Enhancing Brain MRI Tumor Detection: Exploring Vision Transformers and Fine-Tuned Convolutional Neural Network Architecture for Improved Performance

The timely identification of brain tumors using medical imaging plays a pivotal role in enhancing patient prognoses and shaping effective treatment approaches. This research investigates the efficacy of Convolutional Neural Networks (CNNs) and our proposed model based on Vision Transformers in classifying brain tumors from MRI scans. The study leverages a dataset of MRI images, preprocesses them to enhance their suitability for deep learning, and employs both CNNs and Vision Transformers for classification tasks. To assess performance, a confusion matrix was plotted, accuracy percentage was calculated, and accompanying graphs were generated. Furthermore, the models were rigorously tested on dual Kaggle GPU T4 to evaluate their real-world computational efficiency. Comparative analysis reveals insights into the strengths and limitations of both architectures in the context of medical image analysis. The findings contribute to the ongoing discourse on the application of deep learning techniques for medical diagnosis and highlight the potential of Vision Transformers in capturing global features from medical images. The outcomes of this research not only inform the choice of architecture for brain tumor detection but also open avenues for future investigations aimed at hybrid models and enhanced analysis of medical image segmentation methodologies.