COMPUTER AIDED DETECTION OF BRAIN TUMOR USING MAGNETIC RESONANCE IMAGING-MRI

Abstract

Medical image widely use Active contour models (ACMs) or snake for segmentation purposes, specifically to extract tumor boundaries in brain tumor MRI images. ACMs practical applications are limited due to issues related with initialization and weak convergence to boundary concavities. Here comparative study is presented to analyze three different Active contour models (ACMs) of snakes implemented on Brain tumors MR imagery. Where the three ACMs are named as: Gradient Vector Flow (GVF), Boundary vector field (BVF), Generalized Boundary Vector Field (GBVF). The assessment of results with tumor MR imagery demonstrates that GBVF method is more perfect and robust for brain tumor segmentation. Selected imagery for testing provides difficult and typical problems found in different cross-sectional views of brain tumor MRI segmentation. The comparative results emphasize both the potential and weaknesses of these models.