Little did Mr. Yang (pseudonym) expect that, after suffering a severe car accident and being paralyzed for 14 years, he would be able to grasp a cup with his right hand once again.
In his home, cameras captured the moment when he used a pneumatic robotic glove to grip a bottle of orange juice. It's evident that Mr. Yang is thrilled; this signifies a qualitative change for him—he hasn't been able to eat or drink independently for over a decade.
Mr. Yang successfully achieved brain-controlled grasping through wireless minimally invasive brain-machine interface. Image source: Tsinghua University
If there's anyone to thank, it's the technological advancements of this era, and the technology of "brain-machine interfaces" which, though still in the early stages of small-scale clinical trials, has become a hot topic. It was unfortunate that Mr. Yang encountered a severe car accident; however, he was also fortunate because on October 24, 2023, he became the first paralyzed patient on this planet to undergo successful wireless minimally invasive brain-machine interface surgery.
"Combining internal and external, wireless minimally invasive"
Mr. Yang's brain-machine interface surgery was successfully performed at Xuanwu Hospital in Beijing, under the leadership of Xuanwu Hospital President Zhao Guoguang and Director Shan Yongzhi. The device, named "NEO Wireless Minimally Invasive Brain-Machine Interface," was designed and developed by a team led by Professor Hong Bo from the School of Medicine at Tsinghua University, along with their spin-off company, "Borui Kang Technology."
Rome wasn't built in a day. Professor Hong Bo told the "China Science Daily" that the team had been conducting related research and development since 2013, spanning over a decade. "This has been a long journey, as we aimed to address the technical challenges of brain-machine interfaces, from a scientific concept to engineering implementation and clinical application."
In other words, the simple grasping action controlled by Mr. Yang's pneumatic mechanical glove represents the culmination of a decade-long effort by the entire team: "The signals need to be accurately captured, decoded correctly, and the execution devices need to function properly in order to create a complete application of controlling a mechanical glove through a brain-machine interface."
Professor Hong Bo vividly describes the brain-machine interface device and surgery used on Mr. Yang as "combining internal and external, wireless minimally invasive."
"Minimally invasive" is the most prominent keyword in this approach. Since the development of brain-machine interface technology, there have been generally two approaches: invasive and non-invasive (also known as invasive and non-invasive), differing in whether the electrodes for reading brain signals are inserted into brain tissue. However, Professor Hong Bo's team's design is somewhere in between, termed "semi-invasive."
He explained that the brain can "emit" signals both internally and externally, i.e., brainwaves, but the power of these emissions gradually attenuates. If the brain is likened to an auditorium, the invasive brain-machine interface approach is like placing several microphones evenly in the auditorium, while the non-invasive approach is like placing a speaker only outside the auditorium. Their semi-invasive approach takes a middle ground: placing the microphones inside the door.
This is exactly what Mr. Yang's situation looks like. The brain-machine interface device on his head consists of three parts: electrodes, an internal device, and an external device. The research team first located Mr. Yang's motor cortex using magnetic resonance imaging, which is where the craniotomy (only opening the skull without damaging the dura mater) was performed. Then, electrodes for signal acquisition were placed outside the dura mater, between the skull and the brain cortex. The internal device responsible for processing and communication was implanted about 6-10 millimeters into the skull, and the external device, which can be attached to the scalp, both receives and transmits brainwave signals while also providing power to the internal device through the scalp.
"We are definitely not doing this to show off, but to achieve a balance between the performance of brain-machine interface signals and the trauma to patients, maximizing the benefits for the patients," said Professor Hong Bo.
He named their implanted device the "NEO Wireless Minimally Invasive Brain-Machine Interface System," which utilizes near-field wireless power supply and communication technology. To achieve minimal invasiveness as much as possible, the entire system, integrating 329 components, was made to be as small as two one-yuan coins, "solving very complex engineering problems inside."
The results were very satisfying: Mr. Yang was discharged from Xuanwu Hospital on the 10th day after surgery.
Proudly, Professor Hong Bo said, "So far, our solution is the only one in the world that allows patients to go home 7-10 days after brain-machine interface surgery."
After returning home, Mr. Yang's brain-machine interface device worked normally. After about a month of training, his right hand, with the help of the pneumatic mechanical arm, was able to grasp a mineral water bottle. Of course, the instruction signals for grasping came from Mr. Yang's brain cortex - they should have been relayed to his fingers through neurons inside the body, but now, after being collected, transmitted, and decoded, the signals ultimately controlled the pneumatic arm.
"Colliding" with Musk
On January 29, 2024, the joint team of Tsinghua University and Xuanwu Hospital held a summary meeting of the clinical trial phase, announcing a breakthrough in the rehabilitation of the first patient with a brain-machine interface.
Unexpectedly, both China and the United States, across the Pacific Ocean, coincidentally made a historic breakthrough in the exploration of brain-machine interface technology.
On January 30, Elon Musk announced that his brain-machine interface company Neuralink had completed the first human brain device implantation surgery and tweeted that "preliminary results are promising." On February 20, Musk announced on the social media platform X that the first human patient implanted with a brain chip by Neuralink had successfully controlled a computer mouse with their thoughts.
Neuralink's Brain-Machine Interface Implant
Image Source: Neuralink
This marks the commencement of human trials for Neuralink's brain-machine interface, following two rejections and a delay of four years by the FDA.
However, it's important to note that Neuralink's approach involves a more invasive, fully implantable solution, which carries higher trauma risks. In comparison, the semi-implantable brain-machine interface clinical trials conducted by Tsinghua University - Xuanwu Hospital appear to be much milder.
