An Efficient Scheme for Brain Tumor Detection of MRI Brain Images Using Euclidean Distance

Abstract

Brain tumor is one the most common or major reason for death among individuals. It is apparent that survival of chances of patient could be expanded if tumor is identified at its initial stage. In this research work, we have developed a simple approach for detection of brain tumor which is based on the method using Euclidean distance classifier and making use of feature vector table and which over comes the limitations of conventional in which combination of supervised and unsupervised learning have been implemented to build cancer detection system, as there is huge overhead in this approach and there is a need to maintain large size training datasets. The new proposed method first convert the image into indexed image, than after de noising it with 3*3 mean filter, it conducts the block wise scanning to get feature set of statistical features in both frequency and time domain and finally based on Euclidean distance measures an optimized tumor part is segmented which is ROI (region of interest) then this segmented part is validated and test to arrive at exact brain tumor part required. The result show high reduction of time, increases specificity with better accuracy in terms of true positive rates.