r20240915

This is a release of the model-spinal-rootlets_mp2rage_r20240915 model providing semantic (i.e., level-specific) segmentation of dorsal and ventral nerve rootlets C2-C8 from MP2RAGE data. The orientation of the input image is expected to be RPI (right-posterior-inferior).
This model is able to segment nerve rootlets from inv1, inv2, and UNIT1 contrasts. In this release, the nnUNet output structure is kept intact.
Update by @valosekj: I re-uploaded the ZIP archive on 2024-11-13 to include the "image_orientation": "RPI" key-value pair. Context: #79

r20240730

This is a re-release of the model-spinal-rootlets_M5_r20240129 model providing semantic (i.e., level-specific) segmentation of dorsal nerve rootlets C2-C8. The orientation of the input image is expected to be LPI (left-posterior-inferior).
The model was originally released in r20240129. In this release, the nnUNet output structure is kept intact, thus allowing it to be used by sct_deepseg and also in the 3DSlicer software.

For details about the model can be found in https://doi.org/10.1162/imag_a_00218.

r20240523

Note

I updated the asset (model-spinal-rootlets_ventral_D106_r20240523.zip) of this release on 2024-09-05 to keep the nnUNet output structure intact in order to make the model compatible with sct_deepseg and 3DSlicer software.

This release contains the ventral D106 model (fold_all) providing semantic (i.e., level-specific) segmentation of dorsal and ventral nerve rootlets C2-C8. The orientation of the input image is expected to be LPI (left-posterior-inferior).
For details about the ventral model D106, see #43.

⚠️ Instructions on how to use the ventral model are available in this README.
⚠️ Please note that the ventral model is still under development and is not yet available in the Spinal Cord Toolbox (SCT).
⚠️ For the stable model for dorsal rootlets only, use SCT v6.2 or higher (please refer to this README).

What's Changed

• README.md: Add citation info for arxiv preprint by @joshuacwnewton in #34
• Update README with SCT v6.2 by @valosekj in #35
• Datasets V1 and V2 for training of the ventral model by @LouisThomasLapointe in #38
• Ltl/new datasets by @LouisThomasLapointe in #41
• Add codebase for PAM50 rootlets by @valosekj in #44
• Add instructions and script on how to use the ventral rootlets model by @valosekj in #47

New Contributors

• @joshuacwnewton made their first contribution in #34
• @LouisThomasLapointe made their first contribution in #38

Full Changelog: r20240129...r20240523

r20240129

This release contains the M5 model (fold_all) providing semantic (i.e., level-specific) segmentation of dorsal nerve rootlets C2-C8. The orientation of the input image is expected to be LPI (left-posterior-inferior).
For details about the model M5, see here: #23 (comment).

r20231108

A release related to the ISMRM 2024 abstract. It contains inter-rater_variability folder with scripts used for the inter-rater variability analysis.