2008 IGT Project Week Nonrigid MR-ULS Registration Algorithms for Neurosurgery, Prostate MRgFUS and Brachytherapy

Back to 2008_IGT_Project_Week#Projects

Team: Sandy Wells, Tina Kapur, Matt Toews, Nathan McDannold, Ben Schwartz, Steve Haker, Michel Audette, Nathan McDannold

Goals: A round table discussion of issues relating to multimodal (MR-US) registration for image-guided therapy.

Agenda

1. Self-introductions
2. Definition of relevant research issues
3. Visual Inspection of data sets brought by participants
4. Conclusions Wrap-up

Non-exhaustive List of Issues for Discussion

1. Goals of Non-Rigid MR-US Registration?
   • Use coarse, cost-effective inter-operative US imagery to update detailed preoperative MR imager, to enhance visualization during therapy.
2. Requirements
   • Clinical requirements: registration precision?
   • Is MR-US registration necessary?
3. State-of-the-art
   • Existing systems?
   • Real-time?
   • Accuracy?
4. Image Registration
   • MR-US similarity measurement: mutual information, correlation ratio?
   • Transforms: rigid, deformable? Outlier detection?
   • 3DMR-2DUS slices? 3DMR-3DUS? 3DMR-3DUS reconstuction?
   • Initialialization: 3D tracking of US probe?
   • Freehand US?
   • Robust?
5. Updating Pre-operative Imagery
   • What information is required? Rigid/non-rigid deformation? Resolution?
   • Interpolation?
6. Challenges
   • Poor US image quality, occlusion, resection, poor visibility of anatomy
   • Computational complexity
7. Data
   • Phantom data?
   • Data sharing for benchmarking?
8. Approach-based path planning
   • Neurosurgical guidance can/should integrate knowledge of surgical approach: frontal, occipital, pterional, transnasal...
   • Use knowledge of _specific_ critical tissues & likely eloquent areas in path planning
   • Curvilinear neurosurgical path planning.