Prosthetic control system based on motor imagery

Area of Science:

• Neuroscience
• Biomedical Engineering
• Rehabilitation Technology

Background:

• Brain-computer interfaces (BCIs) offer functional replacement by interpreting brain activity to control external devices.
• Electroencephalography (EEG) signals, particularly those related to motor imagery, are crucial for intent classification in BCIs.

Purpose of the Study:

• To enhance the accuracy of intent classification in BCIs by processing EEG signals from motor imagery.
• To develop a functional prosthetic control system utilizing a refined EEG signal processing pipeline.

Main Methods:

• EEG signals were decomposed using db4 wavelet basis and denoised with a soft threshold method.
• Sample entropy was employed for feature extraction from denoised signals, focusing on event-related synchronisation/desynchronisation (ERS/ERD) periods.
• Classifiers including a backpropagation (BP) neural network were evaluated for intent classification.

Main Results:

• Wavelet denoising and sample entropy feature extraction significantly improved signal quality and feature relevance.
• The backpropagation (BP) neural network achieved the highest classification accuracy at 93%.
• A wirelessly controlled prosthetic system was successfully demonstrated, enabling hand and wrist movements.

Conclusions:

• The developed BCI system, utilizing EEG and BP neural networks, demonstrates high accuracy for prosthetic control.
• This technology holds significant potential for restoring daily living activities for amputees.
• The study provides a valuable foundation for future advancements in neurorehabilitation and assistive device development.