Brain-computer interface

A brain-computer interface (BCI) relies on direct measures of brain activity, provides feedback to the user, is processed in real time, and relies on intentional control. BCIs measure central nervous system (CNS) activity, converting it into artificial output in order to replace, restore, enhance, supplement, or improve natural CNS output and changing the ongoing interactions between the CNS and the external and internal environment. BCI systems have applications in neurorehabilitation, assistive device technology, cognitive enhancement, and human-to-computer communication. BCIs are used for communication or control of external prosthetic devices in people living with conditions such as spinal cord injury, amyotrophic lateral sclerosis (ALS), locked-in syndrome (LIS), and multiple sclerosis (MS). BCIs can be used for functional electrical stimulation of muscles in a paralyzed person or of peripheral nerves to restore bladder function. BCIs can monitor brain activity during prolonged demanding tasks and detect lapses of attention and alert the person. BCIs are used in research to study CNS function.

Invasive electrodes entail risk due to surgery and show gradual degradation of recorded signals.

• Intracranial electroencephalography, also known as electrocorticography (ECoG)
• Single-neuron action potentials
• Multi-unit activity (MUA)
• Local field potentials (LFPs)

• Electroencephalography (EEG)—Electrodes on the scalp measure electrical brain activity due to the flow of electric currents during synaptic excitations of neural dendrites
• Functional magnetic resonance imaging (fMRI)
• Magnetoencephalography (MEG)
• Near-infrared spectroscopy (NIRS)
• Electromyography

• Neural dust—Electronic sensors placed into the cortex that are interrogated remotely using ultrasound and also powered by ultrasound
• Chronically implanted neural electrodes
• Neural electrodes both stimulate and measure nerve signals
• Microelectrodes, microelectrode arrays (MEA)
• Penetrating electrodes and non-penetrating electrodes
• Microelectromechanical systems (MEMS)
• Sharp glass electrodes and patch-clamp electrodes
• Vertical nanowire electrode arrays (VNEAs)

iBCIs use microscale probes with ninety-six fine-tipped microelectrodes in a 4x4 mm array that is inserted into the cortex. Sensors receive all-or-none output of single neurons, known as the action potential, as well as summed voltage fluctuations from small and large numbers of neurons, called field potentials. These include and wireless implants transcutaneous systems.

Motor imagery is the imagining of a movement rather than actual execution of movement. Motor imagery activates areas of the brain that are responsible for generating actual movement. Motor imagery techniques do not use external signals and are called endogenous BCIs. Steady-state visual evoked potential (SSVEP) and P300 BCIs are exogenous BCI paradigms that use external stimuli such as flickering LEDs or auditory beeps to evoke discriminative brain patterns.

Within motor imagery, sensorimotor rhythms (SMR) have been used by patients with tetraplegia, spinal cord injury, and amyotrophic lateral sclerosis (ALS). In SMR imagination of kinesthetic movements of large body parts such as hands, feet, and tongue can result in modulation of brain activity, such as event-related desynchronization (ERD) in mu (8-12 Hz) and beta rhythms (18-26 Hz), and relaxation results in event-related synchronization (ERS). EEG electrodes above the sensorimotor cortex record ERD and ERS modulations and can be used to control prosthetic devices and move the cursor on a computer. Training using the SMR paradigm can last up to several weeks. SMR can distinguish motor activities corresponding to large body parts, but the decoded motor information does not include magnitude or direction kinematics parameters such as position, velocity, or acceleration.

Imagined body kinematics (IBK), sometimes referred to as natural imaginary movement, is a paradigm independent from SMR because of the different training and analysis protocols. IBK originated from invasive BCI technology but is extracted from low-frequency SMR signals. Subjects are asked to imagine the continuous movement of only one body part in multi-dimensional space and signals are decoded in the time domain. IBK requires less training time compared with SMR.

P300-based BCIs are communication tools in which thought is used to input texts or commands, control smart home controls, and for brain painting. The P300 is the one of the most studied event-related potentials (ERP). EEG signals of a specific event type are averaged to derive an ERP. In ERP P300 is a positive deflection with a time delay of about 250 ms to 750 ms after the onset of a rare and unexpected stimulus. P300 speller devices use a matrix of letters, numbers, and symbols that flash in sequence and the subject is required to focus their attention on the intended character, which is determined by the speller, based on its row and column. Visual P300 BCI can be used by most subjects easily with high accuracy; disadvantages include fatigue from the high level of attention and visual focus issues and inability of use for people with visual impairment.

