Pre-Segmentation for the Computer Aided Diagnosis System

This paper aims to introduce the method for detection and classification of an abnormality of the brain region into infarct, haemorrhage. Initially, the image is pre-processed to remove film artifact, noise using morphological operations. And the brain image is cropped to eliminate the extra parts. Before the abnormality region (infarct, haemorrhage) is segmented, the image is pre-segmented. In this stage, the image is subdivided into four quadrants using split and merge technique to find the region that has the possibility for inclusion of abnormal areas. Thus, the unnecessary regions are no need to search and segment. This paper proposes pre-processing and pre-segmentation steps for the detection of abnormal region in the brain image. Then, abnormalities can be segmented by grouping homogeneous regions according to the predefined criteria. The accuracy of the segmentation results can be assessed by the specialist.