Classification of Motor Functions from Electroencephalogram (EEG) Signals Based on an Integrated Method Comprised of...

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Brain-computer interfaces (BCIs) hold promise for controlling prosthetic devices.
• Electroencephalogram (EEG) signals offer a non-invasive method for capturing brain activity.
• Classifying motor intentions from EEG is crucial for effective prosthetic control.

Purpose of the Study:

• To develop and evaluate an algorithm for classifying six motor functions from EEG signals.
• To investigate the efficacy of combining Common Spatial Pattern (CSP) filtering and Continuous Wavelet Transform (CWT) for motor imagery classification.
• To assess the potential of EEG-based BCIs for controlling upper limb prosthetic devices.

Main Methods:

• An algorithm integrating CSP filtering and CWT was applied to EEG data from grasp-and-lift events.
• EEG signals from motor cortex and parietal regions were processed through band-pass filtering (7-30 Hz) and CSP filtering.
• Scalograms generated by CWT were converted to RGB images and classified using GoogLeNet.

Main Results:

• The algorithm achieved high classification performance with average precision (94.8%), sensitivity (93.5%), specificity (94.7%), and accuracy (94.1%).
• The average area under the ROC curve was 0.985, indicating robust classification capabilities.
• The study demonstrated successful classification of six motor functions from EEG signals.

Conclusions:

• The proposed CSP and CWT-based algorithm shows excellent performance in classifying motor EEG signals.
• This approach shows significant potential for developing advanced EEG-based BCIs for prosthetic limb control.
• The findings support the use of EEG as a viable input signal for controlling upper limb prosthetics.