\u0130leri d\u00fczey melanom hastalar\u0131nda sa\u011fkal\u0131m tahmininde yapay zek\u00e2 destekli yeni bir yakla\u015f\u0131m geli\u015ftirildi. ECOG-ACRIN Kanser Ara\u015ft\u0131rma Grubu ara\u015ft\u0131rmac\u0131lar\u0131, binlerce y\u00fcksek \u00e7\u00f6z\u00fcn\u00fcrl\u00fckl\u00fc dijital melanom t\u00fcm\u00f6r g\u00f6r\u00fcnt\u00fcs\u00fcnde \u00fc\u00e7\u00fcnc\u00fcl lenfoid yap\u0131lar\u0131 (TLS) tespit etmek i\u00e7in yapay zek\u00e2 tabanl\u0131 y\u00f6ntemler kulland\u0131. Bu geli\u015fme, TLS tespitinde geleneksel patolojik tekniklerin kar\u015f\u0131la\u015ft\u0131\u011f\u0131 zorluklar\u0131 a\u015farak, do\u011fruluk ve tutarl\u0131l\u0131k a\u00e7\u0131s\u0131ndan \u00f6nemli bir ilerleme sa\u011flad\u0131. TLS, ameliyat edilebilir ileri evre melanomlu hastalar i\u00e7in olumlu prognostik bir biyobelirte\u00e7 olarak \u00f6ne \u00e7\u0131k\u0131yor. Ancak, bu yap\u0131lar\u0131n rutin patolog de\u011ferlendirmesine al\u0131nmas\u0131; zaman al\u0131c\u0131, yoruma a\u00e7\u0131k olmas\u0131 ve de\u011fi\u015fkenlik nedenleriyle s\u0131n\u0131rl\u0131 kal\u0131yordu.<\/p>\n
\u00dc\u00e7\u00fcnc\u00fcl lenfoid yap\u0131lar, t\u00fcm\u00f6r mikro\u00e7evresinde geli\u015fen \u00f6zel ba\u011f\u0131\u015f\u0131kl\u0131k h\u00fccresi k\u00fcmeleri olarak tan\u0131mlan\u0131yor. T h\u00fccreleri, B h\u00fccreleri ve dendritik h\u00fccrelerden olu\u015fan bu yap\u0131lar, kronik inflamasyon veya t\u00fcm\u00f6r geli\u015fimine ba\u011fl\u0131 olarak ortaya \u00e7\u0131k\u0131yor. \u00c7e\u015fitli kanser t\u00fcrlerinde artm\u0131\u015f ba\u011f\u0131\u015f\u0131kl\u0131k h\u00fccresi infiltrasyonu ve daha iyi hasta sonu\u00e7lar\u0131yla g\u00fc\u00e7l\u00fc korelasyon g\u00f6steren TLS, ancak tespit zorluklar\u0131 nedeniyle rutin patoloji i\u015f ak\u0131\u015f\u0131na tam olarak entegre edilemedi. ECOG-ACRIN ekibinin geli\u015ftirdi\u011fi yapay zek\u00e2 destekli y\u00f6ntem, karma\u015f\u0131k doku g\u00f6r\u00fcnt\u00fclerinin otomatik analizi yoluyla TLS tespitini kolayla\u015ft\u0131rmak \u00fczere tasarland\u0131.<\/p>\n
Ara\u015ft\u0131rmada retrospektif olarak 376 ileri evre y\u00fcksek riskli melanom hastas\u0131n\u0131n dijital H&E boyal\u0131 patoloji slaytlar\u0131 ile RNA dizileme verileri birle\u015ftirildi. Bu kombinasyon sayesinde TLS varl\u0131\u011f\u0131 ile genel sa\u011fkal\u0131m aras\u0131nda anlaml\u0131 bir ili\u015fki kuruldu. Veriler, imm\u00fcn kontrol noktas\u0131 inhibit\u00f6rleri ve sitokin tedavilerinin de\u011ferlendirildi\u011fi E1609 klinik denemesinden elde edildi. Elde edilen b\u00fcy\u00fck veri taban\u0131, gelecekte prognostik belirte\u00e7lerin ve tedavi stratejilerinin geli\u015ftirilmesi i\u00e7in sa\u011flam bir temel olu\u015fturuyor.<\/p>\n
