Research map / Probabilistic and scientific modeling / Differential equations and PDEs
Two-Parameter Flow Map Learning for Continuous-Time Diffeomorphic Image Registration
Mohammadjavad Matinkia, Nilanjan Ray
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 |
|---|---|---|---|
| add cocycle regularization with lambda=10 (vs lambda=0) (ablation) | TPFM-DIR · OASIS | Dice (%): 79.48 -> 87.08 | Improved |
| add cocycle regularization (lambda=10 vs 0) (ablation) | TPFM-DIR · OASIS | folding |J|<0 (%): 3.1263 -> 0.0019 | Improved |
| add cocycle regularization (lambda=10 vs 0) (ablation) | TPFM-DIR · LungCT | TRE (mm): 4.02 -> 2.19 | Improved |
| increase lambda from 10 to 20 (ablation) | TPFM-DIR · OASIS | Dice (%): 87.08 -> 85.02 | Hurt |
| increase lambda from 10 to 20 (ablation) | TPFM-DIR · OASIS | folding |J|<0 (%): 0.0019 -> 0.0 | Improved |
| reduce lambda from 10 to 5 (ablation) | TPFM-DIR · ACDC | Dice (%): 87.42 -> 84.52 | Hurt |
| compose 4 local maps at inference instead of single step (ablation) | TPFM-DIR · OASIS | folding | Improved |
| train CorrMLP and TransMorph in TPFM-DIR setting | CorrMLP/TransMorph · OASIS, IXI, Mindboggle, AbdomenCT | Dice and folding | Improved |
| use TPFM-DIR instead of scaling-and-squaring on same backbones | TransMorph, CorrMLP · OASIS, IXI, Mindboggle, AbdomenCT | Dice | Improved |
| use TPFM-DIR over best prior method | TPFM-DIR · OASIS | Dice (%): 85.07 -> 87.08 | Improved |
| use TPFM-DIR over best prior method | TPFM-DIR · LungCT | TRE (mm): 2.48 -> 2.19 | Improved |
| avoid numerical integration in training | TPFM-DIR · OASIS inference time | inference time (s): 0.96 -> 0.38 | Improved |
Closest papers by meaning
- Learning Diffeomorphism for Medical Image Registration with Time-Embedded Architectures Using Semigroup Regularization (CVPR 2026 · similarity 0.92)
- Collocation for Diffeomorphic Deformations in Medical Image Registration (TPAMI 2017 · similarity 0.89)
- Learning Deformable Image Registration From Optimization: Perspective, Modules, Bilevel Training and Beyond (TPAMI 2021 · similarity 0.88)
- Riemannian Flow Matching for Brain Connectivity Matrices via Pullback Geometry (NeurIPS 2025 · similarity 0.85)
- How to build a consistency model: Learning flow maps via self-distillation (NeurIPS 2025 · similarity 0.85)
- Uncertainty Estimation for Pretrained Medical Image Registration Models via Transformation Equivariance (NeurIPS 2026 · similarity 0.85)
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.
