Computer Aided Diagnosis of Brain Tumour Using Walrus Optimization Algorithm

Abstract

Brain tumour segmentation is one of the most challenging problems in medical image analysis. To detect brain cancers, radiologists must use a computer-based tumour classification model. In medical imaging research, several computer-aided diagnostic (CAD) models are available to help radiologists with their patients. The reason for brain growth division is to deliver exact outline of brain tumour regions. This study suggests a Walrus Optimisation Algorithm approach for detecting brain tumours using MRI (Magnetic Resonance Imaging). Separating the characteristics into four categories has been utilized: no growth, gliomas, meningiomas, and pituitary cancers. The deep learning based model inception AlexNet and ResNet-18 is trained on an augmented training dataset. The CNN classifier is used for characteristics map improvement, while the LSTM (Long momentary memory) classifier is utilized for order. Besides, the boundary remembered for the classifiers is chosen aimlessly utilizing the Walrus Improvement Calculation to build the presentation of the CNN-LSTM classifier. The performance of the tumour diagnosis is assessed using the metrics: overall classification accuracy of 98.8%, precision of 96.23%, recall of 97.01%, specificity of 98%.