Repository navigation
Data and pretrained model weights for CineDiff cine MRI reconstruction.
Files
| File | Size | Contents |
|---|---|---|
cinediff-data.zip |
~1.1 GB | Prepared OCMR cases and data/cases.json: 56 fully sampled cases (14 slices × R8/R12/R16/R20) in data/fs/ and 15 prospectively undersampled cases in data/us/ |
model-192000.pt |
286 MB | Pretrained CineDiff model weights (milestone 192000) |
Usage
Clone the repository, then download and unpack these files:
- unzip
cinediff-data.zipin the repository folder, so it createsdata/ - save
model-192000.ptinweights/
Then reconstruct a case, for example:
python reconstruct.py --case fs_0008_1_5T_sli0_R16 --output-root outputs/example
Checksums (SHA256)
model-192000.pt:53b9239eb68eeb021115bedd7c21d4d8a94864eef7c05f1069a13a9743dc1535