Yesterday, a brain-machine interface, referred to as BMI;or Brain-Computer Interface (BCI) sparked a worldwide frenzy, and humans were implanted with brain-computer interface chips for the first time. This means that the brain-computer interface industry has finally ushered in a major turning point, this business has become, and human development may also enter a new era.
Elon Musk, the founder of Neuralink, announced on social networking that humans have received brain-computer interface (Neuralink) chip implantation for the first time, and the implanter is recovering well. Preliminary results show that neuronal spike detection is promising.
Musk then added, "When people implant brain-computer interface chips, they can control their phones or computers just by thinking, and they can control almost any device." The initial users will be those who are incapacitated. Imagine if we could get Stephen Hawking to communicate faster than a typist or auctioneer, and we would have achieved our goal. ”
Musk also introduced that Neuralink's first product is called "telepathy".
When the brain is connected to the device
Imagine if you could integrate your mind into a machine, as if a straightforward high-speed connection was created between your mind and the machine. This was the idea that came to Musk during a trip in 2016, when he felt it was too slow to type by hand.
And now, Musk's idea is becoming a reality, and the brain-computer interface is the technology that enables "mental handwriting", creating a direct connection pathway between the brain and external devices. The core is to give full play to the advantages of the human brain, bypassing the body's own organs, and the brain directly interacts with external equipment efficiently.
Why do we develop brain-computer interfaces?
First, it has proven to be the most powerful tool for nerve repair, providing a comprehensive solution for patients with impaired neurological function due to paralysis, stroke, and diseases such as Parkinson's.
Secondly, brain-computer interface, as a core key technology, helps us to fully understand the working mechanism of the brain, which is an important means in the field of international brain science research.
In the future, we are expected to develop brain-computer interfaces into smart devices that go beyond smartphones, allowing people to control the devices around them through their consciousness and giving them extraordinary endurance, speed, precision, and efficiency.
What's even more exciting is that last November, a Science article** showed that animals have the same imagination as humans, so connecting animal brains to devices will also make people have more imagination.
Plugging external pathways into the brain will undoubtedly damage the brain, so according to the degree of invasion and damage, the industry can divide brain-computer interfaces into three categories: invasive, non-invasive and semi-invasive.
Among them, the flexible electrode array has high stability and signal quality, but it is easy to cause tissue damage during implantation. Although flexible electrode arrays can reduce tissue damage, they are difficult to implant and have poor long-term stability. Therefore, how to combine the advantages of the two to develop a rigid-flexible adjustable electrode array technology has become a current research hotspot.
What's different this time?
If we count from the development of electroencephalography (EEG) by German doctor Hans Berg in 1924, brain-computer interface has formed a series of basic technical research and application paradigms after a hundred years of modern technological development.
Major historical events of the brain-computer interface, tabulation of the world of electronic engineering.
However, from the beginning to the end, brain-computer interfaces have lacked invasive clinical applications.
At present, there are three main technical routes for invasive brain-computer interfaces on the market: silicon-based hard electrode system, vascular electrode system and flexible electrode system.
These technological routes have made significant progress over the past decade, but they still face some core issues, such as high-throughput, low-trauma, and long-term in vivo challenges.
In the industry, some people have proposed that Moore's law of brain-computer interface is that the number of brain-computer interface neurons that can be read and written should be doubled every 18 months, and the growth rate of the number of brain-computer interface channels can be in line with the development law of semiconductors.
This is the first time that humans have received a brain-computer interface (neuralink) chip implantation, which is a major breakthrough in the clinical application of high-performance brain-computer technology.
Previously approved traditional implantable brain-computer interfaces, which use hard electrodes called "utah arrays," may cause a rejection response to foreign bodies inside the brain, which is often undesirable if more neural information is needed.
Neuralink uses flexible electrodes to effectively reduce the rejection of the brain, and has 1024 channels of electrodes to collect high-quality neural information.
At present, among the international implantable brain-computer interface companies, there are three companies that have entered the human clinical trial stage, namely Neuralink, Onward and Synchron. Among them, Neuralink belongs to the "cortical penetration" route; Onward belongs to the "cortical surface" route, and there are also micro-spirit medical treatments in China that take this route; Synchron is a "vascular intervention".
