Neurorehabilitation of Paralysis Patients using Brain Computer Interfaces

Abstract

Brain-Computer Interfaces (BCI) acquire and translate brain signals from a patient to carry out some action on an external device. BCIs may collect these data signals through invasive, semi-invasive, or non-invasive techniques. Such techniques include electroencephalography (EEG), functional magnetic resonance imaging (fMRI), and Electrocorticography (ECoG). This review will aim to understand the development of BCIs and their associated methods of use throughout the years and highlight the potential BCIs are capable of for patients who are paralyzed as a result of stroke or neurological injuries. It will further explore the current viability of BCI technology for commercial use and future areas of research in order to make this technology more readily available to the public.

Introduction

Damage or disorder to the nervous system can cause paralysis, and neurorehabilitative treatments for these conditions do not work for everyone. Stroke is the most common of these, a leading cause of disability in the United States, afflicting almost 800,000 people in the US every year. A major risk factor for stroke is age, with more than 70% of all stroke cases occurring in those 65 and older. The US Census Bureau predicts that within a decade, 21% of all Americans will be in this age range, and thus at high risk for stroke. With the greying of the US population, it is more important more than ever before to review the existing literature on neurorehabilitative treatments in order to develop novel treatments for patients who cannot be treated with conventional methods. Currently, constraint induced movement therapy (CIMT) is an effective treatment for partial paralysis, but it cannot help the 30–50% of stroke patients who have a fully paralyzed limb. For other paralyzing conditions such as spinal cord injury, traumatic brain injury (TBI), and amyotrophic lateral sclerosis (ALS), treatments can be nonexistent or ineffective. Thus, treatments need to be developed to address paralysis that works in nontraditional pathways.^[1][2][3][4]
A powerful new technology, brain-computer interface (BCI), has emerged to become one of the most promising treatments for neurorehabilitation of paralysis . Unlike pharmaceutical drugs or physical therapy, which manipulate biochemical pathways or parts of the body, BCIs read, interpret, and respond to an individual’s brain activity in order to control another device. Often, this technology controls a robotic apparatus, computer, or device to relay signals to another region of the brain. Because of their direct connection to the brain, BCIs have endless potential and have been studied for many different applications ranging from video gaming to clinical treatments.
The groundwork for the first BCI was laid by a series of discoveries about brain physiology in the 19th century leading to Hans Berger’s 1924 invention of EEG, a non-invasive method of measuring electrical activity in the brain. Then, in the 1970s, a lab at UCLA coined the term “brain-computer interface,” using EEG as the method of collecting electrical signals from the brain. Since then, EEG has been widely used for BCIs, and several variants have been developed which provide better signals, most notably electrocorticography (ECoG). Other methods of signal acquisition, such as EMG and fMRI, will be further discussed in this review.^[5][6]
Experiments in the 1980s explored the technology of BCIs while studying the link between motor movements and neuronal signals. One major lab that has studied this is Apostolos Georgopoulos’ lab at Johns Hopkins University, which found that individual neurons worked together to complete a task: in this case, turning a door handle. The topic of motor control is one of the most heavily researched topics using BCIs and a crucial element in treating paralysis.^[7]
BCIs have been developed for treating stroke patients, and implementations of BCIs are varied. Two strategies of neurorehabilitative BCIs are assistive BCIs, which act as a substitute for lost motor functions, and rehabilitative BCIs, which rely on neuroplasticity to encourage the brain to recover lost motor functions. A 2017 review of EEG-based assistive BCIs which enabled users to control robotic hand devices for stroke rehabilitation concluded that BCI-based robotic assistance for stroke recovery was beneficial and promising because they helped improved communication and movement. This indicates promise as these systems continue to be developed and improved. A 2018 systematic review of rehabilitative BCIs developed for post-stroke recovery found that many of the BCIs could alter brain function, and the change in brain signalling corresponded with a positive behavioral change.^[8][9][10]
Aside from the approach to neurorehabilitation, there are many different aspects of the design of BCIs that affect its efficacy. As the population of the US gets older, it is even more crucial that novel treatments such as BCI rehabilitation be studied and developed. This review examines and evaluates BCI’s individual components and potential applications for neurorehabilitation in those paralyzed due to brain injury. Analysis and discussion of various signal acquisition methods, namely fMRI, EEG, and EcoG, are followed by a reflection on the limitations that exist in current research on this technology and possible pathways for future research.

