A biologically-inspired hybrid deep learning approach for brain tumor classification from magnetic resonance imaging using improved gabor wavelet transform and Elmann-BiLSTM network

Brain tumor represents the unnatural growth of cells in the brain and is identified to be one of the deadliest cancers around the globe. The survival rate of this disease varies with the stage at which the cancer is identified. Therefore, it is important to identify accurately the tumor region in the brain and also the tumor type as early as possible to improve the survival rate by appropriate treatment plans. One of the most significant ways to analyze the tumor is through the examination of magnetic resonance imaging (MRI) images of the patients. Since the amount of data being generated is huge, manual techniques are found to be inappropriate with several misclassifications. To reduce misclassifications and to cope with the large amount of data, a deep learning-based classification model is formulated in the proposed work that functions based on five major modules. Initially, the images are skull-stripped and filtered using the guided bilateral filter (GBF). Then, the tumor regions are segmented using thresholding scheme and then the major texture and edge features are collected using the improved Gabor wavelet transform (IGWT). The optimal features are selected using the black widow adaptive red deer optimization (BWARD) algorithm and then the features are fed to the hybrid Elman bidirectional long short term memory (EBiLSTM) network model for classification. The proposed model is simulated in Matlab platform using the brain tumor MRI dataset and the results proved that the proposed model is effective in classifications with an accuracy of 98.4%.