IGMedSeg-Slicer

3D Slicer extension for interactive correction，Integrates Total-Segmentator to achieve one-stop segmentation

IGMedSeg.mp4

Setup

• Start 3D Slicer
• Go to Sample Data module and load CTA Abdomen (Panoramix) data set
• Go to Extension Wizard
• Click Select Extension
• Select IGMedSeg folder
• Go to IGMedSeg module
• Select Input volume -> Panoramix-cropped
• Select Segmentation -> Create new segmentation
• Click Apply
  • When this module is used the first time:
    • It needs to download and install PyTorch and TotalSegmentator Python packages and weights for the AI models. This can take 5-10 minutes and several GB disk space.
  • Expected computation time:
    • With CUDA-capable GPU: 20-30 seconds in fast mode, 40-50 seconds in full-resolution mode.
    • Without GPU: 1 minute in fast mode, 40-50 minutes in full-resolution mode.
• To display the segmentation in 3D: click the Show 3D button
• Select a Segment
• Click AI Edit
• Click on the Red、Yellow、Green interface to perform interactive correction

Advantage

1. Drag-and-Drop Direct Manipulation (Direct Manipulation) By dragging control points with the mouse to directly adjust the boundaries of closed polygons, real-time feedback and incremental modifications are achieved, realizing "what you see is what you get". This avoids the inefficiency of multiple discrete clicks. Moreover, the drag-and-drop interaction is immediately reflected in the slice mask, adhering to the principle of "direct manipulation" interface.
2. Local Correction, Preserving Other Areas Only the current target connected domain undergoes mask difference operation, while other segmented islands remain completely unaffected, thus avoiding the problems of whole block erasure or accidental deletion.
3. Pure In-Memory Rapid Update Utilizing NumPy and VTK PolyData to directly modify the single-slice mask in local memory, there is no need for recalculation of the entire volume, resulting in excellent performance and extremely low latency. Additionally, the model is lightweight, featuring high computational efficiency and speed.
4. No Server Required, No Network Dependency The entire process runs locally within Slicer, without the need for additional installation or startup of AI services, simplifying deployment and enhancing stability.

Acknowledgements

Wasserthal J., Meyer M., , Hanns-Christian Breit H.C., Cyriac J., Shan Y., Segeroth, M.: TotalSegmentator: robust segmentation of 104 anatomical structures in CT images. https://arxiv.org/abs/2208.05868

https://github.com/lassoan/SlicerTotalSegmentator