Kohort analizine g\u00f6re hastalar\u0131n yakla\u015f\u0131k %55\u2019inde TLS g\u00f6r\u00fcld\u00fc ve TLS pozitif hastalar\u0131n be\u015f y\u0131ll\u0131k genel sa\u011fkal\u0131m oran\u0131 %36,23 olurken, TLS negatiflarda bu oran %29,59 olarak belirlendi. Ayr\u0131ca, birden \u00e7ok TLS\u2019ye sahip hastalarda sa\u011fkal\u0131m avantaj\u0131n\u0131n daha belirgin oldu\u011fu saptand\u0131. Bu durum, sadece TLS varl\u0131\u011f\u0131n\u0131n de\u011fil, TLS yo\u011funlu\u011funun da prognostik \u00f6nemi oldu\u011funu g\u00f6steriyor. Hastalar ayr\u0131ca AJCC t\u00fcm\u00f6r evresi, ya\u015f, cinsiyet, uygulanan tedavi y\u00f6ntemi ve t\u00fcm\u00f6r \u00fclserasyon durumu gibi klinik parametrelere g\u00f6re alt gruplara ayr\u0131larak sa\u011fkal\u0131m s\u00fcreci detayland\u0131r\u0131ld\u0131.<\/p>\n
Bu ilerlemelerin merkezinde, dijital histolojik g\u00f6r\u00fcnt\u00fclerde TLS ve germinal merkezleri otomatik tespit etmek \u00fczere geli\u015ftirilmi\u015f a\u00e7\u0131k kaynakl\u0131 derin \u00f6\u011frenme algoritmas\u0131 HookNet-TLS bulunuyor. HookNet, biyomedikal g\u00f6r\u00fcnt\u00fc analizi i\u00e7in tasarlanm\u0131\u015f konvol\u00fcsyonel sinir a\u011f\u0131 mimarileriyle karma\u015f\u0131k doku yap\u0131lar\u0131n\u0131n biti\u015fik segmentasyon ve s\u0131n\u0131fland\u0131rmas\u0131n\u0131 y\u00fcksek \u00e7\u00f6z\u00fcn\u00fcrl\u00fckte ger\u00e7ekle\u015ftiriyor. Ba\u015flang\u0131\u00e7taki uygulamalarda g\u00f6sterdi\u011fi ba\u015far\u0131l\u0131 sonu\u00e7lar\u0131n ard\u0131ndan modelde do\u011fruluk art\u0131r\u0131c\u0131 iyile\u015ftirmeler yap\u0131ld\u0131 ve TLS ile ili\u015fkili germinal merkezlerin g\u00fcvenilir \u015fekilde nicelendirilebilmesi sa\u011fland\u0131.<\/p>\n
HookNet\u2019in performans\u0131n\u0131 tamamlay\u0131c\u0131 olarak, ara\u015ft\u0131rmac\u0131lar Gigapth Whole-Slide Foundation Model\u2019in de \u00f6zellik \u00e7\u0131karma yeteneklerinden yararland\u0131. Bu geli\u015fmekte olan model, dijital patoloji i\u00e7in optimize edilmekte olup, H&E g\u00f6r\u00fcnt\u00fc par\u00e7alay\u0131c\u0131lar\u0131nda temel morfolojik de\u011fi\u015fkenlikleri yakalayan ana bile\u015fen analizi (PCA) kullan\u0131m\u0131yla g\u00f6rselle\u015ftirme ve analiz yetene\u011fini art\u0131r\u0131yor. Gigapth\u2019in erken PCA \u00e7\u0131kt\u0131 g\u00f6rselle\u015ftirmeleri, TLS tespitindeki hassasiyetini destekler nitelikte olup, s\u00fcrekli model ince ayar ve do\u011frulama \u00e7al\u0131\u015fmalar\u0131 s\u00fcrd\u00fcr\u00fclmekte.<\/p>\n
