Abstract
In the present days, for the human body anatomical study and for the treatment planning medical science very much depend on the medical imaging technology and medical images. Specifically for the human brain, MRI widely prefers and using for the imaging. But by nature medical images are complex and noisy. This leads to the necessity of processes that reduces difficulties in analysis and improves quality of output. This paper discuss about an improved segmentation algorithm for different defects detection in brain MRI (Tumour) and have compared the performance of this method with conventional method.