2012 Winter Project Week:DWIPhantom

From NAMIC Wiki
Jump to: navigation, search
Home < 2012 Winter Project Week:DWIPhantom

Key Investigators

  • UNC: Gwendoline Roger, Yundi Shi, Clement Vachet, Martin Styner, Sylvain Gouttard


Current software phantoms are quite abstract, quite far from human brain. The goal of the project is to create a software phantom that is human brain like for evaluating tractography algorithms. It will allow for simulating pathology, such as tumors, TBI, lesions.

Approach, Plan

Our approach is composed of different steps:

  • creation of high resolution atlas:
    • 6 young adults scanned at 1.5mm3, 42 directions
    • High resolution DWI atlas
    • Full brain filtered two tensor tractography (Millions of fibers)
  • co-registration of structural atlas with shape space:
    • 100 healthy (20 in each 18-29, 30-39, 40-49, 50-59, and 60+)
    • Isomap vs (PCA + local mean)
  • creation of "random-sample" phantoms in shape space:
    • Pathology simulation here
  • application to fiber geometry in atlas space
  • creation of DWI with different models:
    • Initial model is CHARMED only


  • Debugged my program which create the DWI phantom:
    • Tried to find parameters that fit well with the model we choose for creation of DWI
    • Obtained first results for DWI phantom with corpus callosum that I need to validate and then improve
  • Discussed the algorithm we use for tractography comparisons with people I met at the meeting

Delivery Mechanism