Difference between revisions of "2014 Summer Project Week: Pectoralis muscle segmentation"

From NAMIC Wiki
Jump to: navigation, search
 
(2 intermediate revisions by the same user not shown)
Line 26: Line 26:
 
<div style="width: 27%; float: left; padding-right: 3%;">
 
<div style="width: 27%; float: left; padding-right: 3%;">
 
<h3>Progress</h3>
 
<h3>Progress</h3>
* Worked on ensuring that the python packages that we need can be compatible with Slicer (thanks to Hans and Steve). Dealt with many compatibility issues between the code that was developed using Canopy python.
+
* Worked on ensuring that the python packages that we need can be compatible with Slicer (thanks to Hans Johnson and Steve Pieper). Dealt with many compatibility issues (numpy, scipy, scikit, pygco) between the code that was developed using Canopy python and Slicer's python .
* Got help with the image-based registration between the test and training cases from Brad using SimpleITK. He gave me an ipython notebook.
+
* Got help with the image-based registration between the test and training cases from Bradley Lowekamp using SimpleITK. He gave me an ipython notebook.
 +
* Incorporating the segmentation code into a python module (but without the atlas creation for now). It successfully runs bit we are still having trouble visualizing the output.
  
  
 
</div>
 
</div>
 
</div>
 
</div>

Latest revision as of 14:29, 27 June 2014

Home < 2014 Summer Project Week: Pectoralis muscle segmentation

Key Investigators

Rola Harmouche, Jorge Onieva, James Ross, Raul San Jose

Project Description

Objective

  • We have developed a method for the segmentation of the pectoralis muscle on chest CT images using graph cuts and a subject-tailored atlas. The atlas is built by selecting a subset of training data that shares high similarity with the test case, by making use of pairwise registrations. Our objective is to develop a slicer python module for the segmentation of pectoralis muscles on 2D CT images.
  • We would like to refine some parts of the segmentation process, particularly, the image-based registration between the test and training cases.
  • We want to modularize and abstract some components of the segmentation, particularly the atlas creation.
  • We also want to incorporate the segmentation module into a nipype work flow for its deployment in high performance computing environments.

Approach, Plan

  • We have previously developed a library of tools ported all of our available tools for the processing and the analysis of chest images (chest imaging platform) and have ported them into slicer CLIs.
  • The python module will make use of the previously developed CLIs within the chestimaging platform, and of recently developed python scripts.
  • We also aim to, if time permits, define nipype interfaces that utilize the python modules for workflows that can be used for clinical purposes and for parameter testing, and to use to those interfaces to find the optimal parameters for the pairwise registration.

Progress

  • Worked on ensuring that the python packages that we need can be compatible with Slicer (thanks to Hans Johnson and Steve Pieper). Dealt with many compatibility issues (numpy, scipy, scikit, pygco) between the code that was developed using Canopy python and Slicer's python .
  • Got help with the image-based registration between the test and training cases from Bradley Lowekamp using SimpleITK. He gave me an ipython notebook.
  • Incorporating the segmentation code into a python module (but without the atlas creation for now). It successfully runs bit we are still having trouble visualizing the output.