A hybrid framework for brain tumor classification using grey wolf optimization and multi-class support vector machine

Abstract

Medical image processing has a vital role in the detection of diseases in human beings. The accuracy for disease detection using any medical image is highly dependent on the image processing methods. Features extraction and reduction are the two key steps during the medical image processing for disease classification. To develop an effective and efficient mechanism with high accuracy for classification of malignant brain tumor from Magnetic Resonance Imaging (MRI) is the objective of the present research. To achieve this, a nature inspired algorithm; namely, Grey Wolf Optimization (GWO) along with a classification method, multiclass Support Vector Machine (MSVM) is used. Further, Results for the classification accuracy obtained from GWO are compared with other two well-known optimization algorithms such as Particle Swarm Optimization (PSO) and Firefly Algorithm (FA).