A clinical benchmark of public self-supervised pathology foundation models

Abstract The use of self-supervised learning to train pathology foundation models has increased substantially in the past few years. Notably, several models trained on large quantities of clinical data have been made publicly available in recent months. This will significantly enhance scientific res...

Full description

Saved in:
Bibliographic Details
Main Authors: Gabriele Campanella, Shengjia Chen, Manbir Singh, Ruchika Verma, Silke Muehlstedt, Jennifer Zeng, Aryeh Stock, Matt Croken, Brandon Veremis, Abdulkadir Elmas, Ivan Shujski, Noora Neittaanmäki, Kuan-lin Huang, Ricky Kwan, Jane Houldsworth, Adam J. Schoenfeld, Chad Vanderbilt
Format: Article
Language:English
Published: Nature Portfolio 2025-04-01
Series:Nature Communications
Online Access:https://doi.org/10.1038/s41467-025-58796-1
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849698493239656448
author Gabriele Campanella
Shengjia Chen
Manbir Singh
Ruchika Verma
Silke Muehlstedt
Jennifer Zeng
Aryeh Stock
Matt Croken
Brandon Veremis
Abdulkadir Elmas
Ivan Shujski
Noora Neittaanmäki
Kuan-lin Huang
Ricky Kwan
Jane Houldsworth
Adam J. Schoenfeld
Chad Vanderbilt
author_facet Gabriele Campanella
Shengjia Chen
Manbir Singh
Ruchika Verma
Silke Muehlstedt
Jennifer Zeng
Aryeh Stock
Matt Croken
Brandon Veremis
Abdulkadir Elmas
Ivan Shujski
Noora Neittaanmäki
Kuan-lin Huang
Ricky Kwan
Jane Houldsworth
Adam J. Schoenfeld
Chad Vanderbilt
author_sort Gabriele Campanella
collection DOAJ
description Abstract The use of self-supervised learning to train pathology foundation models has increased substantially in the past few years. Notably, several models trained on large quantities of clinical data have been made publicly available in recent months. This will significantly enhance scientific research in computational pathology and help bridge the gap between research and clinical deployment. With the increase in availability of public foundation models of different sizes, trained using different algorithms on different datasets, it becomes important to establish a benchmark to compare the performance of such models on a variety of clinically relevant tasks spanning multiple organs and diseases. In this work, we present a collection of pathology datasets comprising clinical slides associated with clinically relevant endpoints including cancer diagnoses and a variety of biomarkers generated during standard hospital operation from three medical centers. We leverage these datasets to systematically assess the performance of public pathology foundation models and provide insights into best practices for training foundation models and selecting appropriate pretrained models. To enable the community to evaluate their models on our clinical datasets, we make available an automated benchmarking pipeline for external use.
format Article
id doaj-art-bf569c2256f3438ea529c413ecf4632a
institution DOAJ
issn 2041-1723
language English
publishDate 2025-04-01
publisher Nature Portfolio
record_format Article
series Nature Communications
spelling doaj-art-bf569c2256f3438ea529c413ecf4632a2025-08-20T03:18:53ZengNature PortfolioNature Communications2041-17232025-04-0116111210.1038/s41467-025-58796-1A clinical benchmark of public self-supervised pathology foundation modelsGabriele Campanella0Shengjia Chen1Manbir Singh2Ruchika Verma3Silke Muehlstedt4Jennifer Zeng5Aryeh Stock6Matt Croken7Brandon Veremis8Abdulkadir Elmas9Ivan Shujski10Noora Neittaanmäki11Kuan-lin Huang12Ricky Kwan13Jane Houldsworth14Adam J. Schoenfeld15Chad Vanderbilt16Windreich Department of AI and Human Health, Icahn School of Medicine at Mount SinaiWindreich Department of AI and Human Health, Icahn School of Medicine at Mount SinaiWindreich Department of AI and Human Health, Icahn School of Medicine at Mount SinaiWindreich Department of AI and Human Health, Icahn School of Medicine at Mount SinaiWindreich Department of AI and Human Health, Icahn School of Medicine at Mount SinaiDepartment of Pathology, Icahn School of Medicine at Mount SinaiDepartment of Pathology, Icahn School of Medicine at Mount SinaiDepartment of Pathology, Icahn School of Medicine at Mount SinaiDepartment of Pathology, Icahn School of Medicine at Mount SinaiDepartment of Genetics and Genomics, Icahn School of Medicine at Mount SinaiDepartment of Clinical Pathology, Sahlgrenska University HospitalDepartment of Clinical Pathology, Sahlgrenska University HospitalDepartment of Genetics and Genomics, Icahn School of Medicine at Mount SinaiDepartment of Pathology, Icahn School of Medicine at Mount SinaiDepartment of Pathology, Icahn School of Medicine at Mount SinaiDepartment of Medicine, Memorial Sloan Kettering Cancer CenterDepartment of Pathology, Memorial Sloan Kettering Cancer CenterAbstract The use of self-supervised learning to train pathology foundation models has increased substantially in the past few years. Notably, several models trained on