2008 Summer Project Week:DWI-DTI PrepTools

Return to Project Week Main Page

Key Investigators

• UNC: Heather Cody Hazlett, Martin Styner, Zhexing Liu
• Utah: Casey Goodlett, Guido Gerig, Tom Fletcher

Objective

We are developing a tool for preparing diffusion weighted and diffusion tensor image data.

Details of the tool functionality were discussed thus far between UNC and Utah. The discussions have been summarized in it's event page:

• The tool will allow a "protocol" based execution via parameter files. The parameter files contain all the settings/preferences/options and this file can be created using the tool's GUI
• This tool will handle the conversion of DICOM data to volumetric image formats (NRRD) using existing ITK-NAMIC classes
• The tool performs automated quality control, corrects data if possible, possibly rejects parts of the data, and provides a reporting facility. The following functionalities will be supported:
  • Motion correction (using Utah modules)
  • Eddy current correction (using Utah modules)
  • Interpolation of data, e.g. to upinterpolate to isotropic data
  • Quality Control
    • Rejection of interslice brightness, white or black slice at level of direction images
    • Local plausibility check with reconstructed DTI. Reject direction image if implausable results
    • FA/MD histogram comparison with atlas
    • Local dot product of principal direction with atlas,
    • Full brain tracking to check for possible flips, permutations of direction axes
    • Focus QC on white matter only, segment WM from baseline image via Slicer-itkEMS implementation
• Furthermore, DTI data is generated (using Slicer/UNC modules), as well as directly derived images, such as FA/MD etc
• Fltk vs KWWidget: Is Fltk appropriate here?

Approach, Plan

The tool will allow a "protocol" based execution via parameter files. The parameter files contain all the settings/preferences/options and this file can be created using the tool's GUI. This tool will handle the conversion of DICOM data to volumetric image formats (NRRD), perform automated quality control, correct data if possible, possibly reject parts of the data and provide a reporting facility.

Progress