Projects:2019s1-155 Brain Computer Interface Control for Biomedical Applications
Project Team
Students
Zhiying Lin
Kayla Wahlstrom
Supervisors
Associate Professor Mathias Baumert
Mr David Bowler
Introduction
Individuals with severed or deteriorating neuronal pathways caused by injury or disease may encounter daily tasks with great difficulty. Research and development throughout the last decades has engendered a new technology, known as a brain-computer interface (BCI), through which the brain can interact with to communicate with the environment. In BCI technology, commands an individual sends to the environment are not processed through the brain's normal output pathways of peripheral nerves and muscles. Rather, a BCI operates by providing output pathways to communicate with different types of applications that can augment, improve or restore the output of the individual's central nervous system.
Motivation
Stroke is one of the leading causes of disability in Australia [2], with 65% of stroke survivors presenting persistent disability following the initial cerebrovascular episode. In addition, motor impairments are the most prevalent post-stroke disability [3]. This decrease in movement coordination and control means the patient’s ability to perform daily tasks unassisted is impaired [2].
Stroke occurs when blood supply to the brain is disrupted, depriving the brain cells of oxygen and vital nutrients. Brain cells not receiving oxygen and these vital nutrients will die, causing damage to the brain known as a cerebral infarct (necrosis, or tissue death) [2]. Tissue death in the motor regions of the brain impairs the brain's ability to execute movement.
Methods such as exercise and training have predominantly been used as stroke rehabilitative strategies to restore motor function, however, training approaches to improve the effects of these methods are currently being investigated and developed for post-stroke disability [3]. Fundamental to this investigation is the knowledge of neuroplasticity and the intrinsic behaviour of the brain and its motor regions following stroke.
This project will investigate a non-invasive brain-computer interface (BCI) to provide users with the ability to control motor movement via an external limb, and with long-term use, aim to promote neuroplasticity to improve or restore motor movement in post-stroke patients.
Previous UofA student Work
Work completed by previous students included design of a flexible headset using an elastic strap to hold the electrodes. Software was also developed, but due to time constraints, some features such as classifiers and the function of Data Tab on the software framework were not fully implemented. Also developed was a new mechanical 3D-printed glove, and testing of the system was completed using third-party software platform OpenViBE.
Objectives
- Redesign the EEG signal acquisition headset. Electrodes will be positioned along the motor cortex, in a configuration that allows for detection of motor imagery of the hand. Electrodes should sit flush with the headset structure.
- Redesign components of the 3D-printed hand orthosis into a wearable design.
- Create a new software framework for signal processing. This will involve implementing a signal acquisition capability, feature extraction and feature classification.
Background
What is a Brain-Computer Interface?
A brain-computer interface (BCI) is a collaboration between the brain and a device that allows signals from the brain to direct some external activity, such as controlling a cursor or a prosthetic limb. This interface enables direct communication between the brain and the controlled object.
Neuroimaging approaches in BCI[1]
1. Electroencephalography (EEG) measures the difference in potential on the scalp due to neural activity, which is the sum of thousands or millions of cortical neurons' postsynaptic excitatory potential and inhibitory potential.
2. Magnetoencephalography (MEG) measures magnetic field differences related to neuron activity.
3. Functional Magnetic Resonance Imaging (fMRI) detects changes in local cerebral blood volume, cerebral blood flow and oxygenation levels during neuron activation.
4. Near Infrared Spectroscopy (NIRS) uses the characteristics of light in the near infrared spectrum to penetrate the skull to a considerable depth for the study of brain metabolism. It can detect the change of haemoglobin concentration in the process of local nerve activity in different wavelengths of weak light intensity.
Method
- Conducted offline analysis in MATLAB using published BCI data. This analysis included feature extraction using common spatial pattern filters and classification using linear discriminant analysis.
- Conducted offline analysis in Microsoft Visual Studio in C++ using published BCI data. Created EEG signal acquisition capability.
- Conducted offline analysis using own acquired signal data.
- Created a graphical user interface.
Results
A 71% classification accuracy was obtained using the published BCI data, while a much lower classification accuracy of 55% was obtained using own acquired data from the headset.
References
1. Byoung-Kyong Min, Matthew J. Marzelli and Seung-Schik Yoo (2010) Neuroimaging-based approaches in the brain–computer interface, Available at: https://www.researchgate.net/publication/46109898 [Accessed: 12/4/2019].
2. Strokefoundation.org.au. (2019). About Stroke. [online] Available at: https://strokefoundation.org.au/About-Stroke [Accessed 14 Apr. 2019].
3. Dimyan, M. and Cohen, L. (2019). Neuroplasticity in the context of motor rehabilitation after stroke. [online] Available at: https://www.nature.com/articles/nrneurol.2010.200 [Accessed 14 Apr. 2019].
