Author: Zhou You
Published in the 1130th issue of "China News Weekly" magazine on March 4, 2024
On the computer screen, a red circular cursor was moving slowly towards a blue circular cursor in the corner, and the two finally overlapped completely.
This is the first time in China that minimally invasive brain-computer interface technology has been used to successfully help patients with high paraplegia achieve the process of controlling cursor movement with their thoughts.
On February 25, Bai Hao, a high-level paraplegic, recreated this scene to China News Weekly at his home in Beijing.
He underwent brain-computer interface implantation surgery at Beijing Tiantan Hospital affiliated to Capital Medical University in December last year and is currently undergoing rehabilitation training.
Hong Bo, a professor at Tsinghua University School of Medicine and deputy director of Tsinghua Artificial Intelligence Research Institute, is the person in charge of this clinical trial. Bai Hao is the second subject of Hong Bo’s team.
Now, Bai Hao can control the mechanical glove through the brain-computer interface to grasp the water bottle.
On February 20, Elon Musk, founder of the American brain-computer interface company Neuralink, stated on the social media platform "Move the mouse on the screen with your thoughts" operation, but it did not release any data or details.
Musk said on X on January 30 that the subject recovered well and the brain signals showed that the brain-computer interface was "very promising."
On the same day, Hongbo's team also officially announced the results of the world's first clinical trial of wireless minimally invasive brain-computer interface implantation. The team's first high-level paraplegic subject successfully achieved brain-controlled drinking.
In January this year, seven departments including the Ministry of Industry and Information Technology and the Ministry of Science and Technology jointly issued the "Implementation Opinions on Promoting Future Industrial Innovation and Development", in which the brain-computer interface is one of the innovative iconic products.
Hong Bo told China News Weekly that there is "no homework to copy" for the clinical application of brain-computer interfaces. We can only move forward step by step, starting with medical devices, with the primary goal of helping the majority of patients with spinal cord injuries, and achieving their goals. desire to return to society.
However, there is still a long way to go before the results of brain-computer interface technology can be transformed from clinical scientific research to clinical treatment.
The "third hand" for patients with high paraplegia
Bai Hao is from Inner Mongolia. In 2019, he suffered a high-level paraplegia due to a car accident. He lost consciousness below the cervical spine and had almost no ability to move. He needs full-time supervision from his parents.
On weekdays, he spends most of his time lying in bed, and has rehabilitation equipment training in the morning and afternoon every day.
Due to being unable to exercise for many years, his bone and muscle strength are weaker than ordinary people, and he can only sit for one or two hours at a time.
Before being implanted with a brain-computer interface, he was used to holding a stylus in his mouth for simple reading and entertainment on a tablet.
Ten days after the operation in December last year, Bai Hao was discharged from the hospital and immediately devoted himself to brain control training.
Bai Hao's father has learned to skillfully tie the brain-computer interface host to Bai Hao's arm and connect the coil to the surgical incision behind his head.
Just like the wireless charging of a mobile phone, as long as the external machine is connected to the power supply, the internal processor can obtain power through wireless transmission. The signal monitoring software on the computer screen can display the EEG signal curve in real time, corresponding to the 8 implanted cells in Bai Hao's brain. electrode.
Hong Bo's team developed the cursor control algorithm. He told China News Weekly that mechanical grasping and cursor chasing may seem simple to operate, but the simulation is very complicated, involving brain electrical signal decoding and various machine learning algorithms.
The curve displayed on the screen contains the specific movement intention in the patient's mind. The more accurate the interpretation, the freer and more precise the patient's control will be.
Hong Bo said that data collection inside the brain and data decoding outside the body jointly build a bridge between the human brain and external devices. This kind of "integration of inside and outside" is the most basic working logic of the brain-computer interface.
Hong Bo said that the brain-computer interface developed by the team is a semi-invasive minimally invasive wireless brain-computer interface.
Semi-invasive means that the electrodes used to collect brain electrical signals are implanted into the cranial cavity but do not directly contact brain cells.
