Lesion Detection Based BT Type Classification Model Using SVT-KLD-FCM and VCR-50

Due to delayed diagnosis and treatment, Brain tumors have a very low survival rate and are easily detected on MRIs. With the advancement of science and technology, ML models attempted to identify and classify BT diseases. However, the models have constraints on lesion identification and accurate region prediction. Hence, to overcome these constraints and other problems in the way to predict BR disease, a novel deep learning model has been proposed. The input MRI dataset has been taken for this paper, and pre-processed for enhanced accuracy at the segmentation stage. Pre-processing includes contrast enhancement, edge enhancement using ACMPO, and Skull removal. Pre-processed results are segmented using SVT-KLD-FCM. Segmentation output was two clusters with and without lesions. Without lesion images were declared as normal, and with lesions are given for feature extraction and then to classifier VCR-50. Classification results in three types of classes namely glioma, meningioma, and pituitary. The proposed model was implemented and achieved accuracy of about 98.77% better results than previous models.