A brain-computer interface (BCI) is a device that directly translates brain signals into actions for computer software, such as a text editor, or a device, such as a TV. Here we explain who the intended users of a BCI are and how a BCI works. We also describe the different types of BCIs with a specific focus on BCIs for individuals who suffer from severe paralysis.
A BCI is a device that consists of sensors that measure brain signals (often in the form of ‘electrodes’), an amplifier to boost the faint brain signals, and a computer that translates the signals into commands to control computer programs and/or devices. The components of BCIs can be made portable and/or wearable. BCI controlled devices span the range from assistive technology for people with paralysis, to internet devices (such as a smart phone) for healthy people and simple computer games or toys (such as a small helicopter).
There are currently many groups busy developing BCIs for a variety of applications and types of users. Many of the applications are intended for temporary use and require no permanent installation. Rehabilitation devices to help people recover from injuries, and game controllers for healthy people, are examples of this. Some BCIs are intended to permanently replace a function that has been lost or impaired due to injury or disease (such as the loss of hand function due to stroke).
In the case of temporary BCIs such as game controllers, several commercial devices are available. All current gaming BCIs are based on technology that measures brain signals with electrodes built into a cap or headband. While these systems can be fun to play with, the performance of such systems is not yet good enough to make them viable replacements for a physical impairment.
Medical BCI applications require a high degree of accuracy because the consequences of device failure are much more severe. For example frequent mistakes in a BCI device used for typing will severely hamper communication, or the unintended movement of a robot arm or wheelchair can cause serious injury. Many groups are working hard to bring BCIs up to the standards of medical applications. These devices are not only being developed with electrodes placed on the head with a cap or headband, but also with sensors that can be implanted in or on the brain directly. In these cases, the risk of surgery is outweighed by the benefit of better performance relative to what has been achieved so far with electrodes placed on the scalp.
People that stand to benefit the most from a BCI are those suffering from severe paralysis. Injuries located closer to the brain generally lead to higher grades of paralysis and more loss of function. In rare cases the paralysis is so extreme that the afflicted person is able to move almost no part of their body. This condition is known as ‘locked-in syndrome’, or LIS. In this condition it is generally not possible to speak and communication is only possible through subtle facial movements such as eye blinks.
BCIs designed to restore the ability of people with LIS to communicate are currently being developed. The expectation for these devices is that they will at some point provide people with LIS with the ability to communicate with an extensive vocabulary, send messages via the internet, and turn appliances on and off without the assistance of care givers 24 hours a day. Most importantly, the devices can also be used to alert caregivers that the user needs assistance.
BCIs can also be valuable for people suffering from less severe paralysis. For example, BCI devices that allow people who are paralyzed from the neck down to control a robot arm by imagining moving their own arm have successfully been tested. However, it remains unclear how much time is needed to improve these systems to the point that they can be made commercially available and affordable for people who are not suffering from extreme paralysis.
A BCI records and interprets or decodes brain signals. Brain cells (neurons) communicate with each other by sending and receiving very small electrical signals. It is possible to ‘listen’ to these signals (generally referred to as ‘brain activity’) with advanced electrical sensors. Healthy people are able to move because the brain sends signals via the central nervous system to the muscles of the body. All interaction of a person (such as speaking or shaking hands) requires precise communication between the brain and muscles. Medical conditions such as stroke or neuromuscular diseases can disrupt or break the communication between the brain and body muscles and lead to paralysis (or the loss of the ability to control one’s body, such as cerebral palsy). However, in many cases the brain is still able to generate the activity for intended movements and a BCI can use the brain activity to control assistive devices.
Measuring brain signals
Brain signals can be measure with various techniques that each have pros and cons. A commonly used technique (for example used for neurological testing in hospitals) is electroencephalograph (EEG). This technique uses electrical sensors (electrodes) that are places on the scalp.
Electrodes can also be placed under the scalp directly on or in the brain tissue. A surgical procedure is necessary to place such electrodes. Electrodes that are placed on the surface of the brain do not damage the brain. The quality of this signal is significantly better that signals recorded from the scalp. It is for this reason that implantable BCIs are now being developed for paralyzed people.
