학술논문
Teacher-student collaborated multiple instance learning for pan-cancer PDL1 expression prediction from histopathology slides.
Document Type
Academic Journal
Author
Jin D; Image Processing Center, Beihang University, Beijing, 102206, China.; Division of AI in Oncology, German Cancer Research Center (DKFZ), Heidelberg, Germany.; Shen Yuan Honors College, Beihang University, Beijing, 100191, China.; Liang S; Image Processing Center, Beihang University, Beijing, 102206, China.; Shmatko A; Division of AI in Oncology, German Cancer Research Center (DKFZ), Heidelberg, Germany.; Arnold A; Charité - Universitätsmedizin Berlin, Institute of Pathology, 10117, Berlin, Germany.; Horst D; Charité - Universitätsmedizin Berlin, Institute of Pathology, 10117, Berlin, Germany.; German Cancer Consortium (DKTK), partner site Berlin, a partnership between DKFZ and Charité-Universitätsmedizin Berlin, Berlin, Germany.; Grünewald TGP; Institute of Pathology, Heidelberg University Hospital, Heidelberg, Germany. t.gruenewald@kitz-heidelberg.de.; Division of Translational Pediatric Sarcoma Research, German Cancer Research Center (DKFZ), German Cancer Consortium (DKTK), Heidelberg, Germany. t.gruenewald@kitz-heidelberg.de.; Hopp Children's Cancer Center (KiTZ) Heidelberg, Heidelberg, Germany. t.gruenewald@kitz-heidelberg.de.; National Center for Tumor Diseases (NCT), NCT Heidelberg, a partnership between DKFZ and Heidelberg University Hospital, Heidelberg, Germany. t.gruenewald@kitz-heidelberg.de.; Gerstung M; Division of AI in Oncology, German Cancer Research Center (DKFZ), Heidelberg, Germany. moritz.gerstung@dkfz.de.; Bai X; Image Processing Center, Beihang University, Beijing, 102206, China. jackybxz@buaa.edu.cn.; State Key Laboratory of Virtual Reality Technology and Systems, Beihang University, Beijing, 100191, China. jackybxz@buaa.edu.cn.; Advanced Innovation Center for Biomedical Engineering, Beihang University, Beijing, 100083, China. jackybxz@buaa.edu.cn.
Source
Publisher: Nature Pub. Group Country of Publication: England NLM ID: 101528555 Publication Model: Electronic Cited Medium: Internet ISSN: 2041-1723 (Electronic) Linking ISSN: 20411723 NLM ISO Abbreviation: Nat Commun Subsets: MEDLINE
Subject
Language
English
Abstract
Programmed cell death ligand 1 (PDL1), as an important biomarker, is quantified by immunohistochemistry (IHC) with few established histopathological patterns. Deep learning aids in histopathological assessment, yet heterogeneity and lacking spatially resolved annotations challenge precise analysis. Here, we present a weakly supervised learning approach using bulk RNA sequencing for PDL1 expression prediction from hematoxylin and eosin (H&E) slides. Our method extends the multiple instance learning paradigm with the teacher-student framework, which assigns dynamic pseudo-labels for intra-slide heterogeneity and retrieves unlabeled instances using temporal ensemble model distillation. The approach, evaluated on 12,299 slides across 20 solid tumor types, achieves a weighted average area under the curve of 0.83 on fresh-frozen and 0.74 on formalin-fixed specimens for 9 tumors with PDL1 as an established biomarker. Our method predicts PDL1 expression patterns, validated by IHC on 20 slides, offering insights into histologies relevant to PDL1. This demonstrates the potential of deep learning in identifying diverse histological patterns for molecular changes from H&E images.
(© 2024. The Author(s).)
(© 2024. The Author(s).)