2014 Project Week:TBISegmentation

• 

  Projects List

• 

  Initial bounding box

• 

  Self training update

• 

  Active learning update

• 

  Final alpha map

• 

  Segmentation of normal tissue classes

Key Investigators

• GE: Marcel Prastawa
• Utah: Bo Wang, Guido Gerig
• USC: Andrei Irimia, Jack Van Horn

Project Description

Objective

• Develop a pipeline combining our TBI segmentation algorithm with other Slicer modules.
• Study and gain understanding of the manual process for segmenting TBI.
• Test the algorithm with users from USC.

Approach, Plan

• Wrap existing code into a Slicer extension.
• Determine approaches for user interaction for semi-automatic segmentation.
• Test and evaluate Slicer modules for TBI processing:
  • Intra and inter time points coregistration.
  • Initial atlas alignment using affine transformation.
  • Skull stripping to obtain initial brain mask.

Progress

• Ran the command line tool with Andrei step by step to show how it works and test it.
• Meeting with Jack and Andrei to discuss sharing the current command line tool with them.
• Joined the TBI DBP meeting to discuss how to integrate the current command line tool into Slicer 3D.
  • Stephen and Steve recommended to do a Python scripting module and talk to Jean-Christophe.
  • Talked to Jean-Christophe and he pointed me a tutorial the wiki of breakout session Slicer4Python.
  • Tried to understand the Slicer4Python tutorial and followed it step by step (still working on).
  • Talked to other Slicer developers to ask for help and suggestions.

Reference

• 4D-PARSeR (Pathological Anatomy Regression via Segmentation and Registration) Code Repo
• 4D Active Cut
• Modeling 4D changes in TBI