Deep Neural Network for EEG Signal-Based Subject-Independent Imaginary Mental Task Classification

Area of Science:

• Neuroscience
• Biomedical Engineering
• Machine Learning

Background:

• Electroencephalography (EEG) signal analysis is crucial for patients with motor impairments.
• Subject-independent frameworks enable mental task identification without subject-specific training data.
• Deep learning models excel at analyzing complex spatial and time-series EEG data.

Purpose of the Study:

• To propose a deep neural network (DNN) model for classifying imagined mental tasks from EEG signals.
• To develop a non-invasive, subject-independent framework for mental task identification.
• To evaluate the model's performance on a benchmark EEG dataset.

Main Methods:

• EEG signals were spatially filtered using a Laplacian surface.
• Principal Component Analysis (PCA) was applied for high-dimensional data reduction and feature extraction.
• A deep neural network (DNN) was trained using averaged Power Spectrum Density (PSD) values for cross-subject classification.

Main Results:

• The proposed DNN model achieved an accuracy of 77.62% in mental task classification.
• The model successfully extracted mental task-specific features from EEG data.
• The non-invasive approach demonstrated effectiveness in identifying imagined tasks.

Conclusions:

• The developed cross-subject classification framework demonstrates superior performance compared to state-of-the-art algorithms.
• The study validates the efficacy of deep learning for accurate mental task identification from EEG signals.
• The proposed method offers a promising solution for brain-computer interfaces in individuals with limited mobility.