The Methods of BCI Data Acquisition

BCIs can be described as a communication bridge between the brain and an external system, and signal acquisition is the first step in building this bridge. The first distinction of BCIs is whether signals are acquired invasively or non invasively. Invasive BCIs use micro-electrodes directly implanted in the cortex of the brain and deliver the greatest quality signals, but they require delicate neurosurgery procedures to function. By contrast, in noninvasive BCIs, electrodes or sensors are simply placed above the patient\’s scalp and the electrical signals produced by the brain are recorded. Therefore, noninvasive techniques are safer and cheaper, and due to this are more common. There are a variety of noninvasive signal acquisition methods, the most prominent and relevant to the neurorehabilitation of paralysis patients being EEG and fMRI. Another signal acquisition method is ECoG, which is classified as a semi-invasive method.^[11]

fMRI

fMRI is a technique used to measure brain activity by analyzing changes in the blood flow. The reasoning behind monitoring blood flow is that it may correlate with increased brain activity in a specific region. Additionally, oxygen is carried to the brain through hemoglobin in capillary red blood cells. Hemoglobin acts differently depending on whether it\’s oxygenated or deoxygenated. When oxygenated, hemoglobin is diamagnetic (repelled by the magnetic field). On the other hand, when hemoglobin is deoxygenated, it’s paramagnetic (attracted by the magnetic field). These differences in the magnetic properties lead to variance in the signal of the blood. Since blood oxygenation changes in response to neuronal activity, the differences are used to measure brain activity.^[12]
fMRI holds several advantages: it’s non-invasive, doesn’t involve radiation, and it has exceptional spatial and reliable temporal resolution. New developments in fMRI have led to possible neurorehabilitative use. By utilizing fMRI to observe the aspects of task-related changes in the neural activation, doctors can provide real time neurofeedback through the form of neural signals to the participant. Neurofeedback provides real time fMRI data so the patient can attempt to self-regulate different functions, in the context of paralysis, movement. There are many advantages of using real-time fMRI based neurofeedback. Unlike other non-invasive techniques, it can be used for specific subcortical regions. The use of fMRI based neurofeedback is still in its beginnings. Future growth in this technology could even lead to neuroplasticity in brain networks for certain neurological conditions.^[13]
Partial paralysis may occur as the result of a stroke, and there has been research about the rehabilitative opportunities behind an fMRI based BCI. The method of treatment with fMRI-BCI involves targeting a certain region of the brain, for example the parts that control movement or memory. Hemiparesis is a type of paralysis in which one side or part of the body loses motor function, and may be caused by a stroke. A study conducted by Yoo and Jolesz describes the method of using fMRI also as a source of neurofeedback. In the study, several patients were asked to conduct a simple motor task, and they were given real time visual feedback from the BCI to understand which parts of the brain were in use and which were not. The results of the study pointed toward patients focusing and successfully activating previously inactive parts of the brain to increase motor function. In the context of paralysis and stroke, an fMRI-BCI could be used to retrain a part of the brain that had lost its function. This application is significant because it is an example of assistive neurorehabilitation, however it is important to note that the patient was not completely paralyzed.^[14][15]

