Efficient Initialization for Constrained Active Surfaces, Applications in 3D Medical Images

A novel method allowing simplified and efficient active surface initialization for 3D images segmentation is presented. Our method allows to initialize an active surface through simple objects like points and curves and ensures that the further evolution of the active object will not be trapped by unwanted local minima. Our approach is based on minimal paths that integrate the information coming from the user given curves and from the image volume. The minimal paths build a network representing a first approximation of the initialization surface. An interpolation method is then used to build a mesh or an implicit representation based on the information retrieved from the network of paths. From this initialization, an active surface converges quickly to the expected solution. Our paper describes a fast construction obtained by exploiting the Fast Marching algorithm. The algorithm has been successfully applied to synthetic images and 3D medical images.

• Type: Conference
• Year: 2004
• Where: ECCV
• Authors: Roberto Ardon, Laurent D. Cohen
• Keywords: Active Surface, Active Surface Initialization, Computer Vision, ECCV 2004, Minimal Paths