TotalSpineSeg

3D Slicer extension for fully automatic spine MRI and CT segmentation using the TotalSpineSeg AI model.

Acknowledgement

This extension is based on the TotalSpineSeg tool. It is adapted from the SlicerTotalSegmentator extension.

If you use this extension in your research, please cite the following papers:

TotalSpineSeg:

Warszawer, Yehuda & Molinier, Nathan & Valosek, Jan & Benveniste, Pierre-Louis & Bédard, Sandrine & Shirbint, Emanuel & Mohamed, Feroze & Tsagkas, Charidimos & Kolind, Shannon & Lynd, Larry & Oh, Jiwon & Prat, Alexandre & Tam, Roger & Traboulsee, Anthony & Patten, Scott & Lee, Lisa Eunyoung & Achiron, Anat & Cohen-Adad, Julien. (2025). TotalSpineSeg: Robust Spine Segmentation with Landmark-Based Labeling in MRI. 10.13140/RG.2.2.31318.56649.

nnU-Net:

Isensee, F., Jaeger, P.F., Kohl, S.A.A. et al. nnU-Net: a self-configuring method for deep learning-based biomedical image segmentation. Nat Methods 18, 203–211 (2021). https://doi.org/10.1038/s41592-020-01008-z

Setup

1. Install 3D Slicer: Download and install the latest version of 3D Slicer.
2. Install Extension:
   • Via Extension Manager: Install TotalSpineSeg extension via the Extension Manager (once available).
   • Manual Installation:
     1. Download or clone this repository.
     2. Open 3D Slicer.
     3. Click on the search icon (magnifying glass) next to the Modules menu in the toolbar (or press Ctrl+F).
     4. Search for Extension Wizard and open it.
     5. Click on Select Extension.
     6. Navigate to and select the SlicerTotalSpineSeg folder (the root folder of this repository).
     7. Confirm that you want to load the extension.
3. GPU Setup (Recommended): A CUDA-capable GPU is highly recommended for fast processing. Ensure you have proper NVIDIA drivers installed.

Usage

1. Open 3D Slicer.
2. Switch to the TotalSpineSeg module (under Segmentation).
3. Installation (First Run):
   • If required Python packages are missing, you will see an Install Dependencies button instead of the main interface.
   • Click it to install the necessary dependencies (TotalSpineSeg, nnU-Net, PyTorch, etc.).
   • Once installation is complete, the main interface will appear.
4. Inputs:
   • Select the Input Volume (choose a loaded volume or load from file using ...).
   • Use the Eye and 3D icons next to the selector to toggle visibility.
   • (Optional) Select a Localizer segmentation (Step 1 or Step 2 output). This is the output of the model on a localizer image, used to ensure accurate instance segmentation on short FOV scans that do not contain C1 or Sacrum.
5. Outputs:
   • Select or create output nodes for the desired segmentations. You can also load existing files using the ... button next to each selector.
   • Each output row has Eye and 3D icons to quickly toggle visibility of results.
   • Available Outputs:
     • Step 2: Segmentation of canal, cord, and instance segmentation of vertebrae and discs.
     • Step 1: Segmentation of canal, cord, vertebrae (binary), and instance segmentation of discs.
     • Cord: Spinal cord soft segmentation (probability map).
     • Canal: Spinal canal soft segmentation (probability map).
     • Levels: Single-voxel vertebral levels at the posterior tip of each disc (similar to SCT).
   • Apply anatomical terminology: Check this to rename segments to standard anatomical names (e.g., "vertebrae_C1"). This applies on file load, node selection, or when toggled.
   • Isotropic output: Check this to keep the output at 1mm³ resolution (as used by the model) instead of resampling back to input space.
6. Click Apply.
   • On the first run, the module will download the model weights. This may take several minutes.
   • Subsequent runs will be faster.

Features

• Multi-Modality: Supports both MRI and CT scans.
• Comprehensive Segmentation:
  • Step 1: Segmentation of spinal canal, spinal cord, vertebrae (binary), and instance segmentation of discs.
  • Step 2: Segmentation of canal, cord, and instance segmentation of vertebrae and discs.
  • Localizer Support: Use model outputs from localizer images to handle short FOV scans (missing C1 or Sacrum).
• Anatomical Terminology: Automatically renames segments to standard anatomical terms.

Troubleshooting

• First Run Delay: The first run downloads large model files. Please be patient.
• CUDA Errors: Ensure your PyTorch installation matches your CUDA version. Use "PyTorch Util" module in Slicer to manage PyTorch installation if needed.
• Memory Issues: If you run out of memory, try cropping the volume using "Crop Volume" module before segmentation.