EEG Signals based motor control using machine learning

Project Title

EEG Signals based motor control using machine learning

Project Area of Specialization Wearables and ImplantableProject Summary

The human brain is a complex organ, which releases ionic discharge when each neuron sends an impulse. This discharge is called Action Potential. The value of Action Potential is different for each decision that the brain makes. Hence, by analyzing action potential pattern, brain activity can be understood and interpreted. Electroencephalography is an electro-psychological monitoring method by which the brain waves can be recorded. By placing electrodes on the scalp, the ionic discharges can be measured using a module of OpenBCI. This project aims to record brain signals using the OpenBCI module and extract useful features from the raw signal. The features will then be processed and used with a machine learning model for control a motor function. Initially, a neural network model will be utilized which can then further extended to a deep learning model.

Project Objectives

• Recording and cleaning of EEG signals
• Extract useful features from the EEG data.
• Train a suitable machine learning model to perform motor control with EEG signals.

Project Implementation Method

Project is implemented by recording live EEG readings from subjects via electrodes. Electrodes are placed on 16 positions on the scalp of the subject, and the data is sent to the Cyton and Daisy boards of OpenBCI module. These boards are connected to a Wi-Fi module, which sends the data from boards to the computer, which is connected to the same network. When the connection is established, the OpenBCI GUI previews the data at run time. The data is saved for later use, by which, the data cleansing is done. Notch filter of 50Hz can be applied via GUI and feature extraction is done by processing in MATLAB. The data is verified by connecting the motor to the system via an embedded system. The project is being undertaken with feedback from physicians from Ayub Teaching Hospital Abbottabad so that a better understanding of the technical requirements and human needs is acquired.

Project is implemented by recording live EEG readings from subjects via electrodes. Electrodes are placed on 16 positions on the scalp of the subject, and the data is sent to the Cyton and Daisy boards of OpenBCI module. These boards are connected to a Wi-Fi module, which sends the data from boards to the computer, which is connected to the same network. When the connection is established, the OpenBCI GUI previews the data at run time. The data is saved for later use, by which, the data cleansing is done. Notch filter of 50Hz can be applied via GUI and feature extraction is done by processing in MATLAB. The data is verified by connecting the motor to the system via an embedded system.

The project is being undertaken with feedback from physicians from Ayub Teaching Hospital Abbottabad so that a better understanding of the technical requirements and human needs is acquired.

Benefits of the Project

Successful implementation of the project will provide a useful framework which can then be incorporated in aid equipment for disabled individuals. The project will also help in providing a better understanding of EEG signals. Besides, it will provide a useful baseline and low indigenous solutions for further extension of action control using EEG signals.

Technical Details of Final Deliverable

Firstly, the project will provide a decent size EEG data for local individuals. The project will provide a framework for processing of the EEG data and machine learning based tool for decision making based on the action potential. More specifically, a working prototype will be presented to perform motor control using EEG signals.

Final Deliverable of the Project HW/SW integrated systemType of Industry Medical Technologies Wearables and ImplantablesSustainable Development Goals Industry, Innovation and InfrastructureRequired Resources

Elapsed time in (days or weeks or month or quarter) since start of the project	Milestone	Deliverable
Month 1	Literature Background	Feasibility report
Month 2	EEG data	Raw data
Month 3	EEG data	Raw data
Month 4	EEG data	Raw data
Month 5	Data processing	Cleansed Data
Month 6	Data processing	Cleansed Data
Month 7	Data processing	Cleansed Data
Month 8	Feature extraction	EEG data regarding motor startup
Month 9	Feature extraction	Final EEG data regarding motor startup
Month 10	Thesis writing	Thesis