Automated Brain Tumour Detection and Classification Using Equilibrium Optimization Algorithm with Deep Learning Model

stract - Earlier diagnosis and classification of Brain Tumour (BT) with high prognosis accuracy is the most crucial step for detection and treatment to increase the survival rate of the patient. However, the manual inspection of brain MR images is timeconsuming for doctors and radiologists to localize and identify cancer and normal tissues and classify the tumours from clinical imaging datasets. A Computer-Aided Diagnosis (CADx) technique is paramount to address these problems. It must be implemented to facilitate radiologists or doctors and relieve the workload in medical imaging analysis. Lately, researcher
workers have introduced accurate and robust solutions to automate the detection and classification of BT. Classical Machine
Learning (ML) techniques were exploited for the analysis of BT. Deep Learning (DL) has combined feature extraction and
classification into a self-learning method on a large number of labelled datasets, which dramatically improves the performance. Therefore, the study presents an Automated BT Detection and Classification technique on MRIs using Equilibrium Optimization with Deep Learning (ABTDC-EODL). The goal of the ABTDC-EODL approach is to classify BT among adults as well as kids under the age of 10. Primarily, the ABTDC-EODL technique involves a Wiener Filtering (WF) technique to discard the noise that exists in it. To derive features, the ABTDC-EODL technique uses the ShuffleNet model, and the EO algorithm can choose its hyperparameters. Finally, the Stacked Autoencoder (SAE) model was utilized to identify the presence of BT. The ABTDCEODL model can be validated on a benchmark Br35H: Brain Tumour Detection 2020 dataset, which encompasses 1500 tumorous and 1500 non-tumorous images. The obtained values highlight the better results of the ABTDC-EODL algorithm over other existing techniques.