3D Medical Image Segmentation Using Level Set Models and Anisotropic Diffusion

This paper is concerned with the segmentation of structures from 3D medical datasets where an important challenge is the need to overcome problems associated with intensity in homogeneities, noise and proximity to other tissue of similar intensity levels resulting in weak delineating boundaries. Level sets are part of an important class of methods that utilize partial differential equations (PDEs) and have been extensively applied in image segmentation. During the evolution of the curve, an isotropic diffusion is adaptively applied to the image to remove noise while preserving boundary information. The speed of curve evolution and the segmentation result has been significantly improved compared with traditional level set methods and region-based implementation of level sets. The method is tested on 3D CTA datasets containing abdominal and thoracic aneurysm (AAA and TAA). The results demonstrate the capability of the methodology to segment an AAA and TAA from these datasets.