Revolutionizing Alzheimer's disease detection with a cutting-edge CAPCBAM deep learning framework.

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Neurology

Background:

• Early and accurate diagnosis of Alzheimer's disease (AD) is critical for timely intervention and treatment.
• Deep learning models have shown potential in classifying AD from medical images, but advancements are needed for improved accuracy and generalization.

Purpose of the Study:

• To introduce CAPCBAM, a novel deep learning framework integrating Capsule Networks and Convolutional Block Attention Module (CBAM) for enhanced Alzheimer's disease classification.
• To evaluate the performance of CAPCBAM on the Alzheimer's Disease Neuroimaging Initiative (ADNI) dataset.

Main Methods:

• Standardized preprocessing of MRI images.
• Feature extraction using Capsule Networks to preserve spatial hierarchies and intricate feature relationships.
• Refinement of feature maps using CBAM with channel and spatial attention mechanisms to highlight clinically relevant regions.

Main Results:

• CAPCBAM achieved a classification accuracy of 99.95% on the ADNI dataset.
• The framework demonstrated high precision (99.8%), recall (99.8%), AUC (0.99), and F1-Score (99.92%).
• CAPCBAM showed advantages over conventional CNNs in model generalization and reduced information loss.

Conclusions:

• CAPCBAM offers superior classification performance for early Alzheimer's disease detection compared to existing methods.
• The robust integration of Capsule Networks and CBAM leads to improved feature extraction and faster model convergence.
• CAPCBAM presents a promising advancement for the clinical diagnosis of Alzheimer's disease.