In 1999, Jim Jatich, a quadriplegic man who lost movement in his limbs after a diving accident made history in science. He had devoted himself to being the guinea pig of neuroscientists working on movement rehabilitation in paralyzed limbs via brain stimulation. The two decades of experimentation had finally paid off: for the first time, a man was able to partially move his paralyzed hands simply by controlling his thoughts. How exactly did a group of scientists at Case Western Reserve University accomplish this remarkable milestone in the field of neurotechnology? The experiment was quite straightforward: while his brain activity was being recorded, Jatich performed a simple task – to think of specific directions. The direction he chose modulated a component of his brain activity, the beta-rhythm, which in turn controlled the movement of the cursor on the screen, thus providing Jatich with direct feedback and allowing him to learn to control the cursor with different imagined directions. Importantly, the signal from the cursor was subsequently fed into the joystick controlling the movement of his hands via electrodes implanted near the nerves responsible for hand movement. Thus, Jatich was able to control his hand movements – closing and opening the fist – simply by imagining a specific direction in his head.
Almost two decades later, researchers at the same university witnessed yet another breakthrough in the field of neurotechnology. With the help of solely his thoughts and a mobile arm support, Bill Kochevar, a man who following a bike accident became totally paralyzed from the neck down, was able to drink and eat by himself for the first time in eight years. Kochevar simply had to imagine himself performing the intended arm movement, and the sensors implanted in his brain would activate the relevant parts of the motor cortex in order to execute the movement.
These examples showcase the emergence and relevance of an innovative neurotechnology: Brain Computer Interface (BCI)-based neuroprosthetics. According to the U.S. Food and Drug Administration, they are “neuroprostheses that interface with the central or peripheral nervous system to restore lost motor or sensory capabilities”. But to further understand the role and the mechanism of BCI-based neuroprosthetics, it is crucial to understand both of the parts which compose it – naturally, BCIs and neuroprosthetics.
What are “neuroprosthetics”?
A neuroprosthetic is formally defined as a “device that supplants or supplements the input and/or output of the nervous system”. For example,
- cochlear implants are electronic devices used to provide hearing sensations to people with inner ear damage by transforming sound waves into electrical signals received by electrodes implanted in the ear and further sent to the brain;
- visual prostheses are devices which transform, via a camera, visual information into electrical signals used to stimulate the visual cortex through implanted electrodes.
However, the most futuristic and fascinating neuroprosthetics are arguably the ones relying on communication between the brain and a computer. But before moving on to BCI-based neuroprosthetics, let’s define BCIs.
What is a “BCI”?
BCI is a computer system consisting of hardware and software parts that record brain activity, read it and find patterns in the signal in order to control a computer, external devices or the brain itself. Although the most promising use of BCI is in movement rehabilitation in paralyzed individuals, BCI have alternative applications:
- Video games: BCI technology can be used to enable gamers to control their virtual environment by using their thoughts;
- P300 Speller: P300, a component of brain activity which occurs when a person encounters or thinks about a stimulus different from the other ones, is used to create a BCI word speller.
How BCI works
To collect brain activity, a BCI requires either the use of electroencephalography (EEG), a non-invasive technique which records activity via electrodes placed on the scalp, or of electrocorticography (ECoG), a partially invasive technique which uses electrodes placed directly on the brain’s surface. Of course, the collection of brain activity is but one part of the BCI system. The first step is to have the participant perform a mental task which would send a signal to the BCI system – for instance, in neuroprosthetics that replace a paralyzed limb, the individual is often asked to imagine the movement of the limb. After preprocessing the brain signal to get rid of noise and artefacts, the system performs feature extraction: pulling out relevant information from the brain signal in order to manipulate the external device or brain. This extracted information is then fed into a classifier which is a tool for machine learning (a subfield of artificial intelligence which deals with computers teaching themselves new information), used to classify patterns of brain signals into categories of brain states. This process of decoding the recorded activity is also known as multivariate pattern analysis and is according to some, the closest neuroscience has gotten thus far to mind-reading. Exciting, isn’t it? Prior to testing the system, the classifier needs to be appropriately trained in order to properly associate patterns of brain signals with the brain states that they represent. Once the classifier determined what the individual’s intention was, the appropriate signal is sent to the external device, computer or brain to execute the intended action
There are two main types of BCI-based neuroprosthetics: the ones which “close the loop” via brain stimulation, and the ones that do so via a prosthetic limb. The most common method, deep-brain stimulation, consists of implanting electrodes directly on the brain to electrically excite areas that elicit movement, such as the motor or the sensorimotor cortex. Of course, deep-brain stimulation involves surgery and may not be desirable for everyone. Instead, there is a number of non-invasive technologies, which include stimulating specific brain areas by using scalp electrodes to deliver current at low levels (tDCS) or at a specific frequency (tACS), or stimulating brain activity with a magnet (rTMS). Notable cases of brain stimulation include treatment of Parkinson’s disease, epilepsy, auditory hallucinations and potentially depression.
Without doubt, as we have witnessed in Bill Kochevar’s case, BCI-based brain stimulation is fascinating and promising. However, it falls short in certain cases – for instance, when the patient is missing a limb, but also in cases with limb paralysis. Then, during the last step of the process, feedback from the classifier is instead used to control a prosthetic limb. For example, in a typical prosthetic limb-BCI system, the extracted relevant information is the brain signals from the motor cortex involved in imagining the limb movement. These are further decoded by the classifier to determine which action the individual wanted to perform and employed to create the desired movement in the prosthetic limb. One prominent example is the neuroprosthetic limb created by the company BrainGate. In one spectacular case, Cathy Hutchnison, a woman who had been paralyzed for 11 years used BrainGate’s robotic arm and her thoughts to drink coffee from a mug. Thus, neuroprosthetic limbs are enabling impaired individuals to regain some of their independence and hopefully, bring them along some joy and relief.
Closing the loop
So, what does the future hold for bci-based neuroprosthetics? According to researchers, the next step is to enable individuals to feel the prosthetic limb when a movement is completed. Similarly to the way we feel our limbs moving in space when reaching out for an object, in order for the experience to feel as real as possible, it is essential to provide individuals with sensory feedback, and restore the sense of touch. Specialists from different institutions, such as Case Western Reserve University – where it all started – are currently experimenting in amputees to connect electrodes from the prosthetic limb to a cuff surrounding the remaining nerves in the amputated limb, and stimulate them in order to recreate touch sensations. Undoubtedly, the field of neurotechnology still has a long way to go to truly close the loop, but we are definitely on the way towards making more Cathys, Bills and Jims autonomous and happy.