fidoptanatomix + ConvexAdam
ConvexAdam on anatomix and MIND-SSC features, the registration pipeline of the anatomix paper with its command-line defaults.
RegistrationGPUversion 4c86aee, weights 842cfe9every datasetMIT
Run it
With its KonfAI app, at the benchmark's settings (the fidopt app repositories are private for now: a Hugging Face token with access opens them):
pip install konfai-apps konfai-apps infer fideus-labs/fidopt-registration:Anatomix_ConvexAdam -i fixed.nii.gz -i moving.nii.gz -o out --gpu 0
The app on Hugging Face: its model, its settings and how it is scored.
Where it works
Its place on each kind of problem it has results on, among the methods ranked there. A method that leads one problem can trail another: read the one closest to yours.
| Problem | Place | Ranked | Datasets |
|---|---|---|---|
| Light-sheet whole mouse brains, two channels, known warp (out of core) | 1 | of 2 | 1 of 1 |
| Light-sheet microscopy of whole mice, between animals | 2 | of 2 | 1 of 1 |
| Pediatric T1 MRI with automatic regional labels | 2 | of 4 | 1 of 1 |
| Abdomen CT, same patient | not ranked yet | 1 of 3 | |
| Two modalities, same patient | not ranked yet | 1 of 5 | |
| Learn2Reg official validation pairs | not ranked yet | 1 of 4 |
Results
| Dataset | # | Read by | Score | Change | SDlogJ | Failed | Time s |
|---|---|---|---|---|---|---|---|
| ACDC | 2 | Dice | 0.672 | +0.114 | 0.022 | 0 | 9 |
| Pediatric T1 MRI with and without previous ADHD diagnosis | 2 | Dice | 0.593 | +0.182 | 0.055 | 0 | 29 |
| Mouse whole-body light-sheet (AIMOS) | 2 | Dice | 0.320 | +0.108 | 0.096 | 0 | 46 |
| AIND SmartSPIM whole mouse brains, two channels | 1 | TRE µm | 58.73 | −93.96 | 0.072 | 0 | 18 |
| Abdominal-DIR-QA | 4 | Dice | 0.575 | +0.143 | 0.042 | 5 | 724 |
| ThoraxCBCT (Learn2Reg 2023), validation | 5 | Dice | 0.511 | +0.167 | 0.096 | 0 | 117 |
- Reference
- Dey et al., ICLR 2025