Attention Gated-VGG with deep learning-based features for Alzheimer's disease classification

Area of Science:

• Neuroscience
• Medical Imaging
• Artificial Intelligence

Background:

• Alzheimer's disease (AD) is a leading neurodegenerative disorder causing cognitive decline and dementia.
• Early detection of AD is critical for timely intervention and management.
• Current diagnostic methods require improvement for earlier and more accurate identification.

Purpose of the Study:

• To develop and evaluate a novel deep learning model for the accurate classification of Alzheimer's disease.
• To leverage advanced image processing and feature extraction techniques for improved AD detection.
• To assess the efficacy of the Attention Gated-VGG model in differentiating AD patients from controls.

Main Methods:

• Image pre-processing including resizing and median filtering.
• Data augmentation to enhance the training dataset.
• Feature extraction using a Whale Optimization Algorithm (WOA)-based ResNet and Convolutional Neural Network (CNN).
• Classification using the proposed Attention Gated-VGG deep learning model.

Main Results:

• The Attention Gated-VGG model achieved high classification performance.
• Achieved an accuracy of 96.7%, sensitivity of 97.8%, and specificity of 96.3%.
• Outperformed conventional methodologies in AD classification tasks.

Conclusions:

• The Attention Gated-VGG model demonstrates significant potential as a tool for early Alzheimer's disease classification.
• The proposed deep learning approach offers a promising avenue for improving diagnostic accuracy in neurodegenerative diseases.
• This technique could aid clinicians in making earlier and more informed decisions regarding AD patient care.