EEG Signal Classification Using Convolutional Neural Networks on Combined Spatial and Temporal Dimensions for BCI...

Area of Science:

• Neuroscience
• Machine Learning
• Signal Processing

Background:

• Electroencephalography (EEG) signal classification is crucial for Brain Computer Interface (BCI) systems.
• Existing methods often rely on time or frequency domain features, with limited exploration of combined spatial and temporal dimensions.
• Designing efficient algorithms based on prior knowledge for complex brain dynamics is challenging.

Purpose of the Study:

• To develop a novel method for EEG signal classification by integrating spatial and temporal dimensions.
• To utilize a 2D AlexNet Convolutional Neural Network (CNN) for learning EEG features without prior knowledge.
• To improve the accuracy of BCI systems in classifying different mental tasks.

Main Methods:

• EEG signals were transformed into 2D topographic maps, incorporating spatial and temporal information.
• Topographic maps from different time indices were cascaded to form 2D images representing time windows.
• A 2D AlexNet CNN was employed to learn features directly from these 2D EEG images.

Main Results:

• The proposed method achieved an average classification accuracy of 81.09% on the BCI Competition IV dataset 2a.
• This accuracy represents a 4% improvement over previous state-of-the-art methods for the same dataset.
• The conversion to a 2D image classification problem enhanced classification accuracy for BCI systems.

Conclusions:

• Representing EEG signals as 2D topographic maps allows CNNs to learn subtle spatial-temporal features effectively.
• This approach surpasses traditional time or frequency domain features in representing complex mental tasks.
• The study demonstrates the efficacy of a 2D CNN approach for advancing BCI accuracy and capabilities.