Comparative analysis for accurate multi-classification of brain tumor based on significant deep learning models

Brain tumours are a significant health concern, often resulting in severe cognitive and physiological impairments. Accurate detection and classification of brain tumours, including glioma, meningioma, and pituitary tumours, are crucial for effective treatment. In this study, we present a comprehensive approach for brain tumor classification using MRI scans and deep learning models, specifically focusing on the use of Convolutional Neural Networks (CNN), Swin Transformer, and EfficientNet. MRI scans from four categories, including healthy brains, underwent pre-processing using normalisation, resizing, and data augmentation to mitigate problems associated with variability in image quality and tumor manifestation. Every deep learning model was trained on the pre-processed dataset, and their performance was assessed using accuracy, sensitivity, and specificity measures. The findings demonstrate that the Swin Transformer and EfficientNet models achieved superior classification testing accuracy, which are 98.08 % and 98.72 % respectively, surpassing conventional CNNs, which achieve 95.16 % testing accuracy. EfficientNet exhibited an optimal combination between computational economy and classification performance, making it an exemplary choice for resource-limited settings. Our results underscore the capability of sophisticated deep learning architectures to enhance diagnostic precision in brain tumor classification tasks.