Yapay zek\u00e2 destekli bu ara\u00e7lar\u0131n rutin klinik i\u015f ak\u0131\u015flar\u0131na entegrasyonu bir\u00e7ok a\u00e7\u0131dan \u00f6nem ta\u015f\u0131yor. D\u00fc\u015f\u00fck maliyetli, yayg\u0131n kullan\u0131lan H&E boyal\u0131 numunelerin de\u011ferlendirilmesinin otomatikle\u015ftirilmesi, \u00f6nceden subjektif ve zaman alan i\u015flemleri standartla\u015ft\u0131r\u0131yor. TLS tespitinde artan duyarl\u0131l\u0131k ve \u00f6zg\u00fcll\u00fck, AJCC evreleme sisteminde daha do\u011fru prognostik s\u0131n\u0131fland\u0131rmalar yap\u0131lmas\u0131na imk\u00e2n veriyor. Bu da y\u00fcksek riskli melanom hastalar\u0131n\u0131n ki\u015fiselle\u015ftirilmi\u015f imm\u00fcnoterapi kararlar\u0131n\u0131n \u015fekillendirilmesi ve sonu\u00e7lar\u0131n\u0131n iyile\u015ftirilmesine do\u011frudan katk\u0131 sa\u011fl\u0131yor.<\/p>\n
Bu \u00e7al\u0131\u015fma, Ulusal Kanser Enstit\u00fcs\u00fc (NCI) finansman deste\u011fiyle ger\u00e7ekle\u015ferek biyomedikal g\u00f6r\u00fcnt\u00fcleme, makine \u00f6\u011frenimi ve molek\u00fcler onkolojinin sinerjisini ortaya koyuyor. Kritik ba\u011f\u0131\u015f\u0131kl\u0131k mikro\u00e7evresi bile\u015fenlerinin h\u0131zl\u0131 ve tekrarlanabilir \u015fekilde nicelenmesine olanak sa\u011flayan y\u00f6ntem, TLS gibi biyobelirte\u00e7lerin tan\u0131sal paradigmaya kat\u0131lmas\u0131n\u0131n \u00f6n\u00fcn\u00fc a\u00e7\u0131yor. B\u00f6ylece, tedavi algoritmalar\u0131na entegre edilerek terap\u00f6tik kararlar\u0131n s\u0131n\u0131rlar\u0131n\u0131 da geni\u015fletiyor.<\/p>\n
Ara\u015ft\u0131rman\u0131n ba\u015f\u0131ndaki isimlerden Moffitt Kanser Merkezi\u2019nden Prof. Dr. Ahmad A. Tarhini, \u201cA\u00e7\u0131k eri\u015fimli yapay zek\u00e2 ara\u00e7lar\u0131 kullanarak imm\u00fcn yap\u0131lar\u0131n ayr\u0131nt\u0131l\u0131 analizi ile sa\u011fkal\u0131m ve imm\u00fcnoterapi yan\u0131t\u0131n\u0131 tahmin etmede devrim yarat\u0131yoruz\u201d ifadesiyle \u00e7al\u0131\u015fman\u0131n \u00f6nemini vurgulad\u0131. Yard\u0131mc\u0131 ara\u015ft\u0131rmac\u0131 Dr. Xuefeng Wang ise Gigapth gibi temel modellerin analizin hassasiyetini art\u0131rmadaki potansiyeline dikkat \u00e7ekerek, metodolojinin klinik kullan\u0131m kapsam\u0131n\u0131 geni\u015fletmek i\u00e7in devam eden geli\u015ftirme \u00e7al\u0131\u015fmalar\u0131n\u0131 payla\u015ft\u0131.<\/p>\n
TLS\u2019nin h\u0131zl\u0131 ve do\u011fru tespiti, \u00f6zellikle melanomlu hasta- hekim ileti\u015fiminde imm\u00fcnoterapinin m\u00fcmk\u00fcn avantajlar\u0131 hakk\u0131nda daha doyurucu bilgiler sunabilir. Yapay zek\u00e2 y\u00f6ntemleri geli\u015fip olgunla\u015ft\u0131k\u00e7a, bu teknolojilerin sadece onkoloji alan\u0131nda de\u011fil, TLS\u2019nin rol oynayabilece\u011fi di\u011fer imm\u00fcn ili\u015fkili hastal\u0131klarda da \u00e7\u0131\u011f\u0131r a\u00e7mas\u0131 bekleniyor. Bu ba\u011f\u0131\u015f\u0131kl\u0131k h\u00fccresi k\u00fcmeleri, antit\u00fcm\u00f6r ba\u011f\u0131\u015f\u0131kl\u0131k aktivitelerinin dinamik merkezleri olarak \u00f6nem ta\u015f\u0131yor.<\/p>\n