large quantities of clinical data have been made publicly available in recent months. This will significantly enhance scientific research in computational pathology and help bridge the gap between research and clinical deployment. With the increase in availability of public foundation models of different sizes, trained using different algorithms on different datasets, it becomes important to establish a benchmark to compare the performance of such models on a variety of clinically relevant tasks spanning multiple organs and diseases. In this work, we present a collection of pathology datasets comprising clinical slides associated with clinically relevant endpoints including cancer diagnoses and a variety of biomarkers generated during standard hospital operation from three medical centers. We leverage these datasets to systematically assess the performance of public pathology foundation models and provide insights into best practices for training foundation models and selecting appropriate pretrained models. To enable the community to evaluate their models on our clinical datasets, we make available an automated benchmarking pipeline for external use.https://doi.org/10.1038/s41467-025-58796-1
spellingShingle Gabriele Campanella
Shengjia Chen
Manbir Singh
Ruchika Verma
Silke Muehlstedt
Jennifer Zeng
Aryeh Stock
Matt Croken
Brandon Veremis
Abdulkadir Elmas
Ivan Shujski
Noora Neittaanmäki
Kuan-lin Huang
Ricky Kwan
Jane Houldsworth
Adam J. Schoenfeld
Chad Vanderbilt
A clinical benchmark of public self-supervised pathology foundation models
Nature Communications
title A clinical benchmark of public self-supervised pathology foundation models
title_full A clinical benchmark of public self-supervised pathology foundation models
title_fullStr A clinical benchmark of public self-supervised pathology foundation models
title_full_unstemmed A clinical benchmark of public self-supervised pathology foundation models
title_short A clinical benchmark of public self-supervised pathology foundation models
title_sort clinical benchmark of public self supervised pathology foundation models
url https://doi.org/10.1038/s41467-025-58796-1
work_keys_str_mv AT gabrielecampanella aclinicalbenchmarkofpublicselfsupervisedpathologyfoundationmodels
AT shengjiachen aclinicalbenchmarkofpublicselfsupervisedpathologyfoundationmodels
AT manbirsingh aclinicalbenchmarkofpublicselfsupervisedpathologyfoundationmodels
AT ruchikaverma aclinicalbenchmarkofpublicselfsupervisedpathologyfoundationmodels
AT silkemuehlstedt aclinicalbenchmarkofpublicselfsupervisedpathologyfoundationmodels
AT jenniferzeng aclinicalbenchmarkofpublicselfsupervisedpathologyfoundationmodels
AT aryehstock aclinicalbenchmarkofpublicselfsupervisedpathologyfoundationmodels
AT mattcroken aclinicalbenchmarkofpublicselfsupervisedpathologyfoundationmodels
AT brandonveremis aclinicalbenchmarkofpublicselfsupervisedpathologyfoundationmodels
AT abdulkadirelmas aclinicalbenchmarkofpublicselfsupervisedpathologyfoundationmodels
AT ivanshujski aclinicalbenchmarkofpublicselfsupervisedpathologyfoundationmodels
AT nooraneittaanmaki aclinicalbenchmarkofpublicselfsupervisedpathologyfoundationmodels
AT kuanlinhuang aclinicalbenchmarkofpublicselfsupervisedpathologyfoundationmodels
AT rickykwan aclinicalbenchmarkofpublicselfsupervisedpathologyfoundationmodels
AT janehouldsworth aclinicalbenchmarkofpublicselfsupervisedpathologyfoundationmodels
AT adamjschoenfeld aclinicalbenchmarkofpublicselfsupervisedpathologyfoundationmodels
AT chadvanderbilt aclinicalbenchmarkofpublicselfsupervisedpathologyfoundationmodels
AT gabrielecampanella clinicalbenchmarkofpublicselfsupervisedpathologyfoundationmodels
AT shengjiachen clinicalbenchmarkofpublicselfsupervisedpathologyfoundationmodels
AT manbirsingh clinicalbenchmarkofpublicselfsupervisedpathologyfoundationmodels
AT ruchikaverma clinicalbenchmarkofpublicselfsupervisedpathologyfoundationmodels
AT silkemuehlstedt clinicalbenchmarkofpublicselfsupervisedpathologyfoundationmodels
AT jenniferzeng clinicalbenchmarkofpublicselfsupervisedpathologyfoundationmodels
AT aryehstock clinicalbenchmarkofpublicselfsupervisedpathologyfoundationmodels
AT mattcroken clinicalbenchmarkofpublicselfsupervisedpathologyfoundationmodels
AT brandonveremis clinicalbenchmarkofpublicselfsupervisedpathologyfoundationmodels
AT abdulkadirelmas clinicalbenchmarkofpublicselfsupervisedpathologyfoundationmodels
AT ivanshujski clinicalbenchmarkofpublicselfsupervisedpathologyfoundationmodels
AT nooraneittaanmaki clinicalbenchmarkofpublicselfsupervisedpathologyfoundationmodels
AT kuanlinhuang clinicalbenchmarkofpublicselfsupervisedpathologyfoundationmodels
AT rickykwan clinicalbenchmarkofpublicselfsupervisedpathologyfoundationmodels
AT janehouldsworth clinicalbenchmarkofpublicselfsupervisedpathologyfoundationmodels
AT adamjschoenfeld clinicalbenchmarkofpublicselfsupervisedpathologyfoundationmodels
AT chadvanderbilt clinicalbenchmarkofpublicselfsupervisedpathologyfoundationmodels