Feature-level fusion based on spatial-temporal of pervasive EEG for depression recognition

Area of Science:

• Neuroscience
• Computational Psychiatry
• Biomedical Engineering

Background:

• Depression is a prevalent condition with significant disability and fatality rates.
• Current diagnostic methods (questionnaires, interviews) are time-consuming, labor-intensive, and subjective.
• Early identification and intervention are crucial to prevent irreversible brain damage.

Purpose of the Study:

• To develop an accurate, convenient, and effective method for depression recognition.
• To overcome the limitations of traditional subjective diagnostic approaches.
• To establish the social and scientific value of objective depression detection.

Main Methods:

• A novel framework for depression recognition using feature-level fusion of spatial-temporal electroencephalography (EEG) data.
• EEG data mapped to a visibility graph (VG) to extract spatial and temporal features.
• A cascade forest model utilizing feature-level fusion with assigned contribution coefficients for enhanced recognition.

Main Results:

• The proposed feature-level fusion method with contribution coefficients demonstrated superior depression recognition capabilities.
• Achieved a highest accuracy of 92.48%, outperforming single-feature methods and fusion without coefficients.
• Experimental data collected from 26 depressed patients and 29 healthy controls.

Conclusions:

• The developed feature-level fusion method serves as an effective computer-aided tool for rapid clinical depression diagnosis.
• Highlights the potential of integrating EEG data and machine learning for objective mental health assessment.
• Supports the advancement of technology in psychiatric diagnostics.