Focal-DenseNet: A Risk Assessment Framework for Alzheimer's Disease in Heterogeneous MRI Data.

Area of Science:

• Neuroscience
• Artificial Intelligence
• Medical Diagnostics

Background:

• Alzheimer's disease (AD) is a progressive neurodegenerative disorder with significant public health impact.
• Current understanding of AD causes and treatments remains limited, necessitating improved diagnostic tools.
• Effective risk assessment is crucial for AD prevention and management.

Purpose of the Study:

• To develop and evaluate an optimized deep learning model for Alzheimer's disease risk assessment and diagnosis.
• To enhance the accuracy and efficiency of early AD detection through advanced AI techniques.

Main Methods:

• Data preprocessing was performed to ensure quality for subsequent analysis.
• A novel Focal-DenseNet model, integrating DenseNet with a focal loss function, was developed.
• The model was trained and tested, with performance compared against DenseNet and other deep learning models (VGG16, ResNet).

Main Results:

• The Focal-DenseNet model achieved a test set accuracy of 98.98%.
• The proposed model demonstrated superior performance across multiple indicators, including accuracy, AUC, precision, and recall, compared to existing models.
• The focal loss function integration significantly improved diagnostic prediction capabilities.

Conclusions:

• The Focal-DenseNet model offers a highly accurate and efficient tool for early Alzheimer's disease risk assessment and diagnosis.
• This AI-driven approach provides significant clinical value for improving AD prevention and treatment strategies.
• The study contributes to reducing the global health burden of Alzheimer's disease through advanced technological support.