EEG

Electroencephalography (EEG) measures electrical signals produced by the brain using either wet electrodes, which requires a gel, or dry electrodes, which is more convenient but may not be as strong. EEG is the most widely used and the cheapest non-invasive method used in BCIs. Because it is small and relatively easy to set up, there are a number of commercial EEG sets available.
Both stroke and TBI may leave the patient in a state known as locked in state (LIS) , when the person is still conscious, but has lost motor function. The goal of Han et al. was to develop an EEG based endogenous brain computer interface that could allow for online communication for people in LIS. EEG was selected because of its cheaper cost and high resolution compared with fMRI and magnetoencephalography (MEG). The team focused on developing an endogenous BCI model, which would use neural signals stemming from the user instead of external signals, as patients in LIS have very limited motor function. The patient, a 62 year old woman in LIS, was given three mental imagery tasks. The system provided a useful interface for communication, and was the first endogenous BCI to accurately communicate with a patient with LIS. The most significant aspect of this study was the usage of the endogenous system, which was more accurate than previous exogenous interfaces, but requires pre-training sessions to be effective.^[16]

ECoG

ECoG, unlike the aforementioned methods, does not acquire signals noninvasively. It requires a craniotomy because the electrodes are placed directly on the brain surface and not the scalp. Although this procedure is more expensive, it is less risky compared to regular invasive methods, because the incision is not as deep. It also has advantages over non-invasive methods including better signal quality. Severe forms of paralysis may leave patients unable to move or communicate. BCIs provide a potential interface for these patients to communicate. Pandarinath et. al outlines the development and testing of an intracortical brain-machine interface (iBCI) that allowed a patient to type messages to communicate. This was done through a “point and click” method. The authors describe the “point and click” method as similar to using a normal computer mouse, however instead of physically moving the mouse, the iBCI would allow the patient to move the cursor simply from brain signals. To quantify the performance of the iBCI, the typing rate was measured, and the apparatus successfully allowed users to type clear messages. One very intriguing future proposition is connecting the interface of the BCI to common technology, such as a phone or tablet. This could allow for cheaper and simpler communication for paralysis patients.^[17]

Limitations of BCIs and Evaluation of Viability

Currently, brain-computer interfaces are in a very experimental stage of development and commercial implementation. The future of BCIs is ultimately dependent on progress in three main areas: deployment of efficient and stable signal-acquisition hardware, BCI validation and information processing techniques, and proven prior reliability and results-driven BCI technologies for a multitude of user segments. Unfortunately, even with the problems outlined with numerous possible solutions, modern BCIs’ reliability for all but the simplest applications remains inadequate.^[18]
Signal Acquisition hardware is heavily dependent on sensors and associated hardwares that are meant to acquire neural signals. Ideally, for non-invasive machines, BCIs should have dry electrodes that do not require any type of conductive gel nor skin abrasion and are also small and fully portable with comfortable and cosmetically acceptable implantations. As a guideline, these hardwares also need to be able to function for many hours and even days without maintenance, have a small margin of error, and perform without interruption in a variety of environments. These systems, with so much expected of them, need to be sustainable, functional, and reliable for years and possibly even decades.
As BCI technology advances, the efficacy, practicality, and impact on the user’s quality of life have been called into question many times. Current BCI technology is used to treat patients with severe disorders, which account for a very limited user population. With a limited user population, minimal interest has been shown by corporate entities to either promote or produce the product. However, many researchers are confident that EEG based BCI technology will be available for commercial use, thus prompting many upcoming researchers to focus on the sensor development and the overall system performance of current non invasive BCI technology.^[19] BCI technology additionally relies on the electrical signals produced by the brain in order to function. A major challenge that many BCI researchers are encountering is the low Information Transfer Rate (ITR) that current BCI technology has. ITR is the metric used to calculate the performance of BCI technology. Currently, most BCI technology can only achieve relatively low ITR, enough for the BCI technology to produce the most basic tasks,and are unable to detect pathways for either emotion or mental state. However, when higher ITR levels are achieved, BCI technology would be capable of more sophisticated applications.^[20]
Current BCI research has mainly focused on the short term effects of BCI technology. For BCI technology to potentially be produced for commercial use, research needs to focus on the potential health risks and brain damage for prolonged use of BCI tech. With continuous use of various neural pathways with the aid of BCI technology, we have to be sure that the patient would experience any health risks as a result of over sensation in the various neural pathways involved. Current BCI research has focused mainly on BCI effects on motor and visual sensory organs, where further research needs to be conducted upon other neural pathways in order for BCI technology to be reliable in the commercial setting. A future direction could be to integrate BCIs in a way such that they are as reliable as natural muscle-based actions. However, without significant improvements, BCI development will be limited to the most basic communication functions for those with paralysis.

