Related Experiment Video
Updated: May 9, 2025
05:19
Author Spotlight: Therapeutic Benefit of Closed-Loop Deep Brain Stimulation in Depression Treatment
Published on: July 7, 2023
2.1K
A depression detection approach leveraging transfer learning with single-channel EEG.
Chengyuan Sun1, Mingjuan Guan1, Keyu Duan1
1Institute of Artificial Intelligence, Anhui University of Science and Technology, No. 168, Taifeng Street, Huainan 232001, Anhui, People's Republic of China.
Journal of Neural Engineering
|May 2, 2025
Summary
A new deep transfer learning model accurately detects major depressive disorder (MDD) using single-channel electroencephalography (EEG) signals. This approach overcomes limitations of multichannel EEG, enabling more accessible and adaptable depression screening.
Area of Science:
- Neuroscience and Computational Psychiatry
- Machine Learning Applications in Healthcare
Background:
- Major Depressive Disorder (MDD) is a prevalent mental health condition requiring objective diagnostic tools.
- Current electroencephalography (EEG) based depression detection often relies on multichannel signals, limiting practical application.
- Variability in EEG data acquisition and limited datasets pose challenges for model generalizability.
Purpose of the Study:
- To develop a novel depression detection model using single-channel EEG signals.
- To leverage transfer learning and deep neural networks for improved accuracy and adaptability.
- To address the constraints of multichannel EEG and data variability in depression diagnosis.
Main Methods:
- A pretrained ResNet152V2 network was adapted with added layers for feature extraction.
- EEG signal sequences were converted into images, reframing the task as image classification.
- A cross-subject experimental strategy was employed for model training and validation.
Main Results:
- The model achieved near-perfect accuracy (approaching 100%) in identifying depression using single-channel EEG.
- Demonstrated robust performance across four diverse, publicly available depression EEG datasets.
- Showcased significant adaptability to variations in EEG data caused by different acquisition contexts.
Conclusions:
- Deep transfer learning holds substantial promise for analyzing EEG signals in mental health.
- The developed model offers a viable, efficient approach for single-channel EEG-based depression detection.
- This research provides a foundation for future innovations in early diagnosis of psychiatric disorders.
