Brain Computer Interfaces for neurological patients — tools to support rehabilitation and communication

Imagine someone unable to move or to speak — someone who can’t even blink in response to a basic question. Some neurological patients with severe brain lesions or rare forms of stroke may still experience the world around them, but have lost any ability to interact with it. They are literally completely locked-in.

As a graduate student in medicine and neuroscience, I am interested in using brain imaging — specifically neurofeedback — to treat and to support patients. In this post, we’re going to learn about techniques called Brain-Computer interfaces (BCIs) that analyze brain signals in real-time with the aim of supporting patients by allowing them to interact with their environment, communicate, and regain lost capacities. These new methods have the potential to unlock those locked-in patients, while also improving their quality of life.

We will look at three examples for BCI applications, which 1) could enable locked-in patients to communicate after years of silence, 2) help Parkinson’s patients to restore and maintain brain function using neurofeedback training and 3) allow patients who have lost a limb to use a BCI prosthesis to learn to control a robotic limb.

Deciphering imagined thoughts — a new voice for locked-in patients?

“One cannot not communicate” is an axiom in communication theory formulated by the Psychologist Paul Watzlawick. One rare exception are patients who suffer from absolute immobility in locked-in syndrome. Restoring some means of communication for these patients has recently inspired a fascinating series of brain computer interface (BCI) case studies (completely) in locked-in patients.

In one study, patients were asked “yes” or “no” questions and had to imagine the answer. A computer algorithm then learned to distinguish between brain signals of an imagined “yes” and “no” response. The groundbreaking study provided insight into the minds of completely locked-in patients for the first time. For instance, one patient had not been capable of communicating reliably for over 5 years.

Work flow of BCI training in locked-in patients. Image credit: Chaudhary et al. PLOS Biology, 2017. 15(1)

This study enabled basic communication with patients using two different imaging techniques: EEG, which is based on electrical signals emitted by a large number of brain cells (so called populations), and functional Near Infrared Spectroscopy (fNIRS), which is based on differences in blood oxygenation similar to fMRI. Both techniques are more affordable and portable than fMRI and thus, could be more effective for translational applications such as helping locked-in patients to communicate. Altogether, BCIs offer hope for new communication channels with locked-in patients and could one day lend some patients a new voice. [edit 2018–18–12: mistakes happen, and the authors have in the meantime published a correction for their article, which does not affect their overall conclusion. Further, are-analysis of the data run by a scientist who was not author on the original study strongly suggests that the analyses computed by Chaudhary et al., 2017 contain a fundamental error, and questioning the evidence for communication in completely locked-in patients. The authors of the original have however not officially responded to-date to this criticism. I will keep you posted.]

Neurofeedback for Parkinson’s patients — a boost for physiotherapy?

Neurofeedback training is designed to help people learn self-regulating brain processes, such as mental imagery. BCI with neurofeedback can be used for rehabilitation. To give you an example: when we learn new motor skills like mastering a tough ski slope, practicing the skill mentally can speed up the learning curve. Some athletes use motor imagery to perfect their art, and more recently, researchers have studied its potential for motor rehabilitation. For instance, in Parkinson’s disease which is a chronic movement disorder. Parkinson’s patients suffer from slowed, rigid and jerky movements due to a progressive loss of dopaminergic neurons. Motor imagery training combined with physical exercise can improve motor symptoms as shown in stroke patients. It is likely that also Parkinson’s patients can benefit from training. And maybe we could further boost these effects by providing patients with real-time feedback based on brain activation?

Feedback thermometer. Background color instructs when to rest (yellow) and when to imagine movements (green). Neural activity is presented as red filling blocks. Image credit: Subramanian et al. Journal of Neuroscience, 2011; 31 (45).

Encouraging results have come from a recent randomized controlled trial of motor imagery-based real-time fMRI neurofeedback (NF-training). The trial compared a patient group that received only physical exercise training with a group that engaged in a combination of some physical exercise and NF-training. Despite more physical training in the first group, only the second group showed substantially reduced motor symptoms. There is hope that one day neurofeedback training could be offered to Parkinson’s disease patients concurrently with physical exercise, however, more trials are needed to assess its effectiveness. The presented trial also shows that neurofeedback, as a BCI technique, extends beyond psychiatric applications.

Imagining movements leads to action — control of a robotic arm

If I told you that you could control an object using only your imagination, you might think this blog is now going too far. What seems beyond reality is possible with the right BCI! A recent groundbreaking study demonstrated how BCIs can truly unleash the power of imagination. When combined with eye movements, patients who have lost their limbs were able to perform basic movements with a robotic arm using the help of real-time analysis of brain signals. These examples are encouraging but larger trials will be needed to assess what is required to make BCI controlled prosthetics a reality for patients.

Patient with amputated arms is moving a robotic arm with his thoughts. The modular robotic prosthesis was developed by the Johns Hopkins University Applied Physics Laboratory (APL). Photo credit to Beckett Mufson.

I hope these three recent applications for BCIs have given you an idea as to what is possible and how this research could make a difference to a patient’s life. There still remains a large amount of work to be done before taking this field beyond prototypes. Initial clinical trials for real-time fMRI neurofeedback training and BCI spelling devices for stroke patients are ongoing, innovative, and promising for how computers interfacing with the brain can improve the quality of human life.

More material:

TED talk — MRI scanning to make you feel better

Review on BCIs for communication and rehabilitation

NYT article and video: Prosthetic Limbs, Controlled by Thought