Proposed segmentation algorithm for mri brain tumor images

Abstract

A great work in the field of medical imaging research was focused on brain tumors segmentation in the recent years [1]. In many Hospitals, brain tumor measurements are obtained by magnetic resonance images (MRI) radiologists, who measure the mass of the tumor manually by using widely available imaging software. Such approach is a time consuming task and leads to large variations in operator performance. Several methods were proposed and developed for accurate segmentation [2-4] such as thresholding, region based, edge based, and clustering methods etc. In this work, thresholding, region growing and proposed method (region growing + active contour) were experimented on two datasets (the first one includes 18 images from a specialized clinic in Vladimir and the second data set of 52 images from a specialized clinic in Riyadh).