Conclusion

Brain Computer Interface technology has proven to be very useful in medical settings. Allowing recovery and detections of muscular and neurological disorders, BCI technologies have opened new avenues to issues that were thought to be once insurmountable. BCI technologies, like EEGs and fMRIs, permit patients suffering from paralysis to undergo rehabilitation and recover their lost senses, even allowing them to communicate. It has also allowed for detection of various muscular and neurological disorders in the body. Although BCI technology has shown to have many benefits in the medical setting, much more research needs to be done. With the technology still in its experimental stages, experimentation on human patients has yet to be completed. More experimentation needs to be done in order to uncover the long term effects the machinery has on a patient, as well as methods to minimize errors when used. With the foundation of Brain Computer Interfaces set in the past few years, BCI technology – with a few improvements- could prove to be the technology of the future.

Funding Statements/Acknowledgments

Although Atri Surapaneni is listed as the primary author, this paper was a collaborative effort done by Atri Surapaneni, Vishruth Bharath, Ivvone Zhou, Bianca Chan, Shardul Dhongade, Joshua Tsai, and Sahithi Ankireddy. All authors contributed equally to the writing of this paper.
Thank you to Nicole Arra for her advice and support in writing this paper.

References

1. Benjamin, Emilia. “Heart Disease and Stroke Statistics … – Ahajournals.org,” 2019. https://www.ahajournals.org/doi/10.1161/CIR.0000000000000659. Accessed August 25, 2020.
2. R;, Taub E;Uswatte G;Pidikiti. \”Constraint-Induced Movement Therapy: A New Family of Techniques with Broad Application to Physical Rehabilitation–a Clinical Review.\” Journal of Rehabilitation Research and Development. Accessed August 26, 2020. https://pubmed.ncbi.nlm.nih.gov/10659807/.
3. Nasser, Haya El. \”The Graying of America: More Older Adults Than Kids by 2035.\” The United States Census Bureau. October 08, 2019. Accessed August 26, 2020. https://www.census.gov/library/stories/2018/03/graying-america.html.
4. Kelly-Hayes, Margaret. \”Influence of Age and Health Behaviors on Stroke Risk: Lessons from Longitudinal Studies.\” Journal of the American Geriatrics Society. October 2010. Accessed August 26, 2020. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3006180/#:~:text=The risk increases with age,will result in death1.
5. Shih, Jerry J., Dean J. Krusienski, and Jonathan R. Wolpaw. \”Brain-computer Interfaces in Medicine.\” Mayo Clinic Proceedings. March 2012. Accessed August 23, 2020. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3497935/.
6. Haas, L. F. \”Hans Berger (1873-1941), Richard Caton (1842-1926), and Electroencephalography.\” Journal of Neurology, Neurosurgery, and Psychiatry. January 2003. Accessed August 23, 2020. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1738204/.
7. Vidal, J. J. (2020). TOWARD DIRECT BRAIN-COMPUTER COMMUNICATION. Retrieved August 26, 2020, from https://www.annualreviews.org/doi/pdf/10.1146/annurev.bb.02.060173.001105
8. Lurito, J., Petrides, M., Schwartz, A. and Massy, J., 2020. Mental Rotation Of The Neuronal Population Vector. [online] Wexler.free.fr. Available at: <http://wexler.free.fr/library/files/georgopoulos%20(1989)%20mental%20rotation%20of%20the%20neuronal%20population%20vector.pdf> [Accessed 23 August 2020].
