Comprehensive Evaluation of Vision Transformer in MRI-based Brain Tumor Detection

Brain tumor detection using magnetic resonance imaging (MRI) plays a pivotal role in early diagnosis and treatment planning. This study presents a novel approach for brain tumor detection using MRI images, preceded by prepro-cessing steps including contrast enhancement, noise removal, and normalization. Leveraging the Vision Transformer (ViT) architecture, the proposed method is examined across different patch sizes (8×8, 16×16, 32×32, 64×64) to comprehensively evaluate its performance. Evaluation is conducted on the Brain Tumor MRI Dataset, comparing the proposed method with four pretrained models: ResNet-50, DenseNet121, EfficientNetB0, and Inception V3. Performance metrics including accuracy, precision, recall, and F-score are employed for evaluation. Notably, the proposed Vision Transformer achieves an accuracy of 99% with a patch size of 8×8, 97% with a patch size of 16×16, 96% with a patch size of 32×32, and 93% with a patch size of 64×64. These results highlight the superior efficacy of the proposed approach in brain tumor detection compared to the pretrained models. Comparative analysis with state-of-the-art methods further confirms the proposed approach's superiority, indicating its potential for enhancing diagnostic accuracy in clinical settings. This study underscores the promise of Vision Transformer models in medical imaging tasks and their potential for advancing medical diagnostics.