of the base model were replaced by a global average pooling layer, a dense (or fully connected) layer with 256 neurons, a dropout layer with a dropout rate of 0.5, and a final dense layer of 4 neurons. Moreover, the input size of the network was changed to . Reshaping the images does not significantly impact the form of the cellular structures however, it does reduce computational cost [20]. The authors did not incorporate stain normalization into their experiments. Data augmentation techniques such as image flips (horizontal and vertical), a 10% zoom range, and shifts (horizontal and vertical) were used to increase the dataset. These particular techniques were chosen with care because if the augmentation causes too much distortion, the anatomical structures in the image could be destroyed [20], and this may result in the network having difficulty extracting discriminative features during training.

Two forms of transfer learning were used in this experiment. At first, only the dense (fully connected) layers of the model were trained. This technique is referred to as feature extraction since the network is using pretrained features (from ImageNet) to classify the breast cancer histology images. The result of this step is that only the weights of the dense layers were adjusted. This aids in overfitting [20]. Afterwards, a certain number of layers were unfrozen so that the network could be fine-tuned. Early stopping with a patience of 20 epochs, and a checkpoint callback monitoring minimum validation loss were the additional techniques implemented to avoid overfitting. The dataset was randomly split into 70% training, 20% validation, and 10% testing. The test set achieved an accuracy of 90%.

In a study by Kassani et al. [21], five different architectures (Inception-v3, Inception-ResNet-v2, Xception, VGG16, and VGG19) were investigated for the classification of the ICIAR2018 dataset. Two stain normalization methods were observed in this study: Macenko et al. [19] and Reinhard et al. [22]. Data augmentation included vertical flips, contrast adjustment, rotation, and brightness correction. The data was split into 75% and 25% for training and testing, respectively. The images were resized to pixels with the help of bicubic interpolation. For each of the models, features were extracted from specific blocks, particularly the layer after a max-pooling layer. The extracted features were put through a global average pooling layer and then concatenated to form a feature vector which was fed into an MLP (multilayer perceptron) set with 256 neurons for final classification. Of these models, the modified Xception network trained with Reinhard stain-normalized images performed the best, with a reported accuracy score of 94%. Overall, the Xception architecture performed the best for both of the stain normalization methods, and the Reinhard [22] technique produced higher accuracies than Macenko [19]. The other architectures ranked in the following order: Inception-v3, Inception-ResNet-v2, VGG16, and VGG19. Interestingly, the approximate parameters for these architectures are 23 million, 54 million, 138 million, and 143 million, respectively. One could hypothesize that an increase in parameter count translates to a decrease in accuracy of this dataset. This indicates that the bigger architectures may have more difficulty extracting critical features from training images, even if measures are taken to enlarge the dataset being used. The results of this study also emphasize the benefit of incorporating stain normalization into preprocessing and how choosing the correct method improves accuracy significantly. Table 1 shows a comparison summary of related deep learning techniques in the literature.


Histopathology-driven artificial intelligence predicts TMB-H colorectal cancer

BEELD: A representative microscopic image of the hematoxylin and eosin (H&E) stained tumor mutational burden-high colorectal cancer tumor. Digital information from such this neoplastic and also non-neoplastic images is transformed and. sien meer

Credit: Niigata University

Niigata, Japan - Biomarkers are important determinants of appropriate and effective therapeutic approaches for various diseases including cancer. There is ample evidence pointing toward the significance of immune check point inhibitors (ICI) against cancer, and they showed promising clinical benefits to a specific group of patients with colorectal cancer (CRC). Several reports demonstrated the efficacy of biomarkers such as programmed death-1 protein ligand (PD-L1), density of tumor-infiltrating lymphocytes (TILs), and tumor mutational burden (TMB), to determine the patient responsiveness for the efficient use of ICIs as therapeutics against cancer.

A high level of TMB (TMB-H), which reflects elevated total number of non-synonymous somatic mutations per coding area of a tumor genome and normally derived from gene panel testing, is recognized as a promising biomarker for the ICI therapies of various solid cancers. However, in clinical practice, it is not feasible to perform gene panel testing for all cancer patients.

In addition, the studies of Dr. Shimada group also provided means to predict TMB-H CRC only by using the TIL information from the H&E slides from the patients' tumor tissues. However, considering that the patients in the studied cohort were not treated with any ICIs, no conclusions could be drawn regarding their ICI responsiveness following the TMB-H diagnosis and it was suggested that future clinical trials need to be conducted to address whether TIL alone can be useful as a predictive biomarker for the efficacy of ICIs. Dr. Shimada says about the present study: "We have developed artificial intelligence to predict genetic alterations in colorectal cancer by deep learning using hematoxylin and eosin slides. This artificial intelligence is important in solving the cost problems associated with genetic analysis and facilitating personalized medicine in colorectal cancer."

