Projects:DTIPathOfInterest

Back to NA-MIC_Collaborations, MGH Algorithms

Projects DTI Path of Interest

Our objective is to provide software deliverable with robust support for input images with varying slice prescription, voxel size, tensor measurement frame, etc. Ensure compatibility with Slicer file formats and Dartmouth tensor data conventions.

Description

Project DTI path of interest (POI) analysis

Use case "I'd like to be able to find the optimal path between two regions in a tensor image." - Saykin

Impact Medium (Saykin)

Difficulty Low (Snyder)

Vision Software tool which can find optimal and sub-optimal paths between regions of interest in a tensor image, and produce associated statistics and probability density images. Should work on output of Slicer DTI analysis and ROI drawing tools. Tool should generate probability density image, sample statistics from this image, and from user defined image w/ same geometry.

Win conditions

'MGH' Improved robustness and portability of existing POI tool.

'Dartmouth' Software deliverable.

Test Data

Six healthy individuals from the Dartmouth DTI data set.

Steps

1. Requirements extraction (Snyder/Saykin): done
2. Create Wiki project page (Snyder): done
3. Port MGH POI tool to Dartmouth compatible version (Snyder/Tuch): done
4. Test output visualization in Slicer (Snyder): done
5. Process test data with MGH tensor reconstruction and POI analysis tools; use for port validation (Snyder): done
6. Software deliverable installation at Dartmouth (West/Snyder): done
7. Port to ITK (Jen): done
8. Test on expanded data set (West/Jen): in queue
9. Develop path initialization (Jen): in progress
10. Develop model selection for selecting optimal number of control points (Yendiki): in queue
11. Develop algorithm for perturbing control points based on neighboring tensors (Yendiki): in progress
12. Submit to the Insight Journal (Jen): in queue

Progress

• Dartmouth Dicom series -> NIFTI image conversion toolset completed.
• Tensor reconstruction workflow identified and tested.
• Path of interest tool ported.
• Installed at Dartmouth
• Currently in testing/use by John West

• Use case: 'Specify 2 points in a diffusion image and tell how connected they are.'
• Difficulty: High
• Impact: High

Key Investigators

• MGH: Fischl, Yendiki, Helmer, Saykin, West, Jen
• BWH: Gordon Kindlmann, Raul San Jose, Steve Pieper

Project Week Results: Jan 2007 (Slicer Integration), Jan 2007 (Image Format)