A Clinical Benchmark of Public Self-Supervised Pathology Foundation Models

• DOI: 10.48550/arxiv.2407.06508
• Publication Date: 2024-07-08
• Authors: Gabriele Campanella; Shengjia Chen; Ruchika Verma; Jennifer Zeng; Aryeh Stock; Matt Croken; Brandon Veremis; Abdülkadir Elmas; Kuan‐lin Huang; Ricky Kwan; Jane Houldsworth; Adam J. Schoenfeld; Chad Vanderbilt
• Keywords: Foundation (evidence); Benchmark (surveying)

The use of self-supervised learning (SSL) to train pathology foundation models has increased substantially in the past few years. Notably, several trained on large quantities clinical data have been made publicly available recent months. This will significantly enhance scientific research computational and help bridge gap between deployment. With increase availability public different sizes, using algorithms datasets, it becomes important establish a benchmark compare performance such variety clinically relevant tasks spanning multiple organs diseases. In this work, we present collection datasets comprising slides associated with endpoints including cancer diagnoses biomarkers generated during standard hospital operation from two medical centers. We leverage these systematically assess provide insights into best practices for training new selecting appropriate pretrained models.