Related Experiment Video
Updated: Jul 13, 2025
05:19
Author Spotlight: Therapeutic Benefit of Closed-Loop Deep Brain Stimulation in Depression Treatment
Published on: July 7, 2023
2.3K
AMGCN-L: an adaptive multi-time-window graph convolutional network with long-short-term memory for depression
Han-Guang Wang1, Qing-Hao Meng1, Li-Cheng Jin1
1Tianjin University, Tianjin Key Laboratory of Process Measurement and Control, Institute of Robotics and Autonomous Systems, School of Electrical and Information, Tianjin, People's Republic of China.
Journal of Neural Engineering
|October 16, 2023
Summary
This study introduces a novel deep learning model, AMGCN-L, for objective depression diagnosis using electroencephalogram (EEG) signals. The model achieved high accuracy, offering a promising tool for clinical depression detection.
Area of Science:
- Neuroscience and Artificial Intelligence
- Computational Psychiatry
- Biomedical Signal Processing
Background:
- Depression is a prevalent, disabling mental disorder with high recurrence rates.
- Current diagnostic methods rely on subjective questionnaires, highlighting the need for objective approaches.
- Deep learning offers potential for analyzing complex biological signals for improved diagnostic accuracy.
Purpose of the Study:
- To develop and evaluate an objective deep learning-based method for depression detection using electroencephalogram (EEG) signals.
- To investigate the efficacy of combining graph convolutional networks (GCN) and long-short-term memory (LSTM) for analyzing brain functional connectivity and spatiotemporal features in EEG data.
- To provide a supplementary tool for clinical depression diagnosis and treatment.
Main Methods:
- Proposed an adaptively multi-time-window graph convolutional network (GCN) with long-short-term memory (LSTM), termed AMGCN-L.
- The network utilizes an adaptive multi-time-window graph generation block to capture brain functional connectivity across different time periods.
- Employed GCN for spatial feature extraction and LSTM for temporal feature extraction from EEG signals.
Main Results:
- Evaluated AMGCN-L on two public EEG datasets: the patient repository for EEG data and computational tools, and the multi-modal open dataset for mental-disorder analysis.
- Achieved high depression recognition accuracies of 90.38% and 90.57% using tenfold cross-validation on the respective datasets.
- Demonstrated the effectiveness of GCN and LSTM in extracting spatial and temporal features for depression detection.
Conclusions:
- The proposed AMGCN-L model effectively utilizes GCN and LSTM for spatial and temporal feature extraction from EEG signals, respectively.
- Exploration of brain connectivity and spatiotemporal features significantly benefits depression detection.
- This deep learning approach offers a valuable, objective support system for clinical depression diagnosis and subsequent treatment.
