Recalling‐enhanced recurrent neural network optimized with wood pecker mating algorithm for brain tumor classification

Abstract

SummaryBrain tumors are caused by the uncontrollable division and proliferation of abnormal cell groupings inside or around the brain. This cell grouping affects the function of brain activities and destroys healthy cells. Several methods have been used to detect the brain tumor, but none of the methods present adequate accuracy and increasing computational time. To overcome these issues, this article proposes recalling‐enhanced recurrent neural network (RERNN) optimized with woodpecker mating algorithm for brain tumor classification (BTC) to accurately classify the four types of brain tumors, namely, glioma, meningioma, pituitary gland, and normal. The brain MRI images are collected from Brats MRI image data set. The simulation is activated in MATLAB. From the simulation, the proposed BTC‐RE‐RNN–WMA achieves better accuracy 29.98%, 26.74%, 33.27%, higher precision 19.24%, 34.82%, 26.92%, when comparing to the existing models, such as efficient identification with categorization of brain tumor utilizing kernel based SVM for MRI (BTC‐KSVM‐HHO), combined training of two‐channel deep neural network for brain tumor categorization (BTC‐JT‐TCDNN), improved structure for brain tumor analysis utilizing MRI depending on YOLOv2 with convolutional neural network (BTC‐YOLOv2‐CNN) methods.