TLS tespiti ve nicelenmesi, klasik y\u00f6ntemlerde gizli histolojik \u00f6zelliklerin uzman patologlarca de\u011ferlendirilmesini gerektiriyor. Bu s\u00fcre\u00e7 inter-g\u00f6zlemci de\u011fi\u015fkenli\u011fi ve kaynak k\u0131s\u0131tlar\u0131 nedeniyle zorluklar i\u00e7eriyor. HookNet-TLS gibi otomatikle\u015ftirilmi\u015f algoritmalar\u0131n y\u00fcksek do\u011frulukla uygulanmas\u0131, i\u015f ak\u0131\u015f\u0131 verimlili\u011fi ve tan\u0131sal standartla\u015ft\u0131rmada kritik bo\u015fluklar\u0131 kapat\u0131yor. Ayr\u0131ca, HookNet\u2019in Grand Challenge gibi platformlarda a\u00e7\u0131k kaynak olarak sunulmas\u0131, biyomedikal g\u00f6r\u00fcnt\u00fc analizi ve yapay zek\u00e2 topluluklar\u0131n\u0131n i\u015f birli\u011fini ve \u015feffafl\u0131\u011f\u0131 art\u0131r\u0131yor.<\/p>\n
Elde edilen bulgular, 2025 y\u0131l\u0131nda Chicago\u2019da yap\u0131lacak Amerikan Kanser Ara\u015ft\u0131rmalar\u0131 Derne\u011fi (AACR) y\u0131ll\u0131k toplant\u0131s\u0131nda sunulacak. Sunumun, model performans\u0131 ve klinik uygulanabilirlik hakk\u0131nda ayr\u0131nt\u0131l\u0131 bilgi vererek yapay zek\u00e2 tabanl\u0131 patoloji yat\u0131r\u0131mlar\u0131nda yeni bir d\u00f6nemin ba\u015flang\u0131c\u0131n\u0131 tetiklemesi bekleniyor. ECOG-ACRIN ekibi, geli\u015fmi\u015f dijital patoloji kaynaklar\u0131n\u0131 kullanarak melanoma prognostiklerinde makine \u00f6\u011freniminin entegre edilmesinde \u00f6l\u00e7eklenebilir bir yol a\u00e7t\u0131.<\/p>\n
Bu ara\u015ft\u0131rma, ileri evre melanomda hayatta kal\u0131m\u0131 iyile\u015ftirmek i\u00e7in yapay zek\u00e2 destekli TLS tespiti kullanarak, bireye \u00f6zel onkoloji anlay\u0131\u015f\u0131nda \u00f6nemli ad\u0131mlar at\u0131yor ve klinik bak\u0131mda biyobelirte\u00e7lerin rol\u00fcn\u00fc yeni bir seviyeye ta\u015f\u0131yor.<\/p>\n
—<\/p>\n
**Ara\u015ft\u0131rma Konusu**: Yapay zek\u00e2 tabanl\u0131 \u00fc\u00e7\u00fcnc\u00fcl lenfoid yap\u0131lar\u0131n tespitiyle ileri evre melanomda sa\u011fkal\u0131m tahmininin geli\u015ftirilmesi<\/p>\n
**Makale Ba\u015fl\u0131\u011f\u0131**: \u0130\u00e7erikte belirtilmemi\u015f<\/p>\n
**Haberin Yay\u0131n Tarihi**: Belirtilmemi\u015f<\/p>\n
**Web References**:
\n– ECOG-ACRIN Cancer Research Group: www.ecog-acrin.org
\n– Grand Challenge platform: https:\/\/grand-challenge.org\/
\n– HookNet-TLS algoritmas\u0131: https:\/\/grand-challenge.org\/algorithms\/hooknet-tls\/ <\/p>\n
**Resim Credits**: Ahmad A. Tarhini ve \u00e7al\u0131\u015fma arkada\u015flar\u0131<\/p>\n
**Anahtar Kelimeler**: Yapay zek\u00e2, Melanom, G\u00f6r\u00fcnt\u00fc analizi, Biyobelirte\u00e7ler, Deri kanseri, RNA dizileme, \u0130mm\u00fcn infiltrasyon, T\u00fcm\u00f6r mikroyap\u0131s\u0131, \u00dc\u00e7\u00fcnc\u00fcl lenfoid yap\u0131lar, Otomatik patoloji, Dijital patoloji, AI onkolojide<\/p>\n","protected":false},"excerpt":{"rendered":"
\u0130leri d\u00fczey melanom hastalar\u0131nda sa\u011fkal\u0131m tahmininde yapay zek\u00e2 destekli yeni bir yakla\u015f\u0131m geli\u015ftirildi. ECOG-ACRIN Kanser