Paper

EVLF-FM: Explainable Vision Language Foundation Model for Medicine

Elze, Tobias, Zhou, Jun, Hon Lim, Tony Kiat, Darren Shu Jeng Ting, Tan, Marcus, Mehta, Jod, Gutierrez, Laura, Hiok Hong Chan, Nielsen, Christopher S, Cheng, Lionel Tim-Ee, Bai, Yang, Ching Yu Cheng, Lim, Soon Thye, Teo, Zhen Ling, Schmetterer, Leopold, Fukutsu, Kanae, Linh Le Dinh, Yao, Jie, Ang, Marcus, Liu, Nan, Soetikno, Brian T, Ke, Yuhe, Koh, Victor, Klang, Eyal, Chee Leong Cheng, Hussain, Rahat, Aung, Tin, Tan, Iain Beehuat, Tran Nguyen Tuan Anh, Rick Siow Mong Goh, Li, Kelvin Z, Zhou, Yang, Thirunavukarasu, Arun, Chrystie Wan Ning Quek, Li, Zengxiang, Yip, Leonard, Yih Chung Tham, Daniel Shu Wei Ting, Cheng, Haoran, Hong, Ashley, Liu, Yong, Gavin Siew Wei Tan, Tien Yin Wong

2025

Overview

Despite the promise of foundation models in medical AI, current systems remain limited - they are modality-specific and lack transparent reasoning processes, hindering clinical adoption. To address this gap, we present EVLF-FM, a multimodal vision-language foundation model (VLM) designed to unify broad diagnostic capability with fine-grain explainability. The development and testing of EVLF-FM encompassed over 1.3 million total samples from 23 global datasets across eleven imaging modalities related to six clinical specialties: dermatology, hepatology, ophthalmology, pathology, pulmonology, and radiology. External validation employed 8,884 independent test samples from 10 additional datasets across five imaging modalities. Technically, EVLF-FM is developed to assist with multiple disease diagnosis and visual question answering with pixel-level visual grounding and reasoning capabilities. In internal validation for disease diagnostics, EVLF-FM achieved the highest average accuracy (0.858) and F1-score (0.797), outperforming leading generalist and specialist models. In medical visual grounding, EVLF-FM also achieved stellar performance across nine modalities with average mIOU of 0.743 and Acc@0.5 of 0.837. External validations further confirmed strong zero-shot and few-shot performance, with competitive F1-scores despite a smaller model size. Through a hybrid training strategy combining supervised and visual reinforcement fine-tuning, EVLF-FM not only achieves state-of-the-art accuracy but also exhibits step-by-step reasoning, aligning outputs with visual evidence. EVLF-FM is an early multi-disease VLM model with explainability and reasoning capabilities that could advance adoption of and trust in foundation models for real-world clinical deployment.

• Publisher: Cornell University Library, arXiv.org
• Subject: Accuracy / Datasets / Dermatology / Medical imaging / Ophthalmology / Reasoning / Vision