SlicerAISegmentation

This 3D Slicer Extension contains modules on loading CT data (3D volume) and auto-contouring (a.k.a segmenting) multiple organs in the Head and Neck (HaN) Region

Modules

3D Slicer as a tool for medical image analysis is made of many modules, each implementing a unique functionality. A 3D Slicer extension can be composed of multiple modules.

Module 1 - AISegmentation

This is a 3D Slicer Python Scripted Module which accept a 3D CT volume. Mostly, this module handles the UI related aspects of this extension.

Module 2 - AISegmentation CLI

This is a 3D Slicer Command Line Interface(CLI). This can asynchronously run a Neural Network (so as to not block the main 3D Slicer UI thread) and return predicted contours. To be run asynchronously, it has to be called from within module 1

Installation

1. Clone this repository
2. Download 3DSlicer [Version >= 4.11]
3. After installation
   • Open the application
   • Ctrl + F = "Welcome to Slicer".
     • Ctrl + F is a shortcut to find a module within 3D Slicer
   • Click on Customize Slicer in the "Welcome to Slicer" module
   • In the window that open, click on Modules
   • Click the double arrow (>>) correponding to Additional Module Paths
     • Click on Add
   • Add the AISegmentation and AISegmentationCLI module folders
   • Restart 3D Slicer
4. After restart
   • Ctrl + F = "AISegmentation"
   • Click on Load CT Scan and select a CT scan (as a 3D volume)
   • Click on Load AI Contouring Model (and load a tensorflow model directory)
     • After a model is selected, the application show progress bars.
     • Once the progress bars are finised, the predicted contours are shown in the 3D Viewer

Pending

• Adding DSC score (in Module 1)
• Fixed Windowing on HaN CT scans (in Module 1)

References

• 3D Slicer Python FAQ