The method used by the team is to thin the inside of the skull and embed a miniature brain-computer interface processor. The semi-invasive design can effectively avoid the risk of infection and inflammation in the brain.
Currently, the industry is also developing non-invasive and invasive brain-computer interfaces.
It is non-invasive and does not require surgery. It only needs to attach the electrodes for collecting EEG signals to the scalp. This type of equipment has achieved localization of technology and has products on the market.
In 2020, the country's first ultra-wideband EEG collection device independently developed by Boricon Technology Co., Ltd. obtained a medical device registration certificate. The device resembles a mesh woolen cap worn on the user's head and is called "EEG." cap".
Wu Jinsong, chief physician of the Department of Neurosurgery at Huashan Hospital Affiliated to Fudan University and deputy director of the Brain-Computer Interface and Interaction Subcommittee of the Chinese Neuroscience Society, told China News Weekly that non-invasive extraction is a local signal, which is equivalent to the average neuron activity. Although the risk is small, the accuracy is low and not many functions can be achieved.
The invasive method allows electrodes to enter the cerebral cortex and make close contact with neurons to obtain higher-quality neural signals, but the operation is risky and costly.
The brain-computer interface released by Neuralink is invasive.
The company's official website shows that subjects need to implant a coin-sized device into the brain, connected to 1,024 flexible electrodes, and a surgical robot will implant these electrodes into the cerebral cortex like a sewing machine.
The chip wirelessly transmits recorded EEG signals to an application that decodes movement intentions, allowing patients to control devices such as external mice and keyboards via a Bluetooth connection.
Christian Huff, an assistant professor at the School of Mental Health and Neuroscience at Maastricht University in the Netherlands, said in a reply to China News Weekly that the breakthrough of Musk's team is that for the first time, it has achieved a wireless device that completely seals the human brain. implantation, and improved the theoretical accuracy of the signal to the single neuron level.
Neuralink's large number of flexible electrodes can establish electrical signal connections with individual neurons and float with the brain, significantly reducing electrode damage to brain tissue.
Wu Jinsong said that Neuralink’s hardware integration process is excellent and can make the device small enough and low enough power to prevent local heating.
In terms of engineering technology and product technology, there is still a considerable gap between domestic teams.
"The difference between us and Musk lies in the trade-off between signal bandwidth and security, not in the absolute advantages and disadvantages." Hong Bo said that the EEG signal bandwidth represents the interpretability and accuracy of the signal, and the bandwidth of its products is definitely not as good as Neuralink. But invasive products carry high risks of implanting electrodes into the brain.
Since invasive brain-computer interfaces require opening holes in the skull and embedding sensors, once the fit between the incision and the device is not perfect, there is a risk of bacterial infection in the brain.
Secondly, although flexible electrodes avoid damage to brain cells by the electrodes, they still cannot avoid the brain's rejection reaction to foreign objects.
Immune cells will gather on the surface of metal electrodes and form "scabs", which will significantly affect the working efficiency of the electrodes in the long run.
Moreover, with invasive brain-computer interfaces, patients require more time in the hospital for observation after surgery.
Li Xiaojian, a senior engineer at the Shenzhen Institute of Advanced Technology, Chinese Academy of Sciences and founder of Weiling Medical Technology Co., Ltd., told China News Weekly that under ideal conditions, the more and denser the electrodes, the higher the spatial resolution of brain nerve electrical signals, and The more it can reflect the activity of a single neuron.
However, in actual operation, the electrode "occupies a lot of space" and will squeeze the neurons, which is a destructive test.
Animal experiments show that the "scab" problem due to rejection will cause the electrodes to have a short lifespan, and the signal will deteriorate significantly after a few months.
When the signal is insufficient for decoding, the electrode can only be removed surgically, and the brain tissue environment in the area where the electrode was removed has been destroyed and cannot be re-implanted.
The safety and reusability of invasive brain-computer interfaces are still unresolved.
Before finding Bai Hao, the first subject of Hong Bo's team was 54-year-old patient Yang Feng.
