Subspecialty-Specific Foundation Model for Intelligent Gastrointestinal Pathology

TL;DR

This paper introduces Digepath, a specialized foundation model for gastrointestinal pathology that significantly improves diagnostic accuracy and sensitivity across multiple tasks and institutions, advancing AI-driven precision medicine.

Contribution

Developed Digepath, a novel GI-specific foundation model with a dual-phase training strategy, achieving state-of-the-art results and high sensitivity in early cancer detection.

Findings

Achieved state-of-the-art performance on 33 of 34 GI pathology tasks.

Attained 99.70% sensitivity in early GI cancer screening across nine institutions.

Pretrained on over 353 million images from 210,043 slides.

Abstract

Gastrointestinal (GI) diseases represent a clinically significant burden, necessitating precise diagnostic approaches to optimize patient outcomes. Conventional histopathological diagnosis suffers from limited reproducibility and diagnostic variability. To overcome these limitations, we develop Digepath, a specialized foundation model for GI pathology. Our framework introduces a dual-phase iterative optimization strategy combining pretraining with fine-screening, specifically designed to address the detection of sparsely distributed lesion areas in whole-slide images. Digepath is pretrained on over 353 million multi-scale images from 210,043 H&E-stained slides of GI diseases. It attains state-of-the-art performance on 33 out of 34 tasks related to GI pathology, including pathological diagnosis, protein expression status prediction, gene mutation prediction, and prognosis evaluation. We further translate the intelligent screening module for early GI cancer and achieve near-perfect 99.70% sensitivity across nine independent medical institutions. This work not only advances AI-driven precision pathology for GI diseases but also bridge critical gaps in histopathological practice.