![[2602.22545] DisQ-HNet: A Disentangled Quantized Half-UNet for Interpretable Multimodal Image Synthesis Applications to Tau-PET Synthesis from T1 and FLAIR MRI](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2602.22545] DisQ-HNet: A Disentangled Quantized Half-UNet for Interpretable Multimodal Image Synthesis Applications to Tau-PET Synthesis from T1 and FLAIR MRI
Summary
DisQ-HNet introduces a novel framework for synthesizing tau-PET images from MRI scans, enhancing interpretability and preserving anatomical details for Alzheimer's disease analysis.
Why It Matters
This research addresses the challenges of tau-PET imaging accessibility by providing an MRI-based alternative. It enhances the understanding of Alzheimer's disease pathology through improved image synthesis techniques, which can lead to better diagnostic tools and treatment strategies.
Key Takeaways
- DisQ-HNet synthesizes tau-PET from T1 and FLAIR MRI scans.
- The framework uses Partial Information Decomposition for better interpretability.
- It maintains high reconstruction fidelity and preserves disease-relevant signals.
- Modality-specific attribution is achieved through PID-based Shapley analysis.
- The method shows promise for improving Alzheimer's disease diagnostics.
Computer Science > Computer Vision and Pattern Recognition arXiv:2602.22545 (cs) [Submitted on 26 Feb 2026] Title:DisQ-HNet: A Disentangled Quantized Half-UNet for Interpretable Multimodal Image Synthesis Applications to Tau-PET Synthesis from T1 and FLAIR MRI Authors:Agamdeep S. Chopra, Caitlin Neher, Tianyi Ren, Juampablo E. Heras Rivera, Mehmet Kurt View a PDF of the paper titled DisQ-HNet: A Disentangled Quantized Half-UNet for Interpretable Multimodal Image Synthesis Applications to Tau-PET Synthesis from T1 and FLAIR MRI, by Agamdeep S. Chopra and 4 other authors View PDF HTML (experimental) Abstract:Tau positron emission tomography (tau-PET) provides an in vivo marker of Alzheimer's disease pathology, but cost and limited availability motivate MRI-based alternatives. We introduce DisQ-HNet (DQH), a framework that synthesizes tau-PET from paired T1-weighted and FLAIR MRI while exposing how each modality contributes to the prediction. The method combines (i) a Partial Information Decomposition (PID)-guided, vector-quantized encoder that partitions latent information into redundant, unique, and complementary components, and (ii) a Half-UNet decoder that preserves anatomical detail using pseudo-skip connections conditioned on structural edge cues rather than direct encoder feature reuse. Across multiple baselines (VAE, VQ-VAE, and UNet), DisQ-HNet maintains reconstruction fidelity and better preserves disease-relevant signal for downstream AD tasks, including Braak stag...