Alain Pitiot, Arthur W. Toga, and Paul Thompson
Laboratory of Neuro Imaging, Dept. Neurology, Division of Brain Mapping,
UCLA School of Medicine, Los Angeles CA 90095, USA
Summary. We propose a general elastic matching method to evolve deformable templates over 2D images, for automated structure identification. A prototype template, in the form of a B-spline, is warped in a potential field which guides the deformable contour. In contrast to prior approaches, the potential field is adapted both spatially and temporally. This improves the speed and accuracy of the object extraction process. The potential field is initially designed based on spatial characteristics of edge features in the input image. This guarantees efficient dynamics as early as the first iterations. The potential function, which guides the deformable contour, is also logarithmically re-mapped to improve the global consistency and convergence speed of the evolution process (see below).
Automated Extraction of the Corpus Callosum Boundary in Brain MRI.
|RESUME| E-MAIL ME| PERSONAL HOMEPAGE| PROJECTS|