DBP2:MIND:RoadmapProject

Return to Project Week Main Page

Example Results of Lesion Classification

Key Investigators

• H. Jeremy Bockholt
• Mark Scully
• Ross Whitaker
• Steve Pieper
• Vincent Magnotta
• Kilian Pohl
• Brad Davis
• Marcel Prastawa
• Guido Gerig
• Sonja Pujol

Objective

Finish the roadmap module, design and implement a testing strategy and work on the end-to-end tutorial for white matter lesion classification

Approach, Plan

The Slicer3 module will allow co-registration of user-selected T1, T2, and FLAIR images. The module will automatically label all brain voxels into either gray, white, csf, or lesion as tissues types. Lesions will be automatically clustered such that the anatomical location and volume can be summarized.

Review and summarize results of each of classification approaches thus far.

Sketch out methods paper and clinical application paper.

Finalize content to be presented for SFN 2008 poster/presentation.

Work with Sonja on timeline and process for creating tutorial and process for release of tool and data (SFN 2008)?

Answer question, where do roadmap projects that share data, share it from, publication database, BIRN project, XNAT Central, other service, provide own?

Progress

Plan is to release end-to-end tutorial at the 2008 Society for Neuroscience Annual meeting.

Deliverables for this week

1. Intensity Standardization filter based on a bspline deformation of joint histograms to minimize KL-divergence.
2. Integrate visualization of ROIs that lesions impinge upon.
3. interested in working with the unbiased atlas work for lesion location (talk with sylvain)
4. parametization of EM-Segment (talk with kilian)

In terms of fine tuning lesion classification:

1. We worked with Kilian on parameterization of EM-Segment
2. We worked with Marcel on using his method for lesion classification
3. We worked with Vince Magnotta on his method for lesion classification