The aim of this research is to develop vision algorithms for analysis of non-rigid motion from medical imagery. We will continue our algorithm development in snake tracking and image velocity field computation for assessment of tissue deformations and analysis of fluid flow from MRI and X-ray imagery, and extend the techniques to 3D. MRI is unique in its ability to non-invasively and selectively alter tissue magnetization, and create tagged patterns within deforming tissue such as the heart muscle. The resulting pattern defines a time-varying curvilinear coordinate system on the underlying tissue, allowing for precise and quantitative measurement of tissue motion and deformation from a sequence of images. The motion of these visible tag lines is tracked with Dynamic Programming (DP) B-spline snakes, and interacting B-spline deformable grids. Warp algorithms have been developed which continuously map two tissue regions in the plane such that two embedded grids of curves are brought into registration. For fluid flow field analysis, a scalar function formulation for optical flow has been developed. The novel aspect of the new optical flow framework is that it automatically enforces incompressibility on the resulting velocity field. Applications of the techniques to various experimental conditions are planned.
