Computing optical flow in cardiac images for 3D motion analysis

Optical flow measurement is a well established image processing technique for the analysis of motion. Optical flow is computed as the solution to a linear algebraic equation system whose coefficient matrix is formed by a combination of temporal and spatial image derivatives The coefficient matrix is sparse and usually very large, as its dimension is twice that of the original 2D images. Here the authors describe the steps required to build an optical flow linear system for 3D space. They assess the use of the arithmetic reconstruction technique (ART) to solve the huge linear system involved in estimating 3D optical flow. ART does not require any special matrix structure apart from a high degree of sparseness. It has been extensively used in tomographic image reconstruction and has proven to be effective when solving large systems