AN EFFICIENT BRAIN TUMOR DETECTION USING MR IMAGES WITH KERNEL BASED PCA AND SVM

The primary method for detecting brain tumors is through the use of MRI. In this paper, a new hybrid technique for classifying human brain images into benign and malignant tumors. The technique consists of several stages: categorization, feature extraction with feature reduction, and pre-processing and segmentation. The threshold function is used during the segmentation and pre-processing phases. In respect to MRI, the second stage is acquiring characteristics using DWT. In third step, KPCA is utilized to distil MRI data to their most crucial components. Finally, the classification stage employs a classifier called Kernel Support Vector Machine (KSVM) to identify the infected area in the brain tumor. Experimental results have shown that proposed technique achieves good accuracy of 97% in distinguishing normal and abnormal tissues in brain MR images. Overall, this proposed technique is effective and powerful in tumor detection compared to existing frameworks.