EEGDepressionNet: A Novel Self Attention-Based Gated DenseNet With Hybrid Heuristic Adopted Mental Depression...

Area of Science:

• Neuroscience
• Artificial Intelligence
• Medical Informatics

Background:

• Depression is a leading cause of global disability, impacting public and private health.
• Electroencephalogram (EEG) is a valuable tool for assessing brain activity and analyzing depression.
• Manual EEG analysis for depression is time-consuming and labor-intensive, necessitating automated solutions.

Purpose of the Study:

• To develop a novel, automated deep learning system for depression detection using EEG signals.
• To enhance diagnostic efficiency and assist clinicians in identifying depression.
• To improve upon existing methods for automated depression analysis.

Main Methods:

• EEG signals were sourced from public databases.
• Three feature sets were extracted: spectrograms (using 3D-CNN), raw EEG (using 1D-CNN), and spectral features.
• The Chaotic Owl Invasive Weed Optimization (COIWSO) algorithm was used for optimal feature selection and weight fusion.
• A Self-Attention-based Gated Densenet (SA-GDensenet) model, optimized by COIWSO, performed the final depression detection.

Main Results:

• The developed deep learning model achieved an accuracy of 96% in depression detection.
• The system demonstrated superior performance compared to traditional depression detection models.
• Empirical results confirmed the effectiveness of the automated feature extraction and detection approach.

Conclusions:

• The proposed automated deep learning system offers a highly accurate and efficient method for depression detection using EEG signals.
• This novel approach can significantly assist clinicians in the diagnosis and management of depression.
• The integration of advanced deep learning techniques and optimization algorithms shows promise for future mental health diagnostics.