A Brain-Computer Interface (BCI) is a system of direct connection between a brain and a computer. It allows an individual to perform tasks without the action of peripheral nerves and muscles. This type of device allows to control a computer, a prosthesis or any other automated system by thinking, without having to use your arms, hands or legs. The concept was first introduced in 1973 and the first human trials date back to the mid-1990s.
Understand the Brain-Machine Interface
Under development in various laboratories around the world, brain-machine interfaces are devices that should enable people with major disabilities to regain a certain degree of autonomy. Tetraplegic individuals or persons suffering from locked-in syndrome (the patient thinks, but only the eyelids can move) could for example control an exoskeleton thanks to the thought of moving, amputees could control the movements of their prosthesis by only thinking about it. Also, people who have lost their speech could speak via a computer, always thanks to thinking. Potential applications are numerous, including for healthy people with for example the field of video games.
How does it work?
The structure of an BCI includes a system for acquiring and processing brain signals, a system for classifying and translating these signals into commands: writing on screen, wheelchair or prosthesis movement, etc.
In concrete terms, the user focuses his attention on an external stimulation of his choice or imagines performing a movement. This generates characteristic brain activity that can be measured using sensors. These signals are transmitted to a computer which analyses them to extract the useful data, then transforms them into a control system for the machine (prosthesis, exoskeleton, wheelchair, software interface, artificial voice…).
These systems usually operate in a closed loop (or feedback loop), allowing the user to progress in mastering the BCI. The user observes the result of his brain control, then adapts his thinking, gradually refining the precision of the action produced by the system. The research also draws on automatic learning algorithms to make the machine adaptive and capable of refining its interpretation of the user’s brain activities over time.
Record electrical signals
The first step necessary for the functioning of a BCI is to record brain activity. Most often, electrodes are placed on the skull, on the cortex or in the brain to record the electrical signals emitted by neurons during a thought.
There are three recording modes:
- Invasive: A grid of electrodes is implanted in the cortex. It records the signals of a population of neurons with very high spatial accuracy.
- Semi-invasive: An electrode grid is placed under the dura mater, the membrane that surrounds the brain just below the skull. The spatial resolution is slightly lower than with cortical implantation, but the risk of complications is lower and medical applications are quickly possible.
- Non-invasive: The patient wears a tissue helmet equipped with multiple electrodes to measure the electroencephalogram. The spatial resolution is limited, and the recording time is only slightly longer than one day.
The choice of recording mode depends on the purpose and applications. In all cases, the electrodes can be removed in case of a problem.
Software interprets the signals
The electrodes used for recording are connected to external software, which classifies, analyses and interprets the brain signals. After, it will output them as commands that are executed by the controlled machine.
Depending on the task to be performed with the BCI, the recorded brain signals are numerous and deep, and difficult to process. Several dimensions are considered in the analysis: duration of the signals, their frequency and their spatial distribution. Pre-processing and filtering are used to remove background noise from the recorded signals. The signal characterizing the intention is then extracted, and its components are classified to keep only the useful information.
Many research teams are working on the development of Brain-Computer Interfaces for the manipulation of exoskeletons, support devices for people who are totally paralysed, to allow them to stand up, move and perform different movements. But many other applications are also being considered: controlling a wheelchair or limb prosthesis, making people talk or writing a computer. In the United States, researchers have already succeeded in obtaining remote control of a robotic arm by tetraplegic people.