Segmentation of Prostate MRI Volumes Using 3D Shape Model Constrained Tissue Classification
Williamson D, Graham J, Allen P, Hutchinson C
University of York
Benign Prostatic Hyperplasia is a non-cancerous enlargement of the prostate that can cause constriction of the urethra. Evaluation of potential drug treatments designed to shrink the prostate requires a means of quantifying their effect. In an MR image the prostate has two distinct zones and the relative volumes of these has been shown to be sensitive to the progress of BPH and the degree of urinary flow obstruction. Here we present a method of semi-automatic prostate segmentation in MR images using grey level voxel classification, coupled with a 3D point distribution model to provide a smooth, anatomically credible spatial constraint.