ProbMED: A Probabilistic Framework for Medical Multimodal Binding (ICCV 2025)
Probabilistic Modality-Enhanced Diagnosis (ProbMED), a multi-modal Med-VLPM that employs probabilistic contrastive learning to model distributions over embeddings rather than fixed-point, deterministic estimates. ProbMED aligns four distinct modalities—chest X-rays, electrocardiograms, echocardiograms, and clinical text—into a unified probabilistic embedding space.
Installation
Clone the GitHub repository and install dependencies, instructions are found in the repo:
git clone [email protected]:mcintoshML/probMED.git
cd probMED
pip install -r requirements.txt
Full Code Release
The model weights and inference is available with this code base.
We plan to release the full training and evaluation codebase upon the clinical journal submission to facilitate reproducibility, please stay tuned!
License
This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License (CC BY-NC-ND 4.0).
You may share this work for non-commercial purposes, with proper attribution, but you may not modify it or use it commercially.
Citation
If you use ProbMED in your research (ICCV 2025), please cite:
@article{gao2025probmed,
title={ProbMed: A Probabilistic Framework for Medical Multimodal Binding},
author={Gao, Yuan and Kim, Sangwook and You, Jianzhong and McIntosh, Chris},
journal={arXiv preprint arXiv:2509.25711},
year={2025}
}