What do electronics engineers look for?
The key technologies of brain-computer interface include acquisition technology, stimulation technology, paradigm coding technology, decoding algorithm technology, peripheral technology and systematization technology. Among them, electrodes and chips are extremely relevant in electronic engineering. Specific technical details include:
Implantable electrodes
Implantable microelectrodes obtain information about the brain's neural electrical activity by converting ion-based neural electrical signals into electron-supported current or voltage signals.
Microelectrodes implanted into the brain can accurately record the action potentials of individual neurons in the vicinity of the electrodes with high spatial and temporal resolution, enabling real-time monitoring of brain activity.
Conventional implantable microelectrodes are made from hard materials such as metal and silicon, forming rigid electrodes dominated by Michigan electrodes and Utah electrodes. With the continuous development of micro-nano processing technology and electrode materials, microelectrodes tend to be flexible, miniaturized, high-throughput and integrated, forming a diversified development situation dominated by microwire electrodes, silicon-based electrodes and flexible electrodes.
High-performance flexible microelectrodes are of great significance for long-term stable chronic records, and high-throughput microelectrodes will lay an important foundation for expanding whole-brain neuroscience research.
Capture chip
At present, brain signal acquisition technology is moving forward in the direction of miniaturization, lightweight, high-throughput, and distributed acquisition. Among them, the brain signal acquisition chip is the core hardware that directly converts brain signals into digital signals, and it is also a tool for brain signal reading and decoding, and for the diagnosis and regulation of brain diseases.
There are many technical challenges in the design process of custom brain signal acquisition chips. The precision amplifier is the core module in the brain signal acquisition chip, which needs to meet the requirements of multiple technical parameters in the application scenario of brain-computer interface.
For brain signals, the amplitude is weak (tens of UVs to a few mV), the frequency is low (0.5Hz to several kHz), so it is susceptible to external noise interference, which leads to poor signal quality. In order to maintain the best signal quality, some key parameters of the brain signal acquisition module, such as signal noise common-mode rejection ratio (CMRR), power supply rejection ratio (PSRR), gain matching, motion artifacts, etc., need to be optimized.
There is a mutual constraint between multiple brain signal acquisition parameters, and the overall optimization of multiple parameters is one of the core problems in the current brain signal acquisition chip design.
Signal noise is one of the biggest sources of interference in the brain signal acquisition processThe common-mode rejection ratio is a key parameter to measure the response of a system to environmental interferenceThe miniaturization of the acquisition chip is one of the core technical challenges of the implantable brain-computer interface system.
For different brain-computer interface applications and some technical problems faced by acquisition chips, many teams at home and abroad have proposed solutions.
For example, in order to solve the problem of saturation of the output of the chopper amplifier caused by the DC bias between the electrodes in the acquisition process, a DC servo feedback loop technology extracts the DC component of the output end and feeds it back to the input end through an integrator, which effectively suppresses the DC bias between the electrodesFor the ultra-low power consumption requirements of the acquisition chip, a team has designed an ultra-low voltage chopper amplifier based on the inverter structure, which is very suitable for implantable scenarios. In order to solve the problem of chip miniaturization, the digital-analog hybrid feedback technology combined with the amplifier and DAC can greatly reduce the on-chip area of the acquisition chipIn order to solve the problem of common-mode interference in the process of brain signal acquisition, the common-mode feedback technology based on chargepump can effectively resist common-mode disturbance up to 15V by dynamically feeding back the common-mode disturbance signal at the input endFor the wireless power supply of the acquisition chip, the wireless inductive transmission technology of the coil is applied to the implantable brain-computer interface chip, and the wireless power supply of the acquisition chip in the body and the wireless transmission of the collected EEG signal are realized through the external transmission coil, the relay coil and the on-chip coupling coil. Body AreaNetwork (BAN) solves the problem of difficult coil alignment during wireless power supply, and uses the body surface of the subject to wirelessly transmit the collected signals and energy, which is suitable for wearable brain-computer interface scenariosIn terms of improving system integration, there are brain-computer interface system-on-chips that integrate signal acquisition, storage, and AI-based signal classification and recognition modules, achieving a high degree of system integration. For high-throughput implantable brain-computer interface chips, some companies have designed highly integrated acquisition chips with action potential recognition, which are combined with thousands of flexible electrodes to realize the acquisition of high-throughput brain signals. Another way for China's "straight-line overtaking".