He underwent the same brain-computer interface surgery at Beijing Xuanwu Hospital of Capital Medical University on October 24 last year.
This patient also suffered from high paraplegia due to a car accident in 2009, but his right arm mobility is slightly better than that of Bai Hao. He can not only control the robot hand to grasp, but also complete the operation of drinking water, and his grasping accuracy is over 90%.
Hong Bo said that animal experiments and two human trials have proven the safety and effectiveness of this device.
The team’s current goal is to mature the first version of the technology and obtain a nationally approved medical device registration certificate before considering exploring indications.
In the future, as equipment capabilities improve, the application scope of brain-computer interfaces will transition from patients' "third hand" to walking and movement.
In principle, language decoding, brain electrical stimulation to treat epilepsy, depression, ALS, etc. are all potential application areas. In the future, brain-computer interfaces may become universal medical devices, benefiting patients with a variety of neurodegenerative diseases.
Semi-invasive may become a domestic technical route
Hong Bo said that many patients refuse to participate in brain-computer interface clinical trials because they are worried about surgical risks, sequelae, etc. Therefore, in China, it is possible to persuade patients to participate in trials by adopting a semi-invasive compromise plan.
This also means that non-invasive and semi-invasive designs may become the technical route for domestic brain-computer interface industrialization.
Recruiting patients is just one of the difficulties in advancing brain-computer interface technology into clinical applications.
Li Xiaojian said that the brain-computer interface is still in the clinical research stage and is transitioning from animal experiments to human trials and requires some samples and precedents.
The United States leads the world in innovative medical devices and has rich experience in review and supervision.
Nerualink's solution has just entered scientific research-level clinical human trials in the United States. Due to safety concerns, it will be difficult for such a solution to be approved in China.
According to the "Regulations on the Supervision and Administration of Medical Devices" issued by the State Council in 2021, innovative medical devices must pass safety and ethics review before entering human trials.
Hong Bo said that the safety inspection involves dozens of tests such as electrical safety and biocompatibility, and is conducted by specialized agencies.
Ethical review is conducted by the ethical review committee organized by the hospital.
The two trials conducted by Yang Feng and Bai Hao passed ethical review in May and August last year respectively.
In Hong Bo's view, the research team needs to convince the experts of the ethical review committee that the benefits to patients far outweigh the risks.
Invasive brain-computer interfaces obviously do not yet meet such conditions.
Many interviewees said that Neuralink did not provide complete evidence of animal experiment risk control before human trials.
In Hong Bo's view, the ethical review mechanism is not a blocking point for the clinical transformation of brain-computer interfaces. Domestic ethical review standards do not lag behind those in Europe and the United States. As long as there is sufficient animal experiment data and a safety inspection report issued by a third-party organization, the ethics review can be passed. There will be no difficulty in reviewing.
The difficulty is that the overall ethical review standards in China are higher and the conditions for products to enter clinical trials are more stringent.
Li Xiaojian said that invasive brain-computer interface devices are a new technology, and there are only a few institutions in China that can conduct safety inspections. They have limited manpower and equipment, and they often need to queue up to apply, which to some extent also slows down clinical trials. Conversion progress.
In October last year, the "Technology Ethical Review Measures (Trial)" issued by 10 departments including the Ministry of Science and Technology pointed out that seven human experimental technologies, including brain-computer interfaces, need to undergo strict ethical review and must be reviewed by the ethics committee of the clinical unit. Must be reported to the National Health Commission.
On February 2 this year, the Ministry of Science and Technology released the "Ethical Guidelines for Brain-Computer Interface Research" compiled by the Artificial Intelligence Ethics Subcommittee of the National Science and Technology Ethics Committee. This is the first ethical code for brain-computer interfaces in China.
The guidelines point out that invasive brain-computer interface research should have stricter standards than non-invasive ones. On the premise of sufficient scientific evidence, strict and prudent evaluation procedures are required to determine the necessity of using invasive brain-computer interfaces.
