Abstract—A brain tumor is formed by an excessive rise of abnormal cells in brain tissue. Early identification of brain tumors is essential to ensure patient safety. Magnetic Resonance Imaging (MRI) scan is used to diagnose brain tumors. Unfortunately, because of the varying form of tumors and their location in the brain, physicians are unable to provide good tumor segmentation in MRI images. Accurate brain tumor segmentation is required to identify the tumor and offer the appropriate therapy to an individual. In this research, a novel hybrid deep learning technique termed Convolutional Neural Network and ResNeXt (CNN-ResNeXt) is introduced to segregate and classify the tumor automatically. Firstly, the MRI image is collected from the standard datasets known as BRATS 2015, BRATS 2017 and BRATS 2019. Then the collected data is smoothed and enhanced by batch normalization technique and the features are extracted from the smoothened image based on tumor shape position, shape and surface features using AlexNet model. Next, using an Adaptive Whale Optimization (AWO) approach, the optimal features are selected for effective segmentation. Consequently, the image segmentation process is done using CNN–ResNeXt depending on the selected features. Finally, the segmented image is used for the classification which is also done by employing CNN-ResNeXt. Whereas compared to the other existing models, the proposed CNN–ResNeXt model achieved a greater accuracy of 98% for the tumor core class. This demonstrates that the proposed methodology segregates and classifies brain tumors effectively.
 
Keywords—Adaptive Whale Optimization (AWO), AlexNet, brain tumor, Convolutional Neural Network (CNN), Magnetic Resonance Imaging (MRI), and ResNeXt


Copyright © 2023 by the authors. This is an open access article distributed under the Creative Commons Attribution License (CC BY-NC-ND 4.0), which permits use, distribution and reproduction in any medium, provided that the article is properly cited, the use is non-commercial and no modifications or adaptations are made.