In this paper, an efficient method for Magnetic Resonance Imaging (MRI) brain image classification is presented using Stockwell (S)-Transform, Sammon Mapping (SM) and Naïve Bayes (NB) classifier. Initially, the MRI brain images are represented in frequency domain by S-Transform. As the representation in frequency domain provides more detailed information than spatial domain, S-Transform is used for feature extraction. The high dimensional S-Transform feature space increases the complexity. Hence, SM technique is used to reduce it and then classification is made by NB classifier. The performance measures such as sensitivity, accuracy and specificity are computed to evaluate the system. Result shows the better classification accuracy of 94% is obtained by S-Transform based SM technique with NB classifier with 94% of sensitivity and specificity.
Modified Visual Geometric Group Architecture for MRI Brain Image Classification
