Deep Learning-based Neural Network for Automated Brain Tumor Diagnosis from MRI Images

The identification and categorization of brain tumors are crucial issues in medical imaging as they affect treatment choices and diagnosis. While several approaches for brain tumor detection and classification have been developed, many are restricted in their capacity to offer accurate and efficient findings, particularly when it comes to tumor localization and classification within medical imaging. Existing models frequently use distinct pipelines for tumor location and classification, resulting in poor performance and inefficiency. Furthermore, these models frequently struggle with imbalanced datasets and outcomes that are challenging to understand. In this research, we address a significant research gap in existing models by presenting a unique method for classifying brain tumors using RetinaNet, an object detection model. We tested our method using a publicly available dataset of brain MRI scans (Figshare), proving its accuracy in detecting and classifying tumors. Our approach contributes to better clinical decision support by not only achieving high classification accuracy but also providing the precise and comprehensible location of brain tumors.