Embodied artificial intelligence robots are revolutionizing industries worldwide. Hong Kong, aiming to be a global medical technology hub, is leveraging AI-powered surgical robots in the operating room to shatter the limits of human precision. Shadow Li reports from Hong Kong.

Editor’s note: As artificial intelligence becomes increasingly important in the international race to get ahead, embodied AI could be the next big thing. In a series of articles, titled Hands-on AI, China Daily looks into the technology’s applications in different scenarios, with the second installment focusing on surgical robots.

For years, neurosurgeon Danny Chan Tat-ming had faced impossible choices — risking damage to his patient’s brain with rigid endoscopes, or drilling a separate hole into the skull.

He declined both options and sought a way out at international conferences, where he met engineer Liu Hongbin — a pioneer in flexible surgical robots.

The duo teamed up and produced MicroNeuro — the world’s first flexible surgical robot for minimally invasive neurosurgery — overcoming human physical limits while improving outcomes.

The idea stemmed from Chan’s work treating patients with pineal germ cell tumors in the central nervous system in Hong Kong in 2015. The city has the world’s highest incidence of the tumors, with up to 20 cases annually, Chan says.

READ MORE: Chan: Finance, innovation technology the path to HK's future

Chan, an honorary clinical assistant professor at the Chinese University of Hong Kong’s Division of Neurosurgery and who treats about eight patients annually, said they have a 90 percent five-year survival rate after treatment.

To minimize the impact on healthy brain tissue, a surgeon would normally drill two holes in a patient’s skull — one for a biopsy to confirm the tumor and another to drain fluid buildup caused by the tumor. Chan had envisaged a single hole to do both, with a flexible endoscope that could make up to a 60-degree turn in the brain to drain the fluid and then bend backward for the biopsy.

In 2020, Chan pitched his idea of a flexible surgical robot again after several futile attempts, hoping to resolve the problem once and for all. Liu, who had returned from the United Kingdom to join the Centre for Artificial Intelligence and Robotics at the Chinese Academy of Sciences’ Hong Kong Institute of Science and Innovation, immediately aligned himself with Chan’s vision. Their concerted effort led to the creation of MicroNeuro in about a year.

MicroNeuro addresses two key neurosurgical challenges — executing precise maneuvers in the brain’s delicate tissues, and providing real-time visual feedback of these movements.

“The hardest part was translating Chan’s clinical needs into engineering specifications,” Liu says. Chan emphasized safety and stability, which the team materialized through haptic vibration feedback — meaning that each MicroNeuro movement provides physical response.

For visualization, MicroNeuro integrates the CARES Copilot (now in its second generation) — an AI multimodal system combining virtual reality-assisted navigation for real-time endoscopic tracking and allowing surgeons to “see through” brain structures while operating — as well as multidata processing of voices, ultrasound and computed tomography scans.

Dual-endoscope system

MicroNeuro emerged from a critical need in neurosurgery. Standard 7-millimeter-diameter endoscopes, designed for bronchoscopy, proved dangerously large for brain surgery for which tools with a diameter of less than 4 mm are required. What’s more, the tools have to accommodate camera, light and cannula systems to allow instruments for suction or biopsies.

Moreover, rigid endoscopes, with limited range of motion, require surgeons to manually adjust them to reach the lesion at the proper angle — an imprecise and hazardous process that could damage the brain and potentially lead to memory loss in patients.

MicroNeuro’s dual-endoscope system uses a sheath endoscope to track the front endoscope’s path. The system enables drainage and biopsy through a single entry, avoiding a second hole, which would involve removing healthy brain tissue to clear a surgical path.

With MicroNeuro, one surgeon alone could complete the surgery, which traditionally is a two-man job. Following cadaver trials in 2022, human clinical trials are planned for 2027.

Liu admits that commercializing a complete MicroNeuro system remains challenging. To accelerate adoption, his team is unbundling it into modular components.

This month, they will launch a stand-alone hand-held version of the flexible endoscope, slated for commercialization next year. Similarly, the Copilot 2.0 will be marketed separately.

