TubeTK/Events/2010.09.14

Meeting topics

• More detailed project brainstorming / thoughts / questions:
  • Apply slipping-organ motion field regularization to vessel-based registration?
    • Simultaneous registration of multiple vessel trees (e.g. from different organs) with slipping registration?
    • Or combination of vessel-based registration with intensity-based registration?
    • How well do non-rigid registrations based on vessels apply to organ surfaces?
  • Likelihood of vessels sliding along each other?
    • Unlikely?
  • Discontinuities in 4D image registration, ex. fast changes that look discontinuous due to sampling rate?
    • Smoothing in time?
    • Discontinuous because one organ moves much faster than its neighbor? (ex. heart, lung, etc)
  • Non-rigid registration despite occlusions from surgical tools
    • Unobtrusive registration between intraoperative and preoperative images, ex. for registration updates during surgery
    • Ex intraoperative fluoro (both tools and vessels are dark?) (US: tools bright, vessels black)
    • Perhaps rigid registration can overcome tool occlusion, but what to do in non-rigid registration?
      • But how often is non-rigid registration used intraoperatively nowadays? Thinking ahead...
    • Alternative: track tools (tracking system / image-based) and then mask them out so that they are not considered during registration.
      • What would be the advantage to incorporating it into the registration?
    • Or, how well does vessel centerline extraction cope with tools in the field of view?
      • Probably would do ok... tested? (could test artificially)
  • Characterizing sliding motion, for additional constraints / anatomically-based strategies
    • • already extensively studied for respiration

Status

Danielle

• Future:
  • valgrind
  • application for Andinet's diffusion properties
  • slipping organ - draft abstract
• Status:
  • Going thru Marc's notes
• built VMTK / Slicer /w VMTK modules
• read Schmidt-Richberg et al., Slipping objects in image registration: Improved motion field estimation with direction-dependent regularization, MICCAI 2009
• looked over Marc's notes
• brainstorming

Romain

• Preliminary results from tumor microenvironment segmentation
• Slicer built with VMTK modules
• Learning execution model of Slicer
• Reviewing papers from Dr. Bullitt on tortuosity

Romain

Read Noise Reduction in Computed Tomography Scans Using 3-D Anisotropic Hybrid Diffusion With Continuous Switch.

Running the itk anisotropic filters tests