9. Soekadar, Surjo R., Niels Birbaumer, Marc W. Slutzky, and Leonardo G. Cohen. \”Brain–machine Interfaces in Neurorehabilitation of Stroke.\” Neurobiology of Disease. December 07, 2014. Accessed August 23, 2020. https://www.sciencedirect.com/science/article/pii/S0969996114003714.
10. McConnell, A., Moioli, R., Brasil, F., Vallejo, M., Corne, D., Vargas, P. and Stokes, A., 2020. Robotic Devices And Brain-Machine Interfaces For Hand Rehabilitation Post-Stroke. [online] Medicaljournals.se. Available at: <https://www.medicaljournals.se/jrm/content/abstract/10.2340/16501977-2229> [Accessed 26 August 2020].
11. Cervera, María A., Surjo R. Soekadar, Junichi Ushiba, José Del R Millán, Meigen Liu, Niels Birbaumer, and Gangadhar Garipelli. \”Brain-computer Interfaces for Post-stroke Motor Rehabilitation: A Meta-analysis.\” Annals of Clinical and Translational Neurology. March 25, 2018. Accessed August 26, 2020. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5945970/.
12. \”Intro to Brain Computer Interface.\” NeurotechEDU. Accessed August 26, 2020. http://learn.neurotechedu.com/introtobci/.
13. Devlin, Hannah. \”What Is Functional Magnetic Resonance Imaging (fMRI)?\” Psych Central. July 30, 2020. Accessed August 26, 2020. https://psychcentral.com/lib/what-is-functional-magnetic-resonance-imaging-fmri/.
14. DL;, Linden DE;Turner. \”Real-time Functional Magnetic Resonance Imaging Neurofeedback in Motor Neurorehabilitation.\” Current Opinion in Neurology. Accessed August 26, 2020. https://pubmed.ncbi.nlm.nih.gov/27213774/.
15. Sitaram, Ranganatha, Andrea Caria, Ralf Veit, Tilman Gaber, Giuseppina Rota, Andrea Kuebler, and Niels Birbaumer. \”FMRI Brain-computer Interface: A Tool for Neuroscientific Research and Treatment.\” Computational Intelligence and Neuroscience. 2007. Accessed August 26, 2020. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2233807/.
16. Radiology, Department Of. \”Functional MRI for Neurofeedback: Feasibility Studyon A… : NeuroReport.\” LWW. Accessed August 26, 2020. https://journals.lww.com/neuroreport/Abstract/2002/08070/Functional_MRI_for_neurofeedback__feasibility.5.aspx.
17. Han, C., Kim, Y., Kim, D.Y. et al. “Electroencephalography-based endogenous brain–computer interface for online communication with a completely locked-in patient.” J NeuroEngineering Rehabil 16, 18 (2019). Accessed August 26, 2020. https://doi.org/10.1186/s12984-019-0493-0
18. Pandarinath, Chethan, Paul Nuyujukian, Christine H. Blabe, Brittany L. Sorice, Jad Saab, Francis R. Willett, Leigh R. Hochberg, Krishna V. Shenoy, and Jaimie M. Henderson. \”High Performance Communication by People with Paralysis Using an Intracortical Brain-computer Interface.\” ELife. February 21, 2017. Accessed August 26, 2020. https://elifesciences.org/articles/18554.
19. Shih, Jerry J., Dean J. Krusienski, and Jonathan R. Wolpaw. \”Brain-computer Interfaces in Medicine.\” Mayo Clinic Proceedings. March 2012. Accessed August 26, 2020. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3497935/.
20. Rashid, Mamunur, Norizam Sulaiman, Anwar P P Abdul Majeed, Rabiu Muazu Musa, Ahmad Fakhri Ab Nasir, Bifta Sama Bari, and Sabira Khatun. \”Current Status, Challenges, and Possible Solutions of EEG-Based Brain-Computer Interface: A Comprehensive Review.\” Frontiers in Neurorobotics. June 03, 2020. Accessed August 26, 2020. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7283463/.

About the Author

Atri Surapaneni is a high school senior from Northern California. He is interested in neuroscience and chemistry. He is part of a company called Ceural Labs and is focusing on the development of Brain Computer Interfaces.