What is a Brain-Computer Interface (BCI)
A BCI is a neural interface based on electroencephalography (EEG), which is able to make the Central Nervous System (CNS) communicate with an external peripheral through appropriate electrodes. In the biomedical field, this "peripheral" takes on various connotations, from joint prosthetic systems to the remote control of powered wheelchairs and devices capable of processing virtual words and images. There are two types, differentiated by the application methodology i.e., non-invasive and invasive.
The invasive, as can be seen from the name, requires surgery for the installation at the subcutaneous level of a matrix of electrodes of different types than those used in the non-invasive, capable of overcoming other types of problems. The purpose of the latter is not to assist the interaction between the CNS and peripherals, but the actual replacement of some brain functions, in case of damaged or partially inactive areas at the synaptic level. They are mostly used in the USA and less in Europe, for ethical and legal reasons. We, therefore, analyze specifically the functionality of the non-invasive ones as they are "less experimental" than the invasive ones.
To the first category of BCI belong used in Cybathlon, electrodes are placed on a cap of insulating material and come into contact superficially with the subject's epidermis. They are indicated and are mostly used by people suffering from severe or milder quadriplegia, for the control of the above mentioned peripheral devices and for the interaction through eye movement. These kinds of BCI are those most commonly used in European countries, both at the level of "Assistive Technologies" and for research purposes.
How a Brain-Computer Interface works
As previously said, the BCI have as their basic operating principle the detection of the synaptic bioelectric signal, through the use of an EEG whose electrodes are placed superficially on the scalp. The signal they receive is processed and analyzed by an electronic computer, which in turn is responsible for controlling the peripherals connected to it. This detected signal does not carry all the information content useful for its transduction by the computer. Through the graphical analysis of the EEG, it is noted that a strong component of the received impulse is in the form of noise, it derives from the brain. This is due to the fact that the incoming signal is given by cortical neurons, placed perpendicular to the skin surface, while the other neurons somehow "interfere" in the reception of the biosignal. The EEG, therefore, goes to record the results of the electric potentials generated before and after the neuron (pre and post-synaptic potentials).
Among the various types of potential generated, the BCI deals with analyzing in particular ERPs or "Evented-Rated Potentials," i.e., all those brain responses deriving from electrophysiological stimuli, both internal and external, which are the result of the production of this potential by cortical neurons in a group and synchronously. Particularly interesting is the study of a component (of a wave nature) of these ERP called "P300", defined in this way since it occurs 300ms after receiving the stimulus and is responsible for voluntary muscle contraction. For the detection and classification of the P300s within the EEG trace, it is necessary to isolate the component given by the background noise, clean it from disturbances (through appropriate low-pass or high-pass filters so as not to damage the information content that the wave itself carries) and classify it. Downstream, training of the analysis software (Artificial Neural Network) must be conducted, which, for each portion of the track analyzed, must identify whether it is a P300 component or not.
Depending on the peripheral connected to the computer and its use, various types of BCI can be characterized.
The versatility of BCI: how many and which problems can be solved?
As we said previously, the peripherals to which the BCI are connected have the task of characterizing these instruments towards the relative fields of application to which they are addressed. We talked about BCIs connected to EEG, but others are being developed on the basis of other diagnostic systems such as fMRI (functional magnetic resonance imaging), NIRS (near-infrared spectroscopy), and MEG (magnetoencephalography). Three macro-categories of BCI can be identified, which vary according to the applicable interest.
BCI for communication
When the patient intends to communicate, he focuses his attention on an alphanumeric character contained in a six × six matrix whose rows and columns light up randomly about 5-8 times per second. A row and a column uniquely identify the desired character, in correspondence with which the P300 will appear in the EEG trace. The computer averages the answers obtained, and subsequently, the software, through classifiers, determines the desired character. Although it does not require almost any type of training by the patient, it is effective only if combined with visual stimuli and requires a considerable degree of attention, even prolonged over time, which can significantly decrease the accuracy of the application.
One of the major applications of this interface is still used today to allow communication with patients with ALS and has enjoyed considerable success with numerous disabled patients. It should be mentioned that in 2007 there was a rare case of communication with patients in the quadriplegic state through a BCI system based on the NIRS system. Patients were questioned with simple questions that included a "Yes / No" answer, associated respectively with an increase/decrease in blood oxygenation; some of these patients appear to have been able to break the 70% threshold for correct answers. The widespread use of BCI applications for communication allowed the appearance of the first products accessible to all. Future developments include implementations of features such as email management, social networks, and internet browsing.
BCI for motor control of peripherals
This typology has been conceived and developed with the aim of being able to control peripheral devices suitable for replacing or helping the subject in movements that he is no longer able to perform, for example, through a robotic arm.
Interesting results were achieved in 2004 by the IDIAP research center in Switzerland, which demonstrated the possibility of controlling the movements of a small robot through EEG signal acquisition; this opens up the use (as already happens in some countries) of small automata for domestic use to carry out the simplest actions of daily use. To date, the main intended use remains the control of the movement of an electric wheelchair with EEG acquisition, which requires a period of training carried out by giving directional commands to pointers on the screen.
BCI for motor rehabilitation
In addition to the external peripherals that replace the movement of the patient with motor disabilities, BCI systems have also been exploited in order to control devices capable of helping the person in the movement of their paralyzed limbs. Taking advantage of the cerebral plasticity of the motor cortex, which has great recovery capacities following partial or total paralysis, it was possible for a quadriplegic patient to control (with the aid of a BCI system based on precise brain wave rhythms) the muscles of the hand; they, therefore, proved to be possible devices for rehabilitation. Even the invasive ones were subsequently tested for motor rehabilitation.