With such a device typically priced at HK$3 million ($382,000), the Copilot 2.0 is priced at HK$300,000. Hospitals in Beijing and Guangdong have expressed interest.

The Copilot AI system, adaptable to other robots, requires partnerships with device makers and retraining for new applications.

Domain-specific AI

“General AI, like ChatGPT, has limited use in neurosurgery,” as “we need specialized AI tools in our field,” Chan says.

“While AI seems omnipotent, real-world solutions require deep vertical expertise,” Liu says, adding that general AI’s shortcomings — unclear logic, hallucinations and unreliable answers — make doctors hesitant to trust it in making medical decisions.

This has led to Copilot 2.0, providing cited medical evidence for all its recommendations and helping conduct surgical planning and postoperation reviews. The system is being tested at Hong Kong and mainland hospitals.

“For MRI or CT lesion detection, trained AI already surpasses clinicians in accuracy,” says Liu. “But diagnosing complex, multisystem cases still requires human expertise. Until we have achieved theoretical breakthroughs, AI won’t match specialists handling these challenging diagnoses.”

“Neurosurgeons embrace AI as a powerful ally,” Chan says. MicroNeuro’s AI assistance gives them X-ray-like vision and real-time guidance, elevating operation precision.

For trainees, this technology compresses years of hands-on learning into accelerated practice with virtual mentorship. While dismissing replacement fears, Chan sees AI freeing up surgeons to focus on patients and helping to develop better systems. “Most errors that stem from human factors can be eliminated by machines. Soon, patients will trust the precision of AI-assisted surgery like they trust airline safety today.”

Chan’s confidence was justified. “No surgical errors or complications relating to Da Vinci Robotic Surgical Systems have been reported in the past three years, despite 1,200 to 1,500 annual cases across seven Hong Kong public hospitals utilizing the system,” a government source confirmed.

Chan envisions a future in which neurosurgical robots operate with full autonomy. He points to existing autonomous technologies like Gamma Knife radiosurgery and magnetic resonance-guided focused ultrasound as proof of this inevitable evolution.

“These systems already perform precise treatments with minimal human intervention. With embodied AI, we are moving toward completely autonomous neurosurgical suites.”

MicroNeuro’s significance transcends offering a new approach in surgery. It enables neurosurgeons to achieve what was previously impossible — surpassing human limitations, whereas other surgical robots only enhance existing capabilities, Chan says.

The next-generation multiarmed MicroNeuro system is slated for 2027-28, promising even greater advancement.

Underdeveloped field

The World Health Organization estimated that over 1 in 3 people are affected by neurological conditions. A study published in the Journal of Neurosurgery in 2018 estimated that every year, 22.6 million patients worldwide suffer from neurological conditions, and 13.8 million need surgery.

The field remains underdeveloped because of a shortage of specialists and costly equipment, Liu explains.

Neurosurgery remains an untapped niche market, he adds, noting a paradigm shift toward minimally invasive neurosurgery and the new generation’s growing acceptance of tech-assisted treatment.

Chan says AI-powered systems could revolutionize treatment for currently inoperable conditions.

According to a report by The Insight Partners, the value of the global surgical robotics market is projected to reach $28.54 billion by 2031 from $10.25 billion in 2023, at a compound annual growth rate of 13.7 percent.

GBA’s edge

Liu and Chan resemble a combination of AI and surgical robotics.

Liu, who spent 15 years at King’s College London doing research, says he was encouraged by the Guangdong-Hong Kong-Macao Greater Bay Area’s push to become a global innovation hub, which Hong Kong complements well with clinical and technical strengths.

Hong Kong’s edge is its world-class clinicians — innovative, globally influential and  interested in cutting-edge medical devices, making the city an ideal place for developing medical robotics. “Hong Kong surgeons exceed my expectations — not just in skill, but in devotion of innovation, even surpassing Western peers,” Liu says.

Hong Kong’s universities have top researchers, while the nine Greater Bay Area cities on the Chinese mainland excel in fields of engineering, electronics, software and mechanics.

Liu says that the GBA’s 87 million residents offer vast clinical data for AI training, with Shenzhen and Guangzhou hospitals collaborating to refine their surgical robot and AI systems.