Other techniques for measuring brain activity are functional MRI (fMRI), which measures brain activity with a MRI-scanner, and magnetoencephalography (MEG), which measures brain activity with an MEG-scanner. Both of these techniques require large and expensive machines that will not become suitable for home use.
An additional technique is near-infrared spectroscopy (NIRS), which measures brain activity by shining near-infrared light through the skull. NIRS can be made portable and does not require any surgery. However, at this time the quality of the brain activity measurement is not sufficient for use with BCIs.
In general each part of the body has its own ‘control center’ in the brain that is responsible for orchestrating it movements. For example making a fist with your left hand and wiggling your right big toe are controlled by distinct areas in your brain.
The different techniques used to measure brain activity can ‘see’ when different control centers are active. This allows BCIs to detect the movement of body parts from the brain activity. A special quality of the brain is that these control centers are also active why you simply think about making a movement without actually moving.
In general people who suffer from LIS still have fully functioning control centers in the brain. Hence they are able to activate distinct areas in their brain by thinking about, or attempting to make, movements even if they are no longer able to move the part of the body that that area of the brain normally controls.
In addition to movements a number of other brain functions can be detected. For example, there is a small area in the brain that is activated when you do a numeric calculation in your head. Other areas are involved in different aspects of understanding language and speaking. When these areas are active a BCI can detect if a person is adding in their head, or is talking.
Hence, there are many distinct areas in the brain that a person can intentionally turn on and off by performing different mental (for example; by counting backwards in steps of 7 in their head) or physical tasks. The fact that a person suffering from paralysis can also intentionally activate specific areas of their brain by performing mental tasks, makes a BCI a realistic and promising assistive device technology.
What can a BCI do with the brain signals
By placing the electrodes exactly on brain areas that someone can control, we obtain signals that respond to that control. When a signal is detected, it can be converted to a command to operate a device or software. A computer can then be programmed to use this information to perform specific tasks. In this way a person can us a BCI to make a computer mouse ‘click’ every time they count backwards in their head and select from a menu in a computer program, such as an email program. This type of ‘mouse-click’ based control is already commonly used by people suffering from paralysis with special ‘buttons’ that can be activated by whatever type of movement they are still able to make, such as lifting there eyebrows. Thus, a BCI can also be used as a ‘button’ to control the many types of devices designed for button press control.
Various BCIs are commercially available. They are not good enough for use by paralyzed people because they don’t work well enough. For instance if they would be used to type an email, every 5th letter would pretty much be wrong, causing a lot of frustration very quickly. There are some implants but they are limited to the electrodes, and do not yet include the amplifier. They rely on external amplifiers and thus come with a connector fixed on top of the head. One system is fully implantable (neuralsignals.com) but it is not yet clear how well it will work. There is, however, one fully implantable system that is now being tested in Utrecht.
The UMC Utrecht Braincenter has developed an implantable BCI system for people suffering from LIS, based on an advanced device made by the medical device company Medtronic. The Utrecht Neuro-Prosthesis (UNP) is the first and only system that is suitable for unassisted home use. The UNP sends amplified brain signals from a device implanted in the chest via a wireless connection to an external computer. The UNP is compatible with existing available assistive technology.
Multiple labs throughout the world are currently developing additional implantable BCIs. The future will tell which of these devices is the most suitable for which users. At this point it is exciting news that the first test of a fully implantable BCI for home use is underway. Positive results from these test can boost further development of BCIs, which in the future promises to make high performance affordable implantable BCIs available to people suffering from lesser degrees of paralysis.
It is important to realize that a BCI is not able to read a person’s thought or inner speech. BCIs are only able to detect when the specific brain area that they are focused on is active. The current technology is far from being able to detect or interpret the activity of different types of complex brain functions within a specific area. For example a BCI can detect that a person wants to make a mouse click, but it cannot determine the reason for the mouse click based on brain activity alone. However, in the future BCI may be able to issue several different commands at the same time, such as to move the mouse up and to the right.