Comparison of ASM and AAM-based segmentation of prostate image in the CT scans for radiotherapy planning

摘要

In this paper we present a novel method of medical CT data segmentation using explicit atlas-type knowledge and Active Appearance Model (AAM) as a principal segmentation tool. New approach of automatic landmarks creation for AAM is proposed. Obtained results of automatic segmentation of prostate image in the CT scans are compared with true outlines drawn by medical doctors during the radiotherapy planning and with the Active Shape Model (ASM)-based segmentation tested by us before. Using ASM turned out to be better than AAM.