Hybrid classifier: Brain Tumor Classification and Segmentation using Genetic-based Grey Wolf optimization

Abstract

This work uses a novel brain tumor classification technique which comprises 5 steps like “(i) denoising, (ii) skull stripping, (iii) segmentation, (iv) feature extraction and (v) classification”. At first, the image is given in the denoising procedure, whereas the amputation of the noise process is performed by using an entropy-oriented trilateral filter. Subsequently, noise removed image is used to skull stripping procedure through morphology segmentation and Otsu thresholding. Then, the segmentation takes place using the adaptive CLFAHE method. GLCM features are extracted after finishing segmentation. Here, hybrid classification represents the hybridization of 2 classifiers such as FNN and “Bayesian regularization classifier”. The very important involvement lies in the best selecting of hidden neurons in FNN. In this paper, a novel genetic algorithm based GWO (GA-GWO) method is proposed that hybrids the conception. At last, the proposed method performance is evaluated with conventional techniques to show the supremacy of the proposed method.