Brain image clustering by wavelet energy and CBSSO optimization algorithm

Abstract

Previously, the diagnosis of brain abnormality was significantly important in the saving of social and hospital resources. Wavelet energy is known as an effective feature detection which has great efficiency in different utilities. This paper suggests a new method based on wavelet energy to automatically classify magnetic resonance imaging (MRI) brain images into two groups (normal and abnormal), utilizing support vector machine (SVM) classification based on chaotic binary shark smell optimization (CBSSO) to optimize the SVM weights. The results of the suggested CBSSO-based KSVM are compared favorably to several other methods in terms of better sensitivity and authenticity. The proposed CAD system can additionally be utilized to categorize the images with various pathological conditions, types, and illness modes.