Deep Learning-Based Eye-Tracking Analysis for Diagnosis of Alzheimer's Disease Using 3D Comprehensive Visual Stimuli.

Area of Science:

• Neuroscience
• Artificial Intelligence
• Medical Diagnostics

Background:

• Alzheimer's Disease (AD) is a progressive neurodegenerative disorder impacting cognitive functions.
• Early diagnosis of AD is crucial for timely interventions to slow disease progression.
• Current diagnostic methods rely on clinical experience, which has inherent limitations.

Purpose of the Study:

• To investigate the potential of deep learning techniques for diagnosing AD using eye-tracking behaviors.
• To develop and evaluate a novel deep learning model for detecting cognitive abnormalities associated with AD.

Main Methods:

• Collected visual attention heatmaps from participants performing a 3D comprehensive visual task using a noninvasive eye-tracking system.
• Proposed a multilayered comparison convolutional neural network (MC-CNN) to analyze differences in visual attention between AD patients and normal controls.
• Utilized hierarchical residual blocks within MC-CNN to encode eye movement behaviors and integrated them into a distance vector.

Main Results:

• The MC-CNN model achieved an accuracy of 0.84, recall of 0.86, precision of 0.82, and F1-score of 0.83.
• The model demonstrated a high area under the curve (AUC) of 0.90, indicating strong diagnostic performance.
• Significant differences in visual attention patterns were identified between AD patients and normal individuals.

Conclusions:

• The proposed MC-CNN model effectively distinguishes Alzheimer's Disease patients from healthy individuals based on eye-tracking data.
• Eye-tracking behavior analysis, particularly visual attention patterns, holds significant clinical value for early AD detection.
• Deep learning approaches, like MC-CNN, offer a promising avenue for objective and accurate AD diagnosis.