Overall, the studies by Dr. Shimada and associates provide a cost and time effective and reliable method to inform the clinicians if the CRC patient they are managing can benefit from Immune Checkpoint Inhibitor (including inhibitors of the PD-1 protein and its ligand, PD-L1) therapy, without implicating the use of gene panel.

Vrywaring: AAAS en EurekAlert! is nie verantwoordelik vir die akkuraatheid van nuusvrystellings wat op EurekAlert geplaas is nie! deur bydraende instansies of vir die gebruik van enige inligting deur die EurekAlert -stelsel.


HER2 Molecular Marker Scoring Using Transfer Learning and Decision Level Fusion

In prognostic evaluation of breast cancer, immunohistochemical (IHC) marker human epidermal growth factor receptor 2 (HER2) is used for prognostic evaluation. Accurate assessment of HER2-stained tissue sample is essential in therapeutic decision making for the patients. In regular clinical settings, expert pathologists assess the HER2-stained tissue slide under microscope for manual scoring based on prior experience. Manual scoring is time consuming, tedious, and often prone to inter-observer variation among group of pathologists. With the recent advancement in the area of computer vision and deep learning, medical image analysis has got significant attention. A number of deep learning architectures have been proposed for classification of different image groups. These networks are also used for transfer learning to classify other image classes. In the presented study, a number of transfer learning architectures are used for HER2 scoring. Five pre-trained architectures viz. VGG16, VGG19, ResNet50, MobileNetV2, en NASNetMobile with decimating the fully connected layers to get 3-class classification have been used for the comparative assessment of the networks as well as further scoring of stained tissue sample image based on statistical voting using mode operator. HER2 Challenge dataset from Warwick University is used in this study. A total of 2130 image patches were extracted to generate the training dataset from 300 training images corresponding to 30 training cases. The output model is then tested on 800 new test image patches from 100 test images acquired from 10 test cases (different from training cases) to report the outcome results. The transfer learning models have shown significant accuracy with VGG19 showing the best accuracy for the test images. The accuracy is found to be 93%, which increases to 98% on the image-based scoring using statistical voting mechanism. The output shows a capable quantification pipeline in automated HER2 score generation.


Bespreking

One limitation of percutaneous image-guided ablation when compared with surgery is the lack of pathologic evidence that the target lesion was completely ablated with sufficient tumor-negative margins. Most studies indicate that positive surgical margins have been associated with a higher risk of local recurrence and shorter overall survival (28,29), a conclusion that has been verified in recent studies performed in the context of modern preoperative chemotherapy (30,31).

Similarly, after RF ablation, the minimal ablation margin is a factor associated with local tumor control (23,32–35). Treatment effectiveness is routinely assessed with postprocedural imaging (US, CT, or MR imaging) to evaluate if the intended ablation margin has been reached, and if it has not, to direct further treatment (23,36,37). Several methods have been proposed for precise calculation of the ablation margin, including fusion imaging (32) and use of anatomic intrahepatic landmarks (23). All of these methods suffer from differences when comparing images from scans performed at different times. Despite progress in fusion and registration software, limitations of these techniques still present a challenge when attempting accurate calculation of the margin of the ablation zone around a previously treated tumor (23,32). Despite the use of these imaging modalities, incomplete tumor treatment and LTP remain common after RF ablation (7–10).

The sole use of imaging in the assessment of ablation margins presumes that the entire depicted ablation zone contains necrotic or nonviable cells however, few studies have validated image findings with tissue characteristics after ablation. In the early radiologic-pathologic correlation study performed by Goldberg et al (38), imaging failed to depict peripheral residual untreated tumor foci, even in four of five cases in which initial tumor and ablation zone sizes were identical. Verder het (a) prior studies performed to evaluate tissue collected from the electrodes used for ablation and (b) our study, in which we preformed direct assessment of the ablation zone with biopsy, show that viable tumor cells may be present within the ablation zone, even when postprocedural imaging displays sufficient ablation margins and findings compatible with radiographic technical success (12,13,16). It appears that these incompletely treated viable tumor cells within the ablation zone “escape” the spatial resolution of available anatomic and functional imaging modalities. In addition, the assessment of tissue examinations in the ablation zone introduces an objective tool with which to assess ablation effectiveness. This assessment is less vulnerable to operator variability and technical limitations, such as those described in the assessment and calculation of the ablation margin (23), and it may be easier to reproduce. Nevertheless, biopsies and sampling errors can also occur, and residual tumor may not be detected in some cases. This can explain the few recurrences seen in our study even after negative biopsy results and sufficient margins were obtained.