Original ideas, rapid engineering and a large market for fast product iterations — three things that drive innovation — are all essential. Hong Kong, a relatively small market, must work with the mainland, which offers unmatched engineering talent and scale, Liu adds.

Chan sees Hong Kong’s potential as a surgical robot testing hub, leveraging its research excellence, world-class surgeons and the GBA’s engineering strengths to create a complete ecosystem.

A similar embodied AI flexible surgical robot for neurosurgery is being developed by Curve Robotics, a startup incubated by the Chinese University of Hong Kong. Chen Jibiao, member of Curve Robotics’ research team, says its surgical robot is set to debut as early as the end of this year or in 2026.

Call for regulatory body

Of the 300 biotech firms at Hong Kong Science Park, 10 are focusing on AI surgical robots.

However, Benny Lo Ping-lai, of the Hong Kong MedTech Association, says that the lack of a local regulatory body for medical-innovation validation hinders medical technology startups’ development.

Without such a regulator, devices must seek approvals from external authorities, such as the United States Food and Drug Administration and the National Medical Products Administration on the mainland, before registering in Hong Kong — a difficult process that would involve trials in those regions. The FDA’s approval often needs Caucasian trial data, while the NMPA requires data from mainland hospitals.

The Hong Kong Special Administrative Region government recognizes the need for its own medical registration regulatory body. Chief Executive John Lee Ka-chiu’s 2023 Policy Address proposed establishing the Hong Kong Centre for Medical Products Regulation to provide evaluations for local medtech innovations.

“Securing the first approval is always the hardest,” Lo says. “But once this is achieved, subsequent market access becomes easier as startups can use this evaluation to knock on other markets for approval.”

The proposed CMPR would be part of the SAR’s broader medtech hub ambitions. As an interim measure, the “1+” mechanism, introduced in November 2023, streamlined drug and device registration, allowing approval with just one external authority’s endorsement plus local clinical data for life-threatening conditions.

Fred Tung, a member of Hong Kong Health Bureau’s Advisory Committee on Health and Medical Innovation Development, says the CMPR would help local startups and mainland companies expand globally, provided they get approvals from both the CMPR and NMPA.

Yet, unlike other markets that prioritize local innovations, Hong Kong’s Hospital Authority — the largest medical device buyer in the city — dose not show an apparent preference for homegrown products. “It’s like pitting a child against an adult,” says Lo, noting multinationals’ resource advantage in securing HA contracts. With hospital procurement cycles spanning a decade, many startups can’t survive long enough to compete.

ALSO READ: Chan: Family businesses need to respond to demands on sustainability

Braillic — a Hong Kong-based startup that provides augmented-reality neurosurgery navigation — is seeking more funding for validation of its product after having secured HK$1 million in angel funding. “First, we’ll target approval from the NMPA on the mainland, then acquire Europe’s CE marking and, finally, the FDA’s backing,” says Camroo Ahmed, Braillic’s CEO.

Ahmed, who founded Braillic after graduating from the City University of Hong Kong, says the city’s global ties, talent pool and proximity to the vast mainland market are key reasons for the company to stay.

Braillic plans to expand to Shenzhen, but will keep Hong Kong as its hub, leveraging the city’s world-class neurosurgeons.

Liu believes the true value of technology such as AI is augmenting human capabilities — overcoming physical and cognitive limits.

Liu envisions an operation room where surgeons command AI-assisted devices like MicroNeuro, enabling junior doctors to perform like experts. AI surgery planning, he says, will soon be as simple as planning a trip.

While being bullish on AI’s potential to reduce costs and expand scope in neurosurgery, Liu stresses that doctors must have the final say in decision-making.

AI advances, but surgeons grow along with the technology, he says.

The future forward for MicroNeuro’s R&D team

1. Develop a multiarm flexible surgical robot.
2. Scale down the footprint of the robotic arm.
3. Upgrade CARES Copilot as an AI-powered surgical assistant.
4. Incubate a portable hand-held flexible endoscope. 

Contact the writer at stushadow@chinadailyhk.com