Brain MRI segmentation using a weighted based approach

Abstract

Image Segmentation can be considered as one of the most critical and important procedures that are implemented in the field of image processing and analysis. Its objective is to partition the image into semantically, meaningful distinct entities based on specific constraints. It is the most crucial and essential process for better characterization of the region of interest in any image, specifically in medical images. Accordingly, the segmentation procedure may affect all subsequent image analysis processes such as the features extraction, the objects description and representation, and the objects classification. With the increasing of medical imaging modalities and the number of examinations, the manual segmentation of medical images by the radiologists has become a tedious and time consuming process and not efficient. Therefore, it is essential to perform an automated segmentation that does not require much effort and involvement by the user, as the manual segmentation requires. In this process, the region of interest or the anatomical structure must to be extracted from the acquired image in order to perform certain measurements or monitor the effect of a particular treatment administered to the patient. The objective of this project is to develop a segmentation technique that will segment structures or lesions in brain MRI images. Consequently, a quantitative evaluation is performed to study the accuracy of the implemented techniques.