Implications of Convolutional Neural Network for Brain MRI Image Classification to Identify Alzheimer's Disease

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Neurology

Background:

• Alzheimer's disease is a progressive neurodegenerative disorder primarily affecting individuals over 60.
• Early detection through medical image analysis is crucial for timely diagnosis and treatment planning.
• Magnetic Resonance Imaging (MRI) provides valuable data for identifying pathological changes associated with Alzheimer's.

Purpose of the Study:

• To evaluate the efficacy of the TinyNet architecture for classifying Alzheimer's disease using MRI datasets.
• To demonstrate TinyNet's capability in handling small-scale image classification tasks efficiently.
• To improve diagnostic accuracy and reduce computational complexity compared to larger neural networks.

Main Methods:

• Utilized publicly available Alzheimer's disease MRI datasets from Kaggle.
• Trained the TinyNet architecture, optimized for small-scale image classification.
• Employed transfer learning techniques and fine-tuning for enhanced model performance.
• Conducted comparative analysis against existing methods to validate TinyNet's applicability.

Main Results:

• Achieved 98% accuracy on training MRI datasets with a 2% error rate.
• Attained 80% accuracy on validation MRI datasets with a 20% error rate.
• Demonstrated reduced convergence time and improved generalization capabilities.
• Showcased TinyNet's effectiveness despite a lower parameter count than traditional networks.

Conclusions:

• TinyNet is a viable and efficient architecture for Alzheimer's disease classification from MRI scans.
• The model offers a promising approach for early and accurate diagnosis of Alzheimer's disease.
• Further research and fine-tuning can potentially enhance the model's diagnostic performance.