{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,11,5]],"date-time":"2026-11-05T19:38:27Z","timestamp":1793907507452,"version":"4.8.0"},"reference-count":23,"publisher":"Frontiers Media SA","license":[{"start":{"date-parts":[[2026,5,20]],"date-time":"2026-05-20T00:00:00Z","timestamp":1779235200000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":["frontiersin.org"],"crossmark-restriction":true},"short-container-title":["Front. Bioinform."],"abstract":"<jats:p>Whole Slide Images (WSIs) are a core data modality in computational pathology, yet their gigapixel resolution requires weakly supervised approaches such as Multiple Instance Learning (MIL) for prognostic modeling. While recent advances in representation learning have introduced domain-specific and foundation models for histopathology, it remains unclear how the choice of patch-level embedding influences survival prediction performance when combined with different MIL architectures and applied across heterogeneous cancer cohorts. Systematic evaluations addressing this gap are still limited. In this work, we present a comprehensive benchmark aimed at deriving practical guidelines for MIL-based Overall Survival (OS) prediction from WSIs. We compare four representative tile embedding strategies (ResNet50, ProvGigaPath, UNI, and CONCH) across five The Cancer Genome Atlas (TCGA) cohorts: bladder urothelial carcinoma (BLCA), breast invasive carcinoma (BRCA), colon adenocarcinoma (COAD), head and neck squamous cell carcinoma (HNSC), stomach adenocarcinoma (STAD) and one dataset from the Clinical Proteomic Tumor Analysis Consortium (CPTAC), clear cell renal cell carcinoma (ccRCC). Patch-level features are aggregated using three state-of-the-art MIL survival models: Attention-Based MIL (ABMIL), Transformer-based MIL (TransMIL), and Dual-Stream MIL (DSMIL). Furthermore, the analysis is extended to include a comparison with two state-of-the-art slide-level encoders, TITAN and the ProvGigaPath slide encoder, utilizing a Cox Proportional Hazard (CPH) model as the prediction head to benchmark patch-aggregation approaches against end-to-end slide representations. Prognostic performance is assessed using the concordance index (c-index), with interpretability evaluated through risk stratification analysis. Our results show that embedding choice has a substantial and consistent impact on OS prediction accuracy, robustness, and interpretability across cancer types, with domain-specific and foundation models outperforming conventional convolutional baselines. These findings provide evidence-based practical guidelines for designing robust and generalizable WSI-based survival prediction pipelines in computational pathology.<\/jats:p>","DOI":"10.3389\/fbinf.2026.1809049","type":"journal-article","created":{"date-parts":[[2026,5,20]],"date-time":"2026-05-20T05:45:08Z","timestamp":1779255908000},"update-policy":"https:\/\/doi.org\/10.3389\/crossmark-policy","source":"Crossref","is-referenced-by-count":1,"title":["Practical guidelines for multiple instance learning in computational pathology: how embedding choice impacts overall survival prediction"],"prefix":"10.3389","volume":"6","author":[{"given":"Francesca","family":"Miccolis","sequence":"first","affiliation":[{"name":"Department of Engineering Enzo Ferrari, University of Modena and Reggio Emilia","place":["Modena, Italy"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Elisa","family":"Ficarra","sequence":"additional","affiliation":[{"name":"Department of Engineering Enzo Ferrari, University of Modena and Reggio Emilia","place":["Modena, Italy"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Marta","family":"Lovino","sequence":"additional","affiliation":[{"name":"Department of Life Sciences, University of Modena and Reggio Emilia","place":["Modena, Italy"]}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"1965","published-online":{"date-parts":[[2026,5,20]]},"reference":[{"key":"B1","doi-asserted-by":"publisher","first-page":"1301","DOI":"10.1038\/s41591-019-0508-1","article-title":"Clinical-grade computational pathology using weakly supervised deep learning on whole slide images","volume":"25","author":"Campanella","year":"2019","journal-title":"Nat. Med."},{"key":"B2","doi-asserted-by":"publisher","first-page":"330","DOI":"10.1038\/nature11252","article-title":"Comprehensive molecular characterization of human colon and rectal cancer","volume":"487","year":"","journal-title":"Nature"},{"key":"B3","doi-asserted-by":"publisher","first-page":"61","DOI":"10.1038\/nature11412","article-title":"Comprehensive molecular portraits of human breast tumours","volume":"490","year":"","journal-title":"Nature"},{"key":"B4","doi-asserted-by":"publisher","first-page":"576","DOI":"10.1038\/nature14129","article-title":"Comprehensive genomic characterization of head and neck squamous cell carcinomas","volume":"517","year":"2015","journal-title":"Nature"},{"key":"B5","doi-asserted-by":"publisher","first-page":"202","DOI":"10.1038\/nature13480","article-title":"Comprehensive molecular characterization of gastric adenocarcinoma","volume":"513","year":"","journal-title":"Nature"},{"key":"B6","doi-asserted-by":"publisher","first-page":"315","DOI":"10.1038\/nature12965","article-title":"Comprehensive molecular characterization of urothelial bladder carcinoma","volume":"507","year":"","journal-title":"Nature"},{"key":"B7","doi-asserted-by":"publisher","first-page":"850","DOI":"10.1038\/s41591-024-02857-3","article-title":"Towards a general-purpose foundation model for computational pathology","volume":"30","author":"Chen","year":"2024","journal-title":"Nat. Med."