Looking around the world, business giants such as the Defense Advanced Research Projects Agency (DARPA), Facebook, Google, and Amazon are actively deploying in the field of brain-computer interfaces, and achievements are constantly emerging, and high technical barriers have been formed.
My country is no exception. China's "Brain Project", also known as "Brain Science and Brain-like Research", will be fully launched as a "major project of scientific and technological innovation 2030".
Tao Hu, deputy director and researcher at the Shanghai Institute of Microsystem and Information Technology, Chinese Academy of Sciences, once wrote in an article: "Among high-end technologies, I think brain-computer interfaces are one of the areas where China is most likely to catch up or even 'overtake in a straight line'." At present, in terms of the design of brain-computer interface core devices, China is not lagging behind foreign countries at all, and its processing only involves mature semiconductor processes, and these core processing technologies do not face the problem and risk of being "stuck". Therefore, for China, to promote the future development of brain-computer interface, the main thing is to accelerate the allocation of resources and other issues, and all links work together to develop a full-chain independent and controllable brain-computer interface system, so as to provide solutions for the comprehensive development and smooth promotion of China's "Brain Project". ”
In terms of academia, at present, Chinese universities are very active in the research and development of brain-computer interaction technology, and Zhejiang University, Tianjin University, Southern University of Science and Technology, Shanghai Jiao Tong University, Xi'an Jiaotong University and other universities have made achievements.
Throughout China's capital market, it has also been active in the field of brain-computer interface. According to the statistics of Arterial Network, a total of 10 investment and financing events will occur in the field of brain-computer interface in 2023. Among them, Shenzhen Yinghe Brain Science *** hereinafter referred to as "Yinghe Brain Science") led the way in the first half of the year with a financing amount of more than 100 million yuan.
At present, China relies heavily on imports in the field of brain-computer interfaces, especially in the key devices and high-end equipment of implantable brain-computer interfaces. There is a lack of original core technologies in China, and most of the time it is conducting follow-up research, with scattered layout and lack of systematization. In the past two years, the United States has implemented export controls on brain-computer interfaces, which has affected the development of system-level products and core devices. This has not only had an impact on brain science research in China, but also had varying degrees of impact on patients with neurological diseases.
In addition, China's brain-computer interface research and development faces several major challenges:
First, it is difficult to achieve both safety and efficacy, and the unsolved problem limits the large-scale application of brain-computer articulation; The second is the effective bandwidth of the brain-computer interface, that is, how many electrodes are implanted to basically cover the important activities of the brain or meet the needs of specific functions, which is still unknown. Third, the processing of massive neural signals is still a difficult problem; Fourth, brain-computer safety and ethical risks are of general concern to society. It is also a big problem for the commercialization of brain-computer interfaces. Peng Lei, the founder and CEO of Shanghai Brain Tiger Technology, once said at the second Nandu River Forum. The challenges of commercializing brain-computer interfaces, including the cost of purchase and use by users and the value they bring to users. He believes that the combination of brain-computer interfaces with XR technology is a promising direction, because it can achieve no cost on the wear, and can provide more ways to interact, including motor interaction and emotional interaction.
Brain-computer interfaces, as a bridge between people and the outside world, are not absolutely secure, and they are also at risk of being maliciously attacked, which undoubtedly increases the complexity and uncertainty of decision-making and may lead to greater risks. Because the ultimate application object of brain-computer grafting is humans, it is necessary to strictly abide by national laws and regulations to ensure that clinical ethics requirements are met before clinical trials are carried out. As for future risks, we still need to remain vigilant and cautiously assess.