Research map / Image generation / Fast diffusion sampling
GenTract: Generative Global Tractography
Alec Sargood, Lemuel Puglisi, Elinor Thompson, Mirco Musolesi, Daniel C. Alexander
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use diffusion instead of flow matching objective (8 layers) (ablation) | GenTract · HCP test internal comparison (PyAFQ) | AFQ %P: 82.45 -> 85.0 | Improved |
| increase transformer layers from 4 to 8 (diffusion) (ablation) | GenTract · HCP test internal comparison (PyAFQ) | AFQ %P: 81.8 -> 85.0 | Improved |
| increase embedding dimension from 128 to 256 (ablation) | GenTract · HCP test internal comparison (PyAFQ) | AFQ %P: 80.6 -> 85.0 | Improved |
| increase embedding dimension from 256 to 512 (ablation) | GenTract · HCP test internal comparison (PyAFQ) | AFQ %P: 85.0 -> 79.3 | Hurt |
| increase DDIM inference steps from 5 to 10 (ablation) | GenTract · HCP test, inference steps | AFQ %P: 69.7 -> 82.9 | Improved |
| increase DDIM inference steps from 10 to 50 (ablation) | GenTract · HCP test, inference steps | AFQ %P: 82.9 -> 85.2 | No clear effect |
| use GenTract global generation instead of DDTracking | DDTracking · HCP test (clean) | BS %P: 35.20 -> 61.95 | Improved |
| use GenTract instead of TractOracle under Rician noise | TractOracle · HCP test (Rician noise) | BS %P: 22.06 -> 60.32 | Improved |
| use GenTract instead of DDTracking under low-res + noise | DDTracking · HCP test (low-res + noise) | BS %P: 4.44 -> 15.73 | Improved |
| use GenTract instead of TractOracle on out-of-domain TractoInferno | TractOracle · TractoInferno (low-res + noise) | BS %P: 9.74 -> 24.94 | Improved |
| use GenTract instead of DDTracking for bundle coverage | DDTracking · HCP test (clean) | BS bundles: 49.69 -> 36.62 | Hurt |
Closest papers by meaning
- TractoTransformer: Diffusion MRI Streamline Tractography using CNN and Transformer Networks (NeurIPS 2025 · similarity 0.90)
- AnaDiffusion: Anatomically Compositional Latent Diffusion for Controllable 3D Brain MRI Generation (NeurIPS 2026 · similarity 0.82)
- MAGNET: Manifold-Aware Graph Diffusion Network for Connectome Generation (NeurIPS 2026 · similarity 0.82)
- STREAM: Stochastic Riemannian Flow Matching with Anisotropic Decoder for Digital Histopathology Image Generation (NeurIPS 2026 · similarity 0.82)
- Equivariant spatio-hemispherical networks for diffusion MRI deconvolution (NeurIPS 2024 · similarity 0.82)
- Riemannian Flow Matching for Brain Connectivity Matrices via Pullback Geometry (NeurIPS 2025 · similarity 0.82)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