Histopathologic examination of tumors excised immediately after RF ablation has shown regions of altered cellular morphology within the treated volume that do not correspond to coagulative necrosis or classic tumor cell appearance (38). These areas may represent either (a) cells in the early stages of irreversible apoptosis or coagulative necrosis or (b) viable cells maintaining their proliferative potential, allowing them to replicate once the cellular insult is removed (21,22). Thus, it is prudent to further interrogate such cells with available immunohistochemical staining for cytosolic and mitochondrial enzyme activity, especially in the immediate postprocedural setting. Days after RF ablation, the evolution of intracellular cascades leads to more pronounced cellular changes in the direction of irreversible apoptosis or necrosis that can be evaluated with standard hematoxylin-eosin staining (38,39). However, immunohistochemical staining has been considered superior to hematoxylin-eosin staining in the diagnosis of irreversible cellular damage up to 24 weeks after RF ablation, as noted in an earlier study by Morimoto et al (39).

An interesting and unique finding of our study is that immunohistochemistry did not alter the initial classification based on the interpretation of the morphologic (hematoxylin-eosin) stain. All 16 ablation zones containing tumor cells were also positive for Ki-67, OXP, or both. This observation differs from observations in previous studies in which ablated tissue was assessed and in which immunohistochemistry was considered necessary to determine viability and changed specimen classification in two (13%) of 15 cases (12,13). It is possible that this finding represents the acquired experience of study pathologists at our institution in the evaluation of ablated tissue, and it potentially justifies the future investigation of the utility of immediate postablation frozen sections and morphologic assessment to document complete tumor cell necrosis.

Viability of tissue adherent to electrodes was analyzed in a study by Snoeren et al (16), who used the autofluorescence method and glucose-6-phosphate diaphorase staining. They concluded that viable tissue was an independent risk factor for LTP . Despite the different methods used to document viability, the incidence of viable tissue in the Snoeren et al study (29.2%) was slightly greater than that in our study (24%) and in the prior studies by Sofocleous et al (19.1%) (12,13). This observation, in combination with the fact that viability in all studies was an independent predictor of LTP , may indicate that both methods of tissue evaluation possess similar efficacy.

Our study had several limitations. The most important factor limiting the weight of our conclusions was the relatively small number of enrolled patients (n = 47). In addition, pathologic evaluation with biopsies from the center and the margin of the ablation zone does not yield information about necrosis within the entire treated lesion volume, as opposed to the evaluation of resected tumors and their surgical margins. Moreover, specimens were classified as Ki-67 positive even if the evaluated tissue was only slightly positive at immunostaining. Estimation of the labeling percentage of the target CLM was not performed, thereby precluding investigation of any potential correlation between the level and grade of proliferation (Ki-67 positivity) and time to LTP . Another limitation of our study was that the minimal margin analyzed as a predictor for time to LTP was evaluated by using 4–8-week postablation CT images, while the presumed minimal margin targeted with biopsy was estimated by using CT images obtained immediately after ablation. That was a rough estimate suited to the time limitations of the procedure however, for this study, accurate estimation and recording of the minimal margin (using CT landmarks) was performed by using CT images obtained 4–8 weeks after RF ablation, as previously described in detail (23). Software capabilities with fusion of the preablation tumor with the ablation zone are currently under development and are evolving. This will enable accurate calculation of the margin during or immediately after ablation and at subsequent time points. As indicated in our work, a positive viable tumor biopsy had a strong and significant prognostic value, as it was an early biomarker of local tumor recurrence and ablation failure. The addition of biopsy of the ablation zone introduces a more objective and reproducible significant predictive assessment of the ablation zone rather than relying on the imaging findings and margin assessment alone.

Vooruitgang in kennis

■ Ablated tumors with posttreatment biopsies containing tumor cells positive for Ki-67 or OxPhos antibodies are 3.4 times more likely to recur (Bl = .008).


Kyk die video: histopathological lab Hematoxylin Eosin staining (Junie 2022).


Kommentaar:

  1. Jean Baptiste

    Jy is verkeerd. Skryf vir my in PM, bespreek dit.

  2. Dilrajas

    Ek kan baie oor hierdie kwessie praat.

  3. Arashakar

    What a remarkable topic

  4. Stod

    Die rekwisiete kom uit

  5. Dabar

    Ek bedoel jy is verkeerd. Tik ons ​​sal dit bespreek. Skryf vir my in PM.



Skryf 'n boodskap