National Alliance for Medical Image Computing NA-MIC Work of Tannenbaum Group Computer Science and Mathematics Stony Brook University
National Alliance for Medical Image Computing In collaboration with (no particular order): Steven Haker Tauseef ur-Rehman Ayelet Dominitz Eric Pichon Delphine Nain Yi Gao Ivan Kolesov LiangJia Zhu Samuel Dambreville James Malcolm Ganesh Sundaramoorthi Behnood Gholami Marc Niethammer Oleg Michaelovich Namrata Vaswami Peter Karasev Arie Nakhmani Yogesh Rathi Patricio Vela Vandana Mohan Shawn Lankton Gozde Unal Students and Postdocs
National Alliance for Medical Image Computing Assorted Projects Segmentation: Local/Global, Sobolev, Finsler, Steerable, Optimal Control Shape Theory: Spherical Wavelets, OMT Registration: OMT, Particle Filtering, Optimal Control Meshing (hexahedral) Conformal maps (brain warping, colon fly- throughs)
National Alliance for Medical Image Computing KSlice Interactive Segmentation Added Features: ● Editor module ● Inter-slice interpolation ● Control of user input function ● Choice for image cost functional ● Selection of tools for input
National Alliance for Medical Image Computing 3D Interactive Segmentation GrowCut method Easy for user interaction Slow for 3D images Level sets method Flexible to segment complex structures Rely on good initialization 3D interactive segmentation Fast GrowCut for initialization Level sets refinement, Slicer modules e.g. KSlice
National Alliance for Medical Image Computing Comparison MethodSegmentation time (seconds)Memory (MB) Quantitative 1st edit2nd edit3rd editDiceVol. Overlap GrowCut % Proposed GrowCut: Proposed: Lung segmentation: image ROI [ ] 3 rounds of interaction/editing
National Alliance for Medical Image Computing December 15 7 Particle Filtering
National Alliance for Medical Image Computing 8 Particle Filtering
National Alliance for Medical Image Computing Particle Filtering Registration
National Alliance for Medical Image Computing
Longitudinal shape analysis
National Alliance for Medical Image Computing Traumatic Brain Injury
National Alliance for Medical Image Computing Fibrosis distribution analysis AFib recurrence after RF ablation Group 1, cured Group 2, recurrence Hypothesis: Group-wise difference between 1 and 2 Shape and fibrosis (intensity) distribution
National Alliance for Medical Image Computing Results Gray: no-statistical difference. Color region: statistically different regions.
National Alliance for Medical Image Computing Hexahedral Meshes
National Alliance for Medical Image Computing Future Work Compressive Sensing/Mass Spec/Raman Spectroscopy for better tumor margin delineation (Nathalie Agar, Alex Golby, Yi Gao) DECS for neurosurgery/validation (Sonia Pujols, Yi Gao) Microanatomical imaging (Joel Saltz) Radiation oncology (Harini V., Joe Deasy, Greg Sharp, Ivan Kolesov, Yi Gao) Fibrosis analysis (Rob MacLeod, Josh Cates, Yi Gao, LiangJia Zhu)
National Alliance for Medical Image Computing Conclusions Thank you to all the collaborators and especially to Ron Kikinis for giving us this great opportunity! May the Force be with you and Slicer.