Current thinking: how does the brain work?
Image credit: Science Photo Library
The workings of the human brain are still largely a mystery. Electrodes can offer an insight, but some fear sinister intent.
In the 1983 film ‘Brainstorm’, Christopher Walken and Natalie Wood develop a brain-computer interface (BCI) to read and process thoughts, which are then transferred to video tape to be shared with other people. This was used to share happy memories to reconcile the estranged pair, but also to experience a fellow scientist’s heart attack, and it was the target of the military interest for torture and brainwashing.
The idea of being able to read a person’s thoughts and exert control over their thoughts and actions has been a rich seam of material for science fiction and the medical profession alike. Today, there are some insights into the brain, but they are limited. Functional magnetic resonance imaging (fMRI) scans indicate brain activity by measuring the flow of blood to the brain, not what neurons in the brain are doing.
Electrodes placed on the scalp respond to a visual stimulus, an event-related stimulus, or a combination of the two.
Gel on the electrode pads enhances the electrical signal. Typically, the scalp is shaved for closer contact, but a team at Tsinghua University, Beijing, has developed electrodes that operate without gel and which function through a full head of hair. Each electroencephalogram (EEG) electrode is a polyvinyl butyral (PVB)/melamine sponge, embedded with conductive silver nanowires. The nanowires are self-locking, so a skull cap of 10 electrodes is mechanically stable and has a high conductivity of 917 S/m (siemens per metre), which remained unchanged after 10,000 cycles at 10 per cent compression. It was found to detect brain activity as accurately as conventional electrodes, with the silver nanowires creating surface metallisation to produce a high conductivity without adding weight to the cap. Accuracy for steady-state visual evoked potential (SSVEP) measurements was 86 per cent, which is comparable to 88 per cent accuracy for conventional electrodes and conductive gel.
A volunteer wearing the skull cap was able to control a toy car left, right, backwards or forwards by shifting her gaze to an icon on a computer screen. As well as operating through hair, the electrodes were not adversely affected by heat or sweat.
An electrode placed closer to the neurons produces a better-quality signal but this requires an invasive procedure where electrodes are placed on the surface of the brain in electrocorticography (ECoG) monitoring, to record electrical activity from the cerebral cortex. A 2012 study by researchers at the Epilepsy Monitoring Unit at the Albany Medical Center in Albany, New York and the Wadsworth Center of the New York State Department of Health in Albany, New York, found that this provides signals that have a high signal-to-noise ratio, are less susceptible to artefacts, and have a high spatial and temporal resolution. It can be used to map the brain to locate the focus points of epileptic seizures for treatment. The study was also able to identify functional connectivity and finer, task-related spatial temporal actions for the brain-computer interface to decode the patient’s intentions and improve communication, motor execution and planning, auditory processing and visual-spatial attention.
Scientists are monitoring brain activity that controls the brain’s speech regions to decode spoken words and phrases in real time. This work could eventually make it possible for people with speech loss to communicate via brain signals.
Researchers at Harvard University are developing a flexible mesh of electrodes that could be implanted in the brain for targeted relief of symptoms of illness caused by neurodegeneration or ageing.
In Parkinson’s disease, for example, if drugs are not effective in controlling tremors, rigid implants may be used for deep brain stimulation to ease symptoms, but these large implants may impact other areas of the brain and the immune system may be disabled or impaired by immune cells overwhelming and rejecting the implant.
Harvard Professor Charles Lieber has developed flexible mesh electronics that move with the brain and can be implanted to gather data on how individual neurons communicate to more precisely map neural areas and networks of the brain. Following on from that work, Shaun Patel, faculty member at the Harvard Medical School and Massachusetts General Hospital, saw the possibility of using the electrode mesh for control of prosthetic or paralysed limbs. “They could act like neural substitutes, replacing damaged circuitry to re-establish broken communication networks and recalibrate based on live feedback,” he says.
If precise interaction and feedback is achieved, “you could really communicate with the brain in the same way that the brain is communicating within itself”, adds Lieber.
The project’s challenges are scaling up the number of electrodes, processing all the data and returning it for live recalibration to achieve the desired response.
At the University of California, San Francisco’s Epilepsy Center, volunteers preparing for neurosurgery agreed that the ECoG electrodes placed on the surface of their brain to map the origins of seizures in preparation for neurosurgery could also be used to monitor brain activity as the volunteer replied to a set of standard questions.
Using this data, researchers developed a set of machine-learning algorithms with phonological speech models to decode specific speech sounds from the brain activity. The algorithms identified which of the standard responses the volunteer was giving with 61 per cent accuracy. Context of speech improved the algorithm’s accuracy in identifying the question.
Research continues with a goal to reach typing speeds of 100 words per minute using brain activity and to extend the vocabulary to 1,000 words and reduce the error rates for the algorithm to below 17 per cent.
