An efficient deep learning approach to detect brain tumor using MRI images

Abstract

The formation of altered cells in the human brain constitutes a brain tumor. There are numerous varieties of brain tumors in existence today. According to academics and medical professionals, some brain tumors are curable, while others are deadly. In most cases, brain cancer is identified at a late stage, making recovery difficult. This raises the rate of mortality. If this could be identified in its earliest stages, many lives could be saved. Brain cancers are currently identified by automated processes that use AI algorithms and brain imaging data. In this article, we use Magnetic Resonance Imaging (MRI) data and the fusion of learning models to suggest an effective strategy for detecting brain tumors. The suggested system consists of multiple processes, including preprocessing and classification of brain MRI images, performance analysis and optimization of various deep neural networks, and efficient methodologies. The proposed study allows for a more precise classification of brain cancers. We start by collecting the dataset and classifying it with the VGG16, VGG19, ResNet50, ResNet101, and InceptionV3 architectures. We achieved an accuracy rate of 96.72% for VGG16, 96.17% for ResNet50, and 95.55% for InceptionV3 as a result of our analysis. Using the top three classifiers, we created an ensemble model called EBTDM (Ensembled Brain Tumor Detection Model) and achieved an overall accuracy rate of 98.60%.