r20250909

What's Changed

• Integrating more contrasts in the MSD dataset by @plbenveniste in #56
• Evaluation of existing SCT methods by @plbenveniste in #64
• Updated dataset analysis code by @plbenveniste in #66
• Training on compute canada by @plbenveniste in #57
• Code for article writing by @plbenveniste in #63
• Code reorganization by @plbenveniste in #69

Full Changelog: r20250626...r20250909

Release content

This release was done for the 2025 article submission: "TEXT TO BE INSERTED HERE". It contains the weights of the model and the code used to train, test and evaluate the model. The data used is displayed in dataset_2025-04-15_seed42.json. The model was trained using the following nnUNet plans nnUNetResEncUNetL1x1x1_Model2_Plans.json.

The dataset versions are the following:
basel-mp2rage: commit c54b3fb0fad8fa07baedb724ce9d919a73854e6e
bavaria-quebec-spine-ms-unstitched: commit f44f92b39eadd4276f0f689c96e63954051aea32
canproco: commit 7daf5ed2abc4304b45d6c81364ed06df0fc1f7c4
ms-basel-2018: commit cfddb0ed1aee701ef512631edb6016bcc29ef47b
ms-basel-2020: commit 92ec31c142bc01cc37957742fcf318d55b7ef80d
ms-karolinska-2020: commit d81e0c7a5aa8092df274f8d88827b64b4e1b4dc4
ms-nmo-beijing: commit 3359fc242e02367bb2373fd3f8809c1ea8624ced
ms-nyu: commit fd9f1880d639e7ef169daa11396c1f1b1687fde9
ms-rennes-mp2rage: commit 4a6b2d32bf7dca400d96a17d71eb749b0db4b4eb
nih-ms-mp2rage: commit 14b7c9fbc73613ae49cdbacb555a55d8fb071605
sct-testing-large: commit 132f9c8387335601960fbc2968f59b626f8e285c
umass-ms-ge-excite1.5: commit 18125775b55a21aac1ee8fcb0cc60cd1622e5349
umass-ms-ge-hdxt1.5: commit 503f28b65b0d3cdbb99f790daec4a976b35c0ffc
umass-ms-ge-pioneer3: commit 89cf98a82f951c96cd9517cc5e201a36be490dfd
umass-ms-siemens-espree1.5: commit ab24170b5a0e1fbbb63d5631d1a7052ad0197982

The version of the canproco code is the following (for the exclude file)
canproco: commit 52d5b40bd406aa467bc60013f139bc901f22a1a2

SCT version: commit dd0faf4df1a6750e7fc107b8dbb73f96cf1bf507

r20250626

What's Changed

• Removing nnunet from requirements.txt by @plbenveniste in #60

Full Changelog: r20250219...r20250626

Release content:

This release corresponds to the code used for the submission for ESMRMB 2025 “Reinforcing the generalizability of spinal cord multiple sclerosis lesion segmentation models”. It contains dataset_2025-01-17_seed42.json which stores the data split used in this project a well as the fold of the model.

The datasets used are:

• basel-mp2rage: c54b3fb0fad8fa07baedb724ce9d919a73854e6e
• bavaria-quebec-spine-ms-unstitched: f44f92b39eadd4276f0f689c96e63954051aea32
• canproco: 7daf5ed2abc4304b45d6c81364ed06df0fc1f7c4
• ms-basel-2018: cfddb0ed1aee701ef512631edb6016bcc29ef47b
• ms-basel-2020: 92ec31c142bc01cc37957742fcf318d55b7ef80d
• ms-karolinska-2020: d81e0c7a5aa8092df274f8d88827b64b4e1b4dc4
• ms-nmo-beijing: 3359fc242e02367bb2373fd3f8809c1ea8624ced
• ms-nyu: fd9f1880d639e7ef169daa11396c1f1b1687fde9
• ms-rennes-mp2rage: 4a6b2d32bf7dca400d96a17d71eb749b0db4b4eb
• nih-ms-mp2rage: 14b7c9fbc73613ae49cdbacb555a55d8fb071605
• sct-testing-large: 0299da1367ac0958e94d3af39a14a0382f14de00
• umass-ms-ge-excite1.5: 18125775b55a21aac1ee8fcb0cc60cd1622e5349
• umass-ms-ge-hdxt1.5: 503f28b65b0d3cdbb99f790daec4a976b35c0ffc
• umass-ms-ge-pioneer3: 89cf98a82f951c96cd9517cc5e201a36be490dfd
• umass-ms-siemens-espree1.5: ab24170b5a0e1fbbb63d5631d1a7052ad0197982