The team, led by Dr Edward Chang, has launched a project called BCI Restoration of Arm and Voice (BRAVO) to investigate if ECoG neural implants can be used to restore movement and communication in patients who suffer paralysis as a result of a brain injury, brainstem stroke or neurodegenerative disease. Those who are unable to speak may be able to use this real-time speech-decoding technique to train an implanted speech prosthesis or battery-operated electro larynx.
Dr Chang’s team’s study was part of Project Steno, part-funded by Facebook Reality Labs, a research division of the technology giant specialising in augmented and virtual reality. Facebook also provided researchers and engineering support for the project.
At its F8 developers’ conference in 2017, the social media company announced its BCI programme and its intention to create a “non-invasive, wearable device that lets people type by simply imagining themselves talking”. The event was plunged into controversy when it was widely reported that the company intended to develop a headset that could read people’s thoughts. Facebook clarified the intent as “imagining the words they want to say – a technology that could one day be a powerful input for all-day wearable AR glasses”.
Graphene listens in
Scientists are investigating how graphene-based transistors can be used to measure low-frequency neural signals. The implants record electrical activity at low frequencies and amplify the brain’s signals before transmitting them to a receiver.
At the Mobile World Congress 2019, Graphene Flagship partners the Microelectronics Institute of Barcelona, the Catalan Institute of Nanoscience and Nanotechnology (ICN2) and the Institute of Photonic Sciences showed a graphene-based implant that covered a large area of the brain and was able to record brain activity at frequencies below 0.1Hz, much lower than previously possible.
The graphene architecture is able to support more recording sites than a standard electrode array and can cover large areas of the cortex without rejecting the implant or interfering with brain activity when mapping low-frequency brain activity to assess the onset of events such as epileptic seizures or to monitor the progress of a stroke. By being able to ‘listen in’ to the brain at this level, the researchers hope that the technology can be used to help neurologists precisely map interactions for an insight into where and how seizures begin and progress for diagnosis and treatment.
ICN2’s José Antonio Garrido compared the technology with existing electrodes: “Our active graphene-based transistor technology will boost the implementation of novel multiplexing strategies that can increase dramatically the development of a new generation of brain-computer interfaces.”
The final phase of Project Steno is a year-long study to assess if it is possible to restore a patient’s ability to communicate using brain activity.
In Project Steno, the electrodes were placed on the surface of the brain and did not penetrate its tissue. Researchers at Brown University, Rhode Island, USA, used Blackrock Microsystems’ Utah Array microelectrode to develop BrainGate. This is a brain implant that allows paralysed patients to move robotic arms. It is an array of 128 needles that gathers stable neural recordings of potential actions in the motor cortex to transmit movement data. It has up to 256 electrodes and neural recordings are accessible immediately after implantation. The arrays are implanted at high speed by a pneumatically actuated inserter for minimal tissue damage. The speed and implantation depth can be adjusted for different electrode configurations and tissue preparations.
Meanwhile Neuralink, a company founded by prominent entrepreneur Elon Musk, is working on a project to implant thin, flexible threads into the brain to transfer data. The threads are intended to reduce damage; as the brain shifts in the skull, the polymer fibres should accommodate the movement, whereas needles on an array are not mobile.
The threads are 4-6µm wide and an array of 96 threads could carry over 3,000 electrodes. Fibres will be automatically inserted by a neurosurgical robot using a needle, like a sewing machine, to insert the fibres into the soft tissue of the brain. The robot adjusts the position of the needle to avoid blood vessels.
Neuralink is still in its infancy (it was launched in July 2019) and is building on work that has gone before. At its launch, Musk said Neuralink plans to design a chip to amplify and wirelessly transmit the brain’s signals to devices.
In a relatively short period of time, electrodes have moved from the surface of the head, to the surface of the brain, to being implanted mechanically into specific regions of the brain. The accumulation of this research, advances in materials and signal technology brings hope but also some concern – if knowledge is power, could the nightmare of Brainstorm await us?
Non-invasive interface to the brain
A wearable device developed by French start-up company NextMind that can sense and translate signals from the user’s visual cortex into digital commands in real-time to operate devices, won innovation awards at CES 2020 for virtual- and augmented-reality and wearables.
The non-invasive BCI is a lightweight device (approximately 60g) with eight electrode ‘combs’ for direct contact with the scalp. Each electrode is fitted with a microchip to process information. The complete device fits into the back of a cap or headband.
Looking at an object induces a response in the visual cortex and amplifies neuron activity. It captures this neuron activity and machine-learning algorithms create a relationship between the object and the user’s intent. Signals are then output as digital commands to control, for example, a TV, changing channels by focusing on icons on the screen. It can also be used to control a computer, an internet-connected device or an augmented-reality/virtual-reality headset.
Next, it is hoped that the wearer will be able to imagine changing channel, for example, in order for the machine-learning algorithm to transmit this signal as a command. Further ahead still, the user could imagine an image for it to be retrieved from the visual cortex’s memory and displayed on a screen – echoes of ‘Brainstorm’, indeed.
Development kits for the wearable electrode will begin shipping in Q2 2020.
Sign up to the E&T News e-mail to get great stories like this delivered to your inbox every day.