![[2512.09185] Learning Patient-Specific Disease Dynamics with Latent Flow Matching for Longitudinal Imaging Generation](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2512.09185] Learning Patient-Specific Disease Dynamics with Latent Flow Matching for Longitudinal Imaging Generation
Summary
The paper presents a novel framework, $ ext{Δ}$-LFM, for modeling patient-specific disease dynamics using latent flow matching, enhancing the interpretability of disease progression in longitudinal imaging.
Why It Matters
This research addresses critical challenges in understanding disease progression, which is vital for early diagnosis and personalized treatment. By improving the alignment of patient data in latent space, it offers a more interpretable framework that could significantly impact clinical practices and research methodologies in medical imaging.
Key Takeaways
- Introduces $ ext{Δ}$-LFM for modeling disease dynamics in imaging.
- Utilizes flow matching to enhance interpretability of disease progression.
- Addresses alignment issues in latent space for patient-specific data.
- Demonstrates strong empirical performance across MRI benchmarks.
- Offers a new framework for visualizing and interpreting disease dynamics.
Computer Science > Computer Vision and Pattern Recognition arXiv:2512.09185 (cs) [Submitted on 9 Dec 2025 (v1), last revised 13 Feb 2026 (this version, v3)] Title:Learning Patient-Specific Disease Dynamics with Latent Flow Matching for Longitudinal Imaging Generation Authors:Hao Chen, Rui Yin, Yifan Chen, Qi Chen, Chao Li View a PDF of the paper titled Learning Patient-Specific Disease Dynamics with Latent Flow Matching for Longitudinal Imaging Generation, by Hao Chen and 4 other authors View PDF HTML (experimental) Abstract:Understanding disease progression is a central clinical challenge with direct implications for early diagnosis and personalized treatment. While recent generative approaches have attempted to model progression, key mismatches remain: disease dynamics are inherently continuous and monotonic, yet latent representations are often scattered, lacking semantic structure, and diffusion-based models disrupt continuity with random denoising process. In this work, we propose to treat the disease dynamic as a velocity field and leverage Flow Matching (FM) to align the temporal evolution of patient data. Unlike prior methods, it captures the intrinsic dynamic of disease, making the progression more interpretable. However, a key challenge remains: in latent space, Auto-Encoders (AEs) do not guarantee alignment across patients or correlation with clinical-severity indicators (e.g., age and disease conditions). To address this, we propose to learn patient-specific la...