The results of the poster were computed using the “checkpoint_final.pth” models.

r20250219

What's Changed

• Dataset aggregation by @plbenveniste in #46
• Add nnUNet training by @plbenveniste in #47
• Dataset analysis of MSD dataset by @plbenveniste in #48
• Evaluation of existing models on the different sets by @plbenveniste in #49

Full Changelog: r20241101...r20250219

Release content:

This release corresponds to the code used for the ACTRIMS/NAIMS poster 2025 “Automatic multi-contrast MRI segmentation of spinal cord lesions”. It contains dataset_2025-01-17_seed42.json which stores the data split used in this project a well as the fold of the model.

The pipeline used for training is the following:

Because of size constraints, the models were stored in multiple files containing each split. To use them, they should be assembled in the following way:

The datasets used are:

• basel-mp2rage: c54b3fb0fad8fa07baedb724ce9d919a73854e6e
• bavaria-quebec-spine-ms-unstitched: f44f92b39eadd4276f0f689c96e63954051aea32
• canproco: 7daf5ed2abc4304b45d6c81364ed06df0fc1f7c4
• ms-basel-2018: cfddb0ed1aee701ef512631edb6016bcc29ef47b
• ms-basel-2020: 92ec31c142bc01cc37957742fcf318d55b7ef80d
• ms-karolinska-2020: d81e0c7a5aa8092df274f8d88827b64b4e1b4dc4
• ms-nmo-beijing: 3359fc242e02367bb2373fd3f8809c1ea8624ced
• ms-nyu: fd9f1880d639e7ef169daa11396c1f1b1687fde9
• ms-rennes-mp2rage: 4a6b2d32bf7dca400d96a17d71eb749b0db4b4eb
• nih-ms-mp2rage: 14b7c9fbc73613ae49cdbacb555a55d8fb071605
• sct-testing-large: 0299da1367ac0958e94d3af39a14a0382f14de00
• umass-ms-ge-excite1.5: 18125775b55a21aac1ee8fcb0cc60cd1622e5349
• umass-ms-ge-hdxt1.5: 503f28b65b0d3cdbb99f790daec4a976b35c0ffc
• umass-ms-ge-pioneer3: 89cf98a82f951c96cd9517cc5e201a36be490dfd
• umass-ms-siemens-espree1.5: ab24170b5a0e1fbbb63d5631d1a7052ad0197982

The results of the poster were computed using the “checkpoint_best.pth” models.

r20241101

What's Changed

• Evaluation of existing SCT methods by @plbenveniste in #35
• nnUNet training and evaluation by @plbenveniste in #36
• Dataset aggregation by @plbenveniste in #37

New Contributors

• @plbenveniste made their first contribution in #35

Full Changelog: https://github.com/ivadomed/ms-lesion-agnostic/commits/r20241101

Content

This release contains the code, the data used and the model for the ACTRIMS 2025 abstract submission.

The datasets versions are the following:

• basel-mp2rage: commit ddd0d555854d3b2cac205583298addc8f0b45ac2
• canproco: commit 248be65fda551479ce0d3f9f644188cfca1248f0
• ms-basel-2020: commit 92ec31c142bc01cc37957742fcf318d55b7ef80d
• nih-ms-mp2rage: commit 8187361e4f5143ffc6c8d93750a34a57e424a3a8
• umass-ms-ge-excite1.5: commit 18125775b55a21aac1ee8fcb0cc60cd1622e5349
• umass-ms-ge-pioneer3: commit 89cf98a82f951c96cd9517cc5e201a36be490dfd
• bavaria-quebec-spine-ms-unstitched: commit 3e0819435f5b99ddad70aee6c50fdc0db035434b
• ms-basel-2018: commit cfddb0ed1aee701ef512631edb6016bcc29ef47b
• ms-nmo-beijing: commit ac134315b48cb9efcb72cd276bbceeb286103442
• sct-testing-large: commit c26a5d690e2ced34bd5dea61cab66a7cb0eaebed
• umass-ms-ge-hdxt1.5: commit 503f28b65b0d3cdbb99f790daec4a976b35c0ffc
• umass-ms-siemens-espree1.5: commit ab24170b5a0e1fbbb63d5631d1a7052ad0197982

The dataset split for training/testing the model can be found in the file dataset_2024-07-24_seed42_lesionOnly.json.

The model is attached below: it is a 3D nnUNet trained with the ResEnc planner.
The model was trained on RPI images. For inference purposes, the dataset.json file was modified to add "image_orientation": "RPI".