EVLF-FM: Explainable Vision Language Foundation Model for Medicine
Abstract
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.
Cite
@article{arxiv.2509.24231,
title = {EVLF-FM: Explainable Vision Language Foundation Model for Medicine},
author = {Yang Bai and Haoran Cheng and Yang Zhou and Jun Zhou and Arun Thirunavukarasu and Yuhe Ke and Jie Yao and Kanae Fukutsu and Chrystie Wan Ning Quek and Ashley Hong and Laura Gutierrez and Zhen Ling Teo and Darren Shu Jeng Ting and Brian T. Soetikno and Christopher S. Nielsen and Tobias Elze and Zengxiang Li and Linh Le Dinh and Hiok Hong Chan and Victor Koh and Marcus Tan and Kelvin Z. Li and Leonard Yip and Ching Yu Cheng and Yih Chung Tham and Gavin Siew Wei Tan and Leopold Schmetterer and Marcus Ang and Rahat Hussain and Jod Mehta and Tin Aung and Lionel Tim-Ee Cheng and Tran Nguyen Tuan Anh and Chee Leong Cheng and Tien Yin Wong and Nan Liu and Iain Beehuat Tan and Soon Thye Lim and Eyal Klang and Tony Kiat Hon Lim and Rick Siow Mong Goh and Yong Liu and Daniel Shu Wei Ting},
journal= {arXiv preprint arXiv:2509.24231},
year = {2025}
}