SSVEP is also referred to as photic driving, since responses are generated in the visual cortex. SSVEP requires highly accurate eye control. In response to flickering stimuli, subjects shift their gaze and their attention. An EEG pattern is formed that is consistent with the flickering frequency of the stimulus on the central retina. Multiple flickering targets with distinct flickering frequencies are presented to the user and the intended target is determined by matching the pattern of EEG activity to the command associated with the particular frequency.

Hybrid BCI platforms combine EEG with one or more physiological measures, such as heart rate by ECG, eye movement with electrooculography (EOG), or a hemodynamic signal recorded by functional near-infrared spectroscopy (fNIRS).

A covert attention paradigm has a subject focus on a centrally located fixation point while following another point such as a cursor without overt eye movement, whereas overt attention requires the subject to use overt eye movements. In both cases, EEG signals are typically recorded from the posterior cortex.

For discrete movement intention, EEG signals are collected before the onset of movement, even in subjects that are not able to physically execute an actual movement.

Auditory exogenous stimulation can be used to evoke auditory steady-state responses (ASSR). Rapid auditory stimuli has been shown to record ASSR maximum amplitude from the vertex of the scalp.

Under the somatosensory paradigm, vibrotactile sensors are placed on the body and stimulated at different frequencies, producing EEG signals. Somatosensory-based BCI systems have been used to assist patients with locked-in syndrome.

Preprocessing is needed to enhance the signal-to-noise ratio and remove artifacts such as the part of EEG signals that come from muscular activity of the head and eye that are unrelated to the brain.

• Covariance Matrix Adaptation Evolution Strategy (CMA-ES)

After brain signals are preprocessed, they are fed into one or more feature extraction algorithms that extract features in the time domain and frequency domain that encode messages or commands.

• Threshold crossings

• Multiunit activity

• Local field potentials
• Amplitude measures
• Band power
• Hjorth parameters
• Autoregressive models
• Wavelets
• Spatial filters
• Short Term Fourier Transform (STFT)
• Auto Regressive Model (AR)
• Wavelet Transform (WT)
• Common Spatial Pattern (CSP)

In BCIs, classification is the translation of features provided by the feature extractor to a category of brain patterns using classification algorithms.

• Linear Discriminant Analysis (LDA)
• Support Vector Machine (SVM)
• Non-linear methods such as neural networks

Over the last five decades, progress in neural recording techniques has allowed the number of simultaneously recorded neurons to double approximately every seven years, mimicking Moore's law. Such exponential growth motivates us to ask how data analysis techniques are affected by progressively larger numbers of recorded neurons.

Number of simultaneously recorded neurons to double approximately every 7 years, mimicking Moore’s law.

For training BCI users, virtual reality is a method of providing feedback. In gaming for the purposes of entertainment, virtual reality headsets used for gaming could include EEG sensors to potentially allow games to respond differently to the user depending on their mood or how they respond to particular elements of the game.

Optogenetics is a pre-clinical neuroscience research tool that has been suggested as an approach for neuroprosthetics and the treatment of brain disorders. Optogenetics can be used for real-time control of genetically engineered brain neurons. Photosensitive proteins open and close membrane channels via light-inducible activation or suppression.

Optogenetics previously utilized optical fibers inserted into the skull, but wireless optogenetics technologies are being developed.Optogenetic approaches could potentially use red or near-infrared light, which has high tissue penetration, delivered by light-emitting diodes (LED).

• RetroSense Therapeutics (acquired by Allergan in 2016) was sponsored by Retina Foundation of the Southwest to perform a clinical trial on human patients with retinitis pigmentosa. Channelrhodopsin-2 was delivered to retina cells in an AAV vector.

• Wireless optogenetic nanoscale device
• Deep brain electrical stimulation for Parkinson's disease (Circuit Therapeutics)
• Science Eye is an optogenetic visual prosthesis for patients with blindness due to loss of light-sensitive photoreceptors (retinitis pigmentosa and dry age-related macular degeneration) under development by the company Science. Optogenetic gene therapy is targeted at optic nerve cells to make them light-sensitive at specific wavelengths. A microLED display panel is inserted over the retina that allows fine control of the optic nerve ganglion cells, which stimulate the brain.