},{"key":"B8","doi-asserted-by":"publisher","first-page":"964","DOI":"10.1016\/j.cell.2019.10.007","article-title":"Integrated proteogenomic characterization of clear cell renal cell carcinoma","volume":"179","author":"Clark","year":"2019","journal-title":"Cell"},{"key":"B9","doi-asserted-by":"publisher","first-page":"187","DOI":"10.1111\/j.2517-6161.1972.tb00899.x","article-title":"Regression models and life-tables","volume":"34","author":"Cox","year":"1972","journal-title":"J. R. Stat. Soc. Ser. B Methodol."},{"key":"B10","doi-asserted-by":"publisher","first-page":"211","DOI":"10.1111\/his.14918","article-title":"Deep learning-based classification and spatial prognosis risk score on whole-slide images of lung adenocarcinoma","volume":"83","author":"Ding","year":"2023","journal-title":"Histopathology"},{"key":"B11","doi-asserted-by":"publisher","first-page":"3749","DOI":"10.1038\/s41591-025-03982-3","article-title":"A multimodal whole-slide foundation model for pathology","volume":"31","author":"Ding","year":"2025","journal-title":"Nat. Med."},{"key":"B12","doi-asserted-by":"publisher","first-page":"102474","DOI":"10.1016\/j.media.2022.102474","article-title":"Benchmarking weakly-supervised deep learning pipelines for whole slide images in computational pathology","volume":"79","author":"Ghaffari Laleh","year":"2022","journal-title":"Med. Image Anal."},{"key":"B13","first-page":"770","article-title":"Deep residual learning for image recognition","volume-title":"Proceedings of the IEEE conference on computer vision and pattern recognition (CVPR)","author":"He","year":"2016"},{"key":"B14","first-page":"2127","article-title":"Attention-based deep multiple instance learning","volume-title":"International conference on machine learning (ICML)","author":"Ilse","year":"2018"},{"key":"B15","doi-asserted-by":"publisher","first-page":"29","DOI":"10.4103\/2153-3539.186902","article-title":"Deep learning for digital pathology image analysis: a comprehensive tutorial with selected use cases","volume":"7","author":"Janowczyk","year":"2016","journal-title":"J. Pathology Inf."},{"key":"B16","first-page":"14318","article-title":"Dual-stream multiple instance learning network for whole slide image classification with self-supervised contrastive learning","volume-title":"Proceedings of the IEEE\/CVF conference on computer vision and pattern recognition (CVPR)","author":"Li","year":"2021"},{"key":"B17","doi-asserted-by":"publisher","first-page":"555","DOI":"10.1038\/s41551-020-00682-w","article-title":"Data-efficient and weakly supervised computational pathology on whole-slide images","volume":"5","author":"Lu","year":"2021","journal-title":"Nat. Biomed. Eng."},{"key":"B18","doi-asserted-by":"publisher","first-page":"863","DOI":"10.1038\/s41591-024-02856-4","article-title":"A visual-language foundation model for computational pathology","volume":"30","author":"Lu","year":"2024","journal-title":"Nat. Med."},{"key":"B19","doi-asserted-by":"publisher","first-page":"170","DOI":"10.1016\/j.media.2016.06.037","article-title":"Image analysis and machine learning in digital pathology: challenges and opportunities","volume":"33","author":"Madabhushi","year":"2016","journal-title":"Med. Image Anal."},{"key":"B20","article-title":"OXA-MISS: a robust multimodal architecture for chemotherapy response prediction under data scarcity","volume-title":"MICCAI workshop on computational pathology with multimodal data (COMPAYL)","author":"Miccolis","year":"2025"},{"key":"B21","first-page":"2136","article-title":"Transmil: transformer based correlated multiple instance learning for whole slide image classification","volume-title":"Advances in neural information processing systems (NeurIPS)","author":"Shao","year":"2021"},{"key":"B22","doi-asserted-by":"publisher","first-page":"1113","DOI":"10.1038\/ng.2764","article-title":"The cancer genome atlas pan-cancer analysis project","volume":"45","author":"Weinstein","year":"2013","journal-title":"Nat. Genet."},{"key":"B23","doi-asserted-by":"publisher","first-page":"181","DOI":"10.1038\/s41586-024-07441-w","article-title":"A whole-slide foundation model for digital pathology from real-world data","volume":"630","author":"Xu","year":"2024","journal-title":"Nature"}],"container-title":["Frontiers in Bioinformatics"],"original-title":[],"link":[{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fbinf.2026.1809049\/full","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,5,20]],"date-time":"2026-05-20T05:45:10Z","timestamp":1779255910000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fbinf.2026.1809049\/full"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,5,20]]},"references-count":23,"alternative-id":["10.3389\/fbinf.2026.1809049"],"URL":"https:\/\/doi.org\/10.3389\/fbinf.2026.1809049","relation":{},"ISSN":["2673-7647"],"issn-type":[{"value":"2673-7647","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,5,20]]},"article-number":"1809049"}}