Segmentation of Brain Tumor Tissue Using Marker Controlled Watershed Transform Method

Abstract

Medical imaging makes use of the technology to noninvasively reveal the internal structure of the human body. By using medical imaging modalities patient’s life can be improved through a precise and rapid treatment without any side effects. The main purpose of this paper is to develop an automatic method that can accurately classify a tumor from abnormal tissues. The images used for tumor segmentation have been obtained from MRI modality. In this research we have developed a novel image segmentation technique based on catchments basins and ridge lines to accurately segment the brain image. This technique is based on immersions techniques for extracting objects from the image based on internal markers and external markers. Direct application of watershed transform leads to over- segmentation due to the presence of noise and other irregularities inherent in digital images. So to avoid this, we have carried out some preprocessing to remove noise and attenuate the curvilinear structures present in the MRI images during acquisition stage. After preprocessing step we calculated the morphological gradient of the input images. Then both internal and external markers of the original images were calculated and finally the watershed transform applied to complete the segmentation process. We have tested our algorithms on images obtained from Brain Atlas data base and found that the results closely match that of the radiologist.