• BrainGate Neural Interface, developed by Cyberkinetics with the Department of Neuroscience at Brown University developed for people with paralysis. Intellectual property is owned by the BrainGate company.
• BrainGate2, successor to BrainGate, is smaller than a contact lens and is surgically implanted into the areas of a disabled user’s motor cortex. Research on BrainGate2 is by the BrainGate Research Team, which is funded from federal and philanthropic sources and separate from BrainGate the company.
• BrainNet
• Cochlear implants
• Deep brain stimulators
• Enten are headphones capable of BCI, developed by Neurable that use cloth EEG sensors.
• IpsiHand Upper Extremity Rehabilitation System (IpsiHand System), by Neurolutions, is an EEG-based Brain-Computer-Interface (BCI) device that is FDA-approved for assisting in rehabilitation for stroke patients with upper extremity (hand, wrist, and arm) disability.
• Kernel Flow is a wearable headset (Kernel) that measures brain activity by changes in blood oxygenation, collecting functional near-infrared spectroscopy (fNIRS) data and EEG data. Kernel Flow is available to neuroscientists and was used in a study on the psychedelic effects of ketamine on the brain. The device is expected to be available as a consumer product in 2024.
• PiEEG is a shield that allows Raspberry Pi to measure EEG signals and function as a BCI. The device is available at low-cost with open-source instructions intended to make neuroscience more accessible.
• Responsive neurostimulation (RNS) utilizes intracranial electroencephalography (EEG) to detect seizures and delivers stimulation to cortical and subcortical brain structures for seizure control
• Speech neuroprosthesis—Researchers at Stanford University developed a speech-to-text BCI that records spiking activity associated with attempted speaking movements using intracortical microelectrode arrays.
• Stentrode is a BMI device developed by the researchers in Melbourne, Australia, who cofounded Synchron. Stentrode is placed in the superior sagittal sinus (blood vessel), next to the motor cortex, through a minimally invasive procedure where it is inserted through a small incision in the neck and fed through the jugular vein. The device receives signals and also delivers currents back to targeted areas of the brain in the form of "focal brain stimulation." Stentrode has applications in Parkinson’s symptoms, paralysis, and as a potential treatment for epilepsy.
• Visual cortical prosthesis, intended to restore visual function using electronic circuitry and electrical impulses

• Argus II (Second Sight)
• IRIS 2 (Pixium)
• Alpha-AMS (Retina Implant)

• BNCI Horizon 2020 is a Coordination and Support Action funded within the European Commission’s Framework Programme 7 that aims to foster collaboration and communication among stakeholders such as research groups, companies, end users, policymakers, and the general public.

• BrainGate research consortium includes researchers from Stanford, Brown, and Case Western Reserve University, investigating BrainGate2

The initiative is supported by several federal agencies, technology firms, academic institutions, and scientists.

• Hand Proprioception and Touch Interfaces (HAPTIX), neural-interface, sense of touch for amputees

• Neural Engineering System Design (NESD), implantable neural interface, bio-electronics

• Neuro Function, Activity, Structure and Technology (Neuro-FAST)

• Next-Generation Nonsurgical Neurotechnology (N3), nonsurgical neural interfaces

• Reliable Neural-Interface Technology (RE-NET)

• Restoring Active Memory (RAM), implantable neural-interface memory prosthesis/memory aid

• Restoring Active Memory – Replay (RAM Replay)

• Revolutionizing Prosthetics program

• Systems-Based Neurotechnology for Emerging Therapies (SUBNETS), closed-loop diagnostic and therapeutic systems for neuropsychological illnesses

Neuroethics are the examination of right and wrong or good and bad aspects of the treatment of, the perfection of, and the unwelcome invasion of and worrisome manipulation of the human brain. Neurolaw refers to the collaboration between neuropsychologists and lawyers. The International Neuroethics Society (INS) is the largest society committed to studying the social, legal, ethical, and policy implications of the advances in neuroscience.

Neuroethical and neurolegal enquiry that concerns ethical-legal challenges in neuroscience and neurotechnology became known as neurorights. Neurorights are defined as the ethical, legal, social, or natural principles of freedom or entitlement related to a person’s cerebral and mental domain. Neurorights came out of the idea of cognitive liberty, the right, and freedom to control one’s own consciousness and electrochemical thought process. The Neurorights Initiative at Columbia University was the first institutional think-tank on neurorights. The Neurorights Network is an international network of scholars working on neurorights.

