BT Detection Using Improved Whale Optimization and Convolutional Neural Networks

Medical image processing was indispensable to a growing need for quick, effective, and systematic Brain Tumour (BT).Pixels are grouped into larger regions via a process called "region growing" that begins at the seed locations.In noisy images where edges are hard to identify, region growth-related methods perform better than edge-related methods.Dataset taken from Kaggle and URI repository as brain MRI images.The input image undergoes morphological edge detection and the image is then enhanced by reconstructing it through erosion and dilation.In this study, we employ an approach that involves median filtering of the image, Otsu automated segmentation, morphological filtration and dilation, and Improved Whale Optimization -Region based Convolutional Neural Network (IWO-RCNN) classification.It used the Weka 3.9 tool to perform the classification after preparing the brain Magnetic Resonance Imaging (MRI) database and carrying out the approach in MATLAB R2015a.We compared our approach to the Brain-Surface Extractor (BSE) and a layer-set technique proposed for the mouse brain they analyzed its performance under increasing Signal to Noise Ratio (SNR) and resolution.According to the data, this approach works better than the competition and is reliable at lower resolutions of partial volume impacts and low SNR.The system has a greater accuracy of 98.7%, precision 96.5% and recall 95.6%.