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

Area of Science:

• Medical imaging analysis
• Artificial intelligence in healthcare
• Neuro-oncology

Background:

• Brain tumors pose significant health risks, necessitating accurate detection and classification.
• Glioma, meningioma, and pituitary tumors require precise identification for effective treatment strategies.

Purpose of the Study:

• To evaluate the efficacy of deep learning models, including Convolutional Neural Networks (CNN), Swin Transformer, and EfficientNet, for brain tumor classification using MRI scans.
• To compare the performance of these models in terms of accuracy, sensitivity, and specificity.

Main Methods:

• MRI scans from four categories (including healthy brains) were pre-processed using normalization, resizing, and data augmentation.
• Three deep learning models (CNN, Swin Transformer, EfficientNet) were trained on the pre-processed dataset.
• Model performance was evaluated using key metrics such as accuracy, sensitivity, and specificity.

Main Results:

• Swin Transformer and EfficientNet achieved superior classification testing accuracies of 98.08% and 98.72%, respectively.
• Conventional CNNs achieved a testing accuracy of 95.16%, lower than the advanced models.
• EfficientNet demonstrated a favorable balance between computational efficiency and high classification performance.

Conclusions:

• Sophisticated deep learning architectures, particularly Swin Transformer and EfficientNet, significantly enhance diagnostic precision for brain tumor classification.
• EfficientNet is a highly suitable model for brain tumor classification, especially in resource-constrained environments.
• The study highlights the potential of advanced AI in improving the accuracy and efficiency of brain tumor diagnosis.