ITK Roadmap 2007 2008/2008-01-17

Status Meeting

Attendees

• Matt Turek, Luis Ibanez, Stephen Aylward

Purpose

• Balance goals of ITK and BW NAC (itk registration optimization)

ITK To-Dos

1. Release in February
   • Move articles from IJ to ITK/Review
   • Move articles from ITK/Review to proper place in ITK
   • Bug fixes
   • Leaders: Luis, Andinet, Matt
2. Confirm ITK operation on VSExpress 2008
   • Leaders: ?
3. RIRE-style projects in conjunction with the IJ
   • Leaders: ?
4. GPU methods in ITK pipeline
   • Leaders: ?
5. Data providence in ITK
   • Leaders: ?

BW NAC To-Dos

1. Registration Optimization
   • Caching
   • Freeze control points if no samples map to their area of influence
     • e.g., if they are in the background
   • Optimize Viola-Wells implementation
     • Take advantage of BSpline transform test in base itkImageToImageMetric class
   • Conduct performance characterization
     • Different size problems
     • Memory and speed
   • Leaders: Luis, Matt, Stephen
2. Registration Helper Classes
   • De-bug
   • Include timing information
   • Develop a simple demonstration of their use
   • Create wiki-based documentation
   • Load/save parameter files
   • Leaders: Stephen
3. Batchmake and Slicer
   • Add support for different experimental designs
     • New looping constructs etc. in BatchMake
     • Custom BatchBoards
     • Leaders: Matthieu Philippe, Julien Jomier

Integration Ideas

How to combine multiple to-dos in a single project/task

1. Batchmake for registration performance characterization
   • Incorporates
     • BWNAC - BatchMake
     • ITK - RIRE
     • BWNAC - Registration Optimization
   • Concept
     • Authors download a batchmake script and code, compile, run and the results are sent to a central batchmake system
     • Illustration: http://insight-journal.org/batchmake/
   • To-dos
     • Simple example batchmake script
     • ID data (Liz's MR data)
     • Display in IJ: How replicate a batchmake batchboard into another page?
     • Setup reporting structure (what measured and reported in each test?)