Examination of Glioblastoma Images by Thresholding Using Heuristic Approach

Abstract

Brain tumor is a deadliest sickness in human community. It affects most of the humans despite of their age, gender and race. Medical imaging procedure is widely adopted to detect and evaluate the brain tumor. In this paper, a semiautomated approach is proposed to examine the high grade brain tumor called the Glioblastoma. During this study, the RGB slices of the brain views, like axial, coronal and sagittal are considered. The integration of the thresholding based on the Otsu and segmentation with the active contour is implemented to extract the tumor section. Initially the thresholding procedure monitored by the Social Group Optimization (SGO) acts as the pre- processing approach to enhance the tumor section and the segmentation procedure act as the post-processing section to extract the tumor. Finally, Haralick texture features are considered to compute the tumor characteristic. The experimental result confirms that, proposed approach helps to achieve better segmentation result on the RGB brain MRI pictures.