Objectively speaking, China's control over the clinical trials of brain-machine interface technology is stricter than that of the United States.
Dr. Li Xiaojian, a senior engineer at the Shenzhen Advanced Technology Institute of the Chinese Academy of Sciences, is an expert in implantable brain-machine interface technology. "Neuralink represents the implantable brain-machine interface technology pathway," he told Science China, stating that this technology has been used in animals for many years and involves implanting electrodes into the brain.
He further explained that the Neuralink team's implantable brain-machine interface requires electrodes to penetrate brain tissue, but only to a shallow depth in the cortex - about a few millimeters, also known as "cortical signal brain-machine interface." Dr. Li stated, "This approach, due to the large area of the cortex and the abundance of information it represents, is suitable for brain control."
Dr. Li Xiaojian succinctly highlighted the advantages of the implantable approach: the electrodes inserted into brain tissue, essentially squeezed into the "stack" of neurons, can easily obtain high-frequency signals (up to several kHz). In contrast, non-implantable approaches, separated by the skull and scalp, can only capture very low-frequency (below 20 Hz) neural impulses. The "semi-implantable" approach by the Hongbo team at Tsinghua University captures signal intensity between these two extremes.
In Dr. Li Xiaojian's view, the "semi-implantable" approach highlights the ingenuity of the Hongbo research team - this conservative yet robust approach directly promotes the clinical exploration and application of brain-machine interface technology in China. Some industry insiders suggest that more aggressive fully implantable approaches advocated by Musk may not be approved for clinical trials in the near term under China's stringent regulations.
"It can be said that the Tsinghua team has made 'combinatorial innovations' on more traditional neural medical devices, which has facilitated the smooth approval of brain-machine interface technology for clinical trials and opened up a new stage of exploration for Chinese brain-machine interface technology," said Dr. Li Xiaojian.
Returning to the scientific discussion, apart from the higher risk of trauma, the fully implantable approach indeed harbors more hidden dangers.
Hongbo explained to Science China that they also considered the risks of fully implantable approaches. Firstly, a fully implantable brain-machine interface system requires an open incision for implantation, which poses a risk of infection. Secondly, electrode insertion into brain tissue may also trigger an immune reaction from glial cells. This could lead to a chain reaction: cells may envelop the electrodes, resulting in deteriorated signals, while the electrodes may pose other risks due to scab formation.
Recently, the UK's Nature website reported that some researchers do not consider the achievements of Musk's Neuralink team as a major innovation. Additionally, they questioned the safety and confidentiality of the device.
Similarly, Miguel Nicolelis, the "father of brain-machine interfaces" and retired professor at Duke University in the United States, expressed pessimism about the future of implantable brain-machine interfaces in an interview with Science China: "Although I invented invasive brain-machine interfaces and have 20 years of patents in this area, I believe that for most patients and commercial enterprises, non-invasive brain-machine interfaces will be the mainstream of development in the next few years (in the field of medical rehabilitation)."
A New Beginning
Science China learned that using the semi-implantable approach, the Hongbo team, in collaboration with Tian Tan Hospital, has already completed the second clinical trial of brain-machine interfaces.
"The second patient is younger, in their thirties, and also suffered from spinal cord injury caused by a car accident. They were successfully implanted by Director Jia Wang's team at Tian Tan Hospital on December 19, 2023," said Hongbo. The condition of this young patient is relatively severe; they cannot move their elbows and cannot operate a mechanical hand, but they can control a screen cursor and hit a blue ball with a red ball on the screen.
"We believe that after a period of adaptation, this patient will be able to independently control computers and smartphones using the brain-machine interface, and flip through books on the screen," said Hongbo.
Remember the patient from Neuralink, Elon Musk's company? They underwent a fully implantable approach and could only "successfully control a computer mouse with their thoughts."
With the success of the aforementioned two patients' trials, the Hongbo team is looking to the future.
"We hope to develop and obtain regulatory approval for third-class implantable medical devices (note: third-class devices are the highest level). This requires large-scale clinical trials for verification and, optimistically, at least two years," envisioned Hongbo. With this approval, their brain-machine interface system can be promoted and used nationwide. "Of course, this requires patience; we are still conducting small-scale clinical trials."
Technologically, the Hongbo team has set higher goals. "We plan to upgrade every two years to continuously improve the performance of the system," he said, using "channel count" as an example. The current brain-machine system only has 8 channels, capable of only simple hand movements. In the future, the number of channels will gradually increase to achieve more brain-machine interface functions.
Of course, he is well aware that there are significant technological challenges behind this - an increase in the number of channels can lead to electrode heating, a problem faced worldwide. "We hope to find a balance between safety, stability, and performance through a large amount of engineering innovation."
The clinical application potential of brain-machine interface technology also includes lower limb rehabilitation, spinal cord injury repair, speech decoding, and more. Hongbo envisions that in the not-too-distant future, brain-machine interface technology will not only help patients like Cai Lei, who has amyotrophic lateral sclerosis, but also treat their diseases.
Furthermore, the clinical exploration of brain-machine interface technology is not only about medical treatment but also about opening a window for human understanding of the brain. Dr. Li Xiaojian told reporters that the advancement of human trials of brain-machine interface technology will bring more brain information that is still unknown, promoting the development of human-like intelligence. Hong Bo also agrees with this argument: "Brain-computer interfaces could potentially be a significant step in human evolution." He says, "What (Old Yang) has done is not the end, but a new beginning."