Ara\u015ft\u0131rma Grubu ara\u015ft\u0131rmac\u0131lar\u0131, binlerce y\u00fcksek \u00e7\u00f6z\u00fcn\u00fcrl\u00fckl\u00fc dijital melanom t\u00fcm\u00f6r g\u00f6r\u00fcnt\u00fcs\u00fcnde \u00fc\u00e7\u00fcnc\u00fcl lenfoid yap\u0131lar\u0131 (TLS) tespit etmek i\u00e7in yapay zek\u00e2 tabanl\u0131 y\u00f6ntemler kulland\u0131. Bu geli\u015fme, TLS tespitinde geleneksel patolojik tekniklerin kar\u015f\u0131la\u015ft\u0131\u011f\u0131 zorluklar\u0131 a\u015farak, do\u011fruluk ve tutarl\u0131l\u0131k a\u00e7\u0131s\u0131ndan \u00f6nemli bir ilerleme sa\u011flad\u0131….<\/p>\n","protected":false},"author":1,"featured_media":3795,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_yoast_wpseo_title":"","_yoast_wpseo_metadesc":"","_yoast_wpseo_focuskw":"","rank_math_title":"","rank_math_description":"","rank_math_focus_keyword":"","_wpan_schema_json_ld":"","_wpan_ai_seo_metadata":"","_wpan_ai_seo_status":"","_wpan_ai_seo_policy":"","_wpan_ai_seo_faq_block":"","_jetpack_memberships_contains_paid_content":false,"footnotes":""},"categories":[28],"tags":[3018,3015,3016,3017,3014],"tmauthors":[],"class_list":{"0":"post-3794","1":"post","2":"type-post","3":"status-publish","4":"format-standard","5":"has-post-thumbnail","7":"category-kanser","8":"tag-immun-kontrol-noktasi-inhibitorleri-tedavi-etkileri","9":"tag-melanom-hastalarinda-sagkalim-tahmini","10":"tag-ucuncul-lenfoid-yapilar-tespiti","11":"tag-yapay-zeka-destekli-patoloji-goruntu-analizi","12":"tag-yapay-zeka-ile-immun-hucre-analizi"},"jetpack_featured_media_url":"https:\/\/haber360.com\/wp-content\/uploads\/2025\/04\/Yapay-Zeka-ile-Immun-Hucre-Analizi-ve-Sagkalim-1745512170.jpg","jetpack_sharing_enabled":true,"_links":{"self":[{"href":"https:\/\/haber360.com\/index.php\/wp-json\/wp\/v2\/posts\/3794","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/haber360.com\/index.php\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/haber360.com\/index.php\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/haber360.com\/index.php\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/haber360.com\/index.php\/wp-json\/wp\/v2\/comments?post=3794"}],"version-history":[{"count":0,"href":"https:\/\/haber360.com\/index.php\/wp-json\/wp\/v2\/posts\/3794\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/haber360.com\/index.php\/wp-json\/wp\/v2\/media\/3795"}],"wp:attachment":[{"href":"https:\/\/haber360.com\/index.php\/wp-json\/wp\/v2\/media?parent=3794"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/haber360.com\/index.php\/wp-json\/wp\/v2\/categories?post=3794"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/haber360.com\/index.php\/wp-json\/wp\/v2\/tags?post=3794"},{"taxonomy":"tmauthors","embeddable":true,"href":"https:\/\/haber360.com\/index.php\/wp-json\/wp\/v2\/tmauthors?post=3794"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}