Convolutional neural networks for multi-class brain disease detection using MRI images.

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Neurology

Background:

• Brain disorders can impair critical functions like thinking, speech, and movement.
• Early detection of brain diseases is vital for effective treatment.
• Manual analysis of magnetic resonance imaging (MRI) for brain abnormalities is challenging and time-consuming, especially for subtle early-stage changes.

Purpose of the Study:

• To develop and evaluate deep learning models for the automatic classification of brain abnormalities in MRI images.
• To compare the performance of various pre-trained deep learning models, including AlexNet, Vgg-16, ResNet-18, ResNet-34, and ResNet-50, for this task.

Main Methods:

• Employed five pre-trained deep learning models: AlexNet, Vgg-16, ResNet-18, ResNet-34, and ResNet-50.
• Utilized these models to automatically classify MRI images into five categories: normal, cerebrovascular, neoplastic, degenerative, and inflammatory diseases.
• Compared the classification performance of the selected state-of-the-art architectures.

Main Results:

• The ResNet-50 model achieved the highest classification accuracy of 95.23% ± 0.6 among the evaluated models.
• This demonstrates the effectiveness of deep learning, particularly ResNet-50, in analyzing brain MRI scans.

Conclusions:

• Deep learning models, specifically ResNet-50, show significant promise for accurate and automated classification of brain abnormalities in MRI.
• The developed model is ready for further testing on larger datasets and can assist clinicians in validating their diagnostic findings.