2014 Project Week:DWIDispersion

From NAMIC Wiki
Jump to: navigation, search
Home < 2014 Project Week:DWIDispersion

Key Investigators

  • Iowa: Hans Johnson
  • Iowa: Kent Williams
  • BWH: C-F Westin
  • BWH: Peter Savadjiev

Project Description


Implement the fiber dispersion computation method of Savadjiev et al. (MICCAI 2012) and the compressed sensing HARDI estimation method of Rathi et al. as Slicer4 extensions. Convert the currently existing Matlab implementation into C++/ITK and test it on various datasets.

Approach, Plan

  • Create a realistic test suite based on Huntington datasets for the Matlab code.
    • [HANS - Monday] Prepare test DWI data set for preparing data
  • Create a VTK reader/writer in matlab, in order to the test the matlab code.
    • [KENT -- DONE] git@github.com:BRAINSia/CompressedSensingDWI.git
    • [PETER - DONE] Build github version of readers/writers
    • [PETER - DONE] Manually test vtk readers/writers, and convert existing test to use the binary forms

  • Convert the Matlab code to C++/ITK
    • [HANS/PETER - DONE] Create design document (i.e. this page and paer/pencil) to describe the intended inputs and outputs.
    • [PETER -- DONE] - Write desired command line interface psuedocode "caluclateDispersion --inputTract <wholebrain.vtp> --mask <roi.nrrd> --outputTract <decoratedversionofinput.vtp> "
    • [HANS/PETER -- DONE] - Create shell program in psydocode/comments with XML and cxx files with description of basic IO programming needed
    • [KENT -- Tuesday/Wednesday] -- Fill in I/O portions of psuedocode and make a compilable application that reads the inputTract, breaks it apart into necessary algorithmic components, and writes out the outputTract.
  • Create a Slicer extension


  • Private git repository created at: git@github.com:BRAINSia/CompressedSensingDWI.git
  • Spent time debugging the vtk readers/writers for MATLAB, in order to test the MATLAB code.
  • Wrote XML interfaces for the fiber tract dispersion extension, and for the compressed sensing HARDI estimation.

DWIDispersion xml interfaces.png

  • Next steps: continue debugging the vtk readers/writers for MATLAB, and begin translating the algorithm implementation from MATLAB to C++.