Reproduce “Towards All-in-One Medical Image Re-Identification” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Yuan Tian, Kaiyuan Ji, Rongzhao Zhang, Yankai Jiang, Chunyi Li, Xiaosong Wang, Guangtao Zhai
The Modality-adaptive Medical Identifier (MaMI) is a framework that re-identifies patients across diverse medical imaging modalities, integrating medical domain knowledge through continuous adaptation and pre-trained foundation models. It consistently outperforms existing methods on 11 datasets, enabling improved history-augmented healthcare and medical privacy protection.
Key insights
- Continuous Modality-based Parameter Adapter (ComPA) effectively enables a single model to adapt to diverse medical imaging modalities without explicit modality labels, outperforming discrete or fixed adaptation methods.
- Integrating Medical Foundation Models (MFMs) provides valuable medical prior knowledge, enhancing re-identification performance by focusing on anatomically relevant features and improving generalization.
- A unified framework can successfully perform medical image re-identification across a wide range of modalities, including cross-modality matching, addressing challenges like intra-patient variability and inter-patient similarity.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.