System and method for using learned discriminative models to segment three dimensional colon image data

A system for segmenting a three dimensional (3D) image of an anatomical structure comprises:

• a processor that receives the 3D image of the anatomical structure, the processor performing the following steps:
• computing a discriminative probability model for each voxel in the 3D image;
• performing thresholding on each discriminative probability model;
• obtaining one or more two dimensional (2D) slices of the thresholded 3D image along X-Y planes;
• selecting seed regions in the 2D slices;
• performing morphological region growing on the selected seed regions;
• obtaining an initial 3D segmentation;
• performing boundary evolution on the initial 3D segmentation;
• and
• removing the segmented anatomical structure in the original 3D image;
• and
• a display for displaying the segmented anatomical structure.