KFCM Algorithm for Effective Brain Stroke Detection through SVM Classifier

Abstract

Ischemic stroke is stated as a loss of neurological brain function due to the sudden loss of blood circulation in the particular area of the brain. Analysis of ischemic strokes is further complicated by the fact that damage often crosses into multiple regions of the brain. MRI image analysis is done by the Neurologist to detect the lesion tissue in the brain image. The technique of manual labeling of ischemic stroke lesion turns out to be time-intensive making an automated method desirable. In the existing work, Segmentation of the Ischemic Stroke image was done by Otsu technique and integrated with SVM classifier. From the results it was inferred that the Accuracy of the technique is 88%, Specificity is of about 66%, and Sensitivity value is 94%.In order to obtain better accuracy in segmentation and for precise detection of the stroke, Kernelized fuzzy C-means clustering with adaptive threshold algorithm has been implemented. The algorithm identifies the distance and intensity of the lesion tissue. The accuracy and segmentation results of the Classifier is measured in the testing and training phase by comparing the similarity and diversity of sample sets by considering different sequences which are analyzed using MATLAB version 7.4.