Neurorights deals with choosing between what is best for individuals and what is best for society. For example, should individuals in the military have neuroenhancing devices. Such devices could help them better serve their country and protect themselves in the line of duty but could compromise their individual identity and privacy. Data protection law, health law, consumer law, and criminal law may be needed to establish legislation.

Chile classified mental data and brain activity as a human right to be legally protected in the world's first neurorights law, which was passed in 2021. The BCI Pioneers is a US-based patient and patient advocate group that works to ensure the conversation around neuroethics are led by patients. The Neurorights Foundation, based in New York, proposed a “technocratic oath” modelled on the Hippocratic oath taken by medical doctors.

A further “neuro-protection” bill is under consideration by Chile’s congress mandates that all neurotech devices, including those intended for consumer wellness or entertainment, are subject to the same regulations as medical devices. The bill states that neural data should be considered equivalent to a human organ, with the same prohibitions against buying or selling brain data. Some critics argue that brain data is not clearly defined, and a broad definition may include behavioral data that is already collected.

• BrainGate Co.—the BrainGate Neural Interface system

• CTRL-labs—an electromyography-based armband that reads nerve signal intentions to move

• Neuralink—implantable brain-computer interface (BCI) called the N1 sensor
• Kernel—hardware and software for implantable devices for people with neurological and degenerative diseases like epilepsy, dementia, and Alzheimer’s disease
• Facebook—a skullcap to allow users to mentally type
• Petal—software and APIs to collect, measure, and analyze EEG data
• Metabrain—neural interfaces to speed up communication between humans and computers
• Truust Neuroimaging—neuroimaging, data processing, and analysis
• Paradromics—implantable chip/neural interface for brains and computers to exchange data, a communication device for people with paralysis
• Neuroloom—living electrodes made with nerve cells grown in microscopic needles interface with the retina to improve the ability to stimulate the retina and restore vision
• Blackrock Microsystems LLC—tools for neural engineering and neuroprosthetics
• Natus Medical—neuromonotoring products
• Emotiv—EEG brain monitoring
• NeuroSky—EEG technology
• Nihon Kohden—EEG technology
• Compumedics Limited—neurophysiology, cardiology, sleep disorder technology
• g.tec medical engineering—invasive and non-invasive BCIs
• Brain Products GmbH—neurophysiological hardware and software
• Advanced Brain Monitoring—B-Alert wireless EEG
• BrainCo, Inc.—cognitive training and prosthetics
• MindMaze SA—rehabilitation, virtual reality medical products to neural recovery
• Neuroelectrics—EEG-based brain monitoring, brain stimulation, home therapy research
• Synchron, Inc.—implantable neural interface, assistive technology for paralysis and neurological conditions
• NextMind—EEG-based BCI for mass market
• BIOS—BCI linked to AI to discover neural biomarkers and use AI-based neural treatments to treat chronic conditions
• NeuroPace—implantable BCI to treat neurological disorders, responsive neurostimulator (RNS) system for epilepsy
• Cadwell Industries, Inc.—neurophysiology devices
• PlatoScience—tDCS for boosting cognition
• Neurosphere—neurofeedback enhanced meditation for enhanced well-being and productivity
• RxFunction—leg sensory neuroprosthesis for neuropathy and balance problems
• ANT Neuro—recording and analysis for EEG, EMG, MRI, TMS, and MEG technology
• Neurable—neurotechnology hardware and software, EEG-based, interpretation of human intent, control of devices, VR games
• Brainmaster—neurofeedback
• Neuroptimal—neurofeedback

The BrainGate Interface System involves implanting a “Utah” microelectrode array (manufactured by Blackrock Neurotech), a ninety-six electrode array on a 4 x 4 platform, onto the cortical surface of the brain. Fourteen people were enrolled in the BrainGate clinical trial between 2004 and 2021. Collectively, there were 12,203 days of safety experience in which 68 device-related adverse events occurred. Six of the adverse events for categorized as serious. No intracranial or deep tissue infections were reported due to the implant. Two participants experienced postoperative seizures four days after implantation, which led to a modification of the trial protocol to include postoperative prophylactic antiseizure medication. One participant experienced new-onset refractory status epilepticus (NORSE). The most common adverse event was local skin reaction or sensitivity around the percutaneous pedestal.