Automatic detection of abnormal regions using guassian mixture model

Automatic detection of abnormal region, such as positions of lesions,disease locus, is an essential processing component for computer-aided medicalimage diagnosis. In this paper, we use gaussian mixture model (GMM,) to simulatethe process of doctors' checking medical images with their accumulated visualexperiences, which is based on the assumption that the normal region is thebackground and abnormal region is the foreground. We employ the brain CTsequences and index the similar slices in the anatomy axial direction. Thismethod was tested on brain CT image with different lesions and receivedfavorable results. ICIC International