Engineering:TCON 09 08 2005

From NAMIC Wiki
Jump to: navigation, search
Home < Engineering:TCON 09 08 2005

Expected Core 3 Attendees: Andy Saykin(Dartmouth), Martina Panzenboeck (UCI), Jim Fallon (UCI), Jessica Turner (UCI)


  • Introduction (Ron)
  • Engineering Progress (30 min) (Note: details for this are available in a wiki page here and a consolidated powerpoint is available here)
    • (15 min) Application Projects from Programming Week and more ...(Steve Pieper)
    • (15 min) Algorithms and Infrastructure Projects from Programming Week (Bill Lorensen)
  • (15 min) Core 3 response to projects content and future direction
  • (10 min) Core 3 suggestions for items for discussion in Boston on September 26th


Background of programmer's week by Bill Lorensen and Ron Kikinis

  • Core 1 and Core 2 meet and split into small teams
  • Setup Sandbox for shared efforts
  • Tina Kapur organized reports
  • Ron was pleased with interactions between Cores
  • Move some emphasis from Core 1 to Core 3 by the Engineering team
    • All-hands meeting marks 1/2 time for Core 3, time to start writing papers

Overview of Applications project by Steve Pieper

  • Structural, Functional and Diffussion Imaging were focus
  • Main focus has been diffusion
  • Quantative Fiber Tracking Project
    • Define quantaties, i.e. fa and ensure consistent implementation
    • Unify into ITK, first instance is clustering using Normalized Cuts
    • Basic algorithm
      • Generate large numbers of tracks
      • Have the algorithm identify a specified numbers of similar tracks via Normalized Cuts
  • DLPFC Semi-automatic Segmentor
    • Baysen classifier, and rules to segment the dorsal-lateral prefrontal cortex
    • Slicer workflow has been defined
    • Rules have been identified for other structures, and new structures should be added soon
  • EPI to structural image registration
    • Use non-rigid registration ITK tools to address this problem
    • Used notion of image orientations to address different acquisition orientations
    • Resulted in better ITK support for orientation, useful in clinical settings
    • Rigid and Affine registration done, exposed in Slicer
    • Soon Demon's and B-Spline based registrations
    • Need to compare to existing tools: SPM and FSL
  • Some of this work has been published in the Insight Journal

Overview of Alogrithms Projects by Bill Lorensen

  • Spherical harmonics added to ITK
  • Database and mechanisms to gather statistics on populations from many different sources
    • Much of the infrastructure exists inside Dart
  • Bayesian Classifier / edge-preserving affine invariant noise remover algorithm
    • Transition from Matlab to ITK
  • 3d adaptive tetrahedral meshing
    • Generate computational stable solid meshes for deformable registration
    • Transition algorithms to ITK
  • 2d/3d landmark detection algorithm
    • Look for corners in the brain, may be useful for registration tasks
  • Flux diffusion algorithm
    • Port anisotropic diffusion algorithm for thin structures to ITK
  • Intensity correction algorithm
    • Port external code to ITK, promote to first class filter

Overview of Engineering Tasks by Bill Lorensen

  • ITKu
    • Expose ITK functions at the commandline
    • Suitable for batch tasks
  • Grid adaption for NAMIC algorithms
    • LONI pipeline and Condor job submission
  • NrrdIO Image format
    • Support for DTI-type structures in files
    • Handle measurement frames for tensors, variable length vectors
  • DTMRI Nightly Testing
    • Nightly regression tests for DTMRI Slicer module
  • LONI Pipeline for UNC shape analysis
    • Adapt UNC scripts to LONI Pipeline
  • Slicer 3.0 Architecture Meeting
    • Still in dream/planning stage
    • Need significant input from Core 3
    • May be completely different from existing Slicer implementation
    • What is useful from all available technologies? How do we take best-of-breed
    • Need to provide architectural support to preserve algorithm development investments
  • ITK Release (2.2)
    • Much has been added at the behest of NAMIC

Core 3 Feedback

  • Andy: development is really exciting from a Core 3 perspective
    • Tools need to migrate into Core 3 day-to-day use
  • Some must-haves are already recorded and will be presented in Boston on Sept 26th.
  • Bill Lorensen invited Core 3 to participate in future tcons