Domestic brain-computer interface companies are constantly making moves in clinical translation.
In 2022, Shanghai Naohu Technology Co., Ltd. received more than 97 million yuan in financing and developed a flexible electrode based on silk materials. Its tissue compatibility and safety exceed similar products.
At present, this flexible brain-computer interface system has been clinically approved by the Food and Drug Administration and is in the recruitment stage of volunteers. It is mainly targeted at people with ALS and high paraplegia.
Wuhan Zhonghua Brain-Computer Integration Technology Development Co., Ltd. has developed an implantable brain-computer interface system. The maximum number of channels can reach 20 times that of Neuralink products. It has been recognized as a "world's first" by an expert group. It has also ended animal experiments and is committed to Human trials proceed.
After eight years of research on implantable brain-computer interface technology for visual and motor systems in the United States, Li Xiaojian returned to China in 2018. The following year, he and his colleagues founded Weiling Medical.
In the past five years, he has witnessed the transformation of brain-computer interfaces from “unpopular” to “popular”.
In his view, the original intention of transforming brain-computer interfaces into medical-grade results is to achieve clinical applications.
Neurodegenerative diseases are still incurable, due to a lack of understanding of the human brain.
For example, Li Xiaojian said that the accuracy of neural electrical impulses is sub-millisecond, but the measurement accuracy of nuclear magnetic resonance used to draw dynamic maps of the brain is second-level, and the time resolution is thousands of times worse.
The lag in human brain scientific research is a fundamental limitation on the application of brain-computer interfaces.
In the layout of the transformation of results, the company must first be committed to promoting human brain scientific research through implantable brain-computer interface technology, and secondly, promote and popularize more mature technologies in medical treatment.
Wu Jinsong believes that at present, most breakthroughs in brain-computer interface technology are in terms of technology, and there is still a lack of a deeper understanding of the scientific principles behind brain signal encoding and decoding.
It is relatively easy to decode signals about body movements, but functions such as language will be more complex, and the encoding and decoding work is very difficult.
As for memory or consciousness, it has not yet been determined whether its essence is an electrical signal. Without knowing the essence, it is impossible to talk about the so-called uploading or immortality of consciousness.
For example, he said that many human emotional reactions are not based on electrical signals, but on the transmission of chemicals such as dopamine between nerve cells. The simulation of such signals has not yet begun.
In Li Xiaojian's view, although brain-computer interface technology is still in the clinical research stage, its industrial supply chain has begun to take shape.
Domestic micro-nano processing technology is relatively mature, the supply chain of electronic components is complete, and consumables for brain-computer interfaces are not a problem.
Software decoding capabilities and supporting computing power are currently sufficient for brain-computer interfaces.
In terms of downstream applications, according to data from the official website of the China Disabled Persons' Federation, there are at least 1.3 million patients with spinal cord injuries in China.
Wu Jinsong said that the treatment logic brought by brain-computer interface is a breakthrough from 0 to 1, which will bring about a qualitative change in the lifestyle of these patients.
The brain-computer interface also brought unexpected joy to Yang Feng.
Hong Bo said that one day two months into the training, Yang Feng said to the researchers: "Your hands are so cold." Everyone present was shocked.
Yang Feng himself seemed at a loss as his hands had not felt hot or cold for more than ten years.
Not only that, he also found that his fingers had a grasping feeling.
Hong Bo said that damaged nerve fibers may recover due to continuous training, which can be confirmed through data. This part of the research is currently waiting to be published.
Andrew Schwartz, a professor of neurobiology at the University of Pittsburgh, also mentioned to China News Weekly that the effectiveness of most brain-computer interfaces will improve with training, which is related to neuroplasticity, a bit like the brain learning to use new of video game controllers and become more and more proficient.
Brain-computer interfaces will help study the human brain, which is where its value lies.
(Bai Hao and Yang Feng are pseudonyms in the article)
"China News Weekly" Issue 8, 2024
Statement: The use of articles from China News Weekly must be authorized in writing.