--- title: "Through a Vein, Not the Skull: How China's \"Third Kind\" of BCI Helped a Paralyzed Man Use His Arm Again" date: 2026-10-02 category: Brain–Computer Interface site: NeuroAI canonical: https://neuroai.site/a/na-nankai-vein-delivered-bci-stroke language: en --- # Through a Vein, Not the Skull: How China's "Third Kind" of BCI Helped a Paralyzed Man Use His Arm Again > Nankai University's team delivered a brain-computer interface to a stroke patient's motor cortex through his own blood vessels — the world's first interventional BCI trial to restore limb movement, done without opening the skull. A 67-year-old man sits in a rehabilitation chair in Tianjin. Six months earlier, a stroke had left his left side paralyzed. Now he slowly raises his left arm and, steady enough, picks up a water cup. He did not think especially hard about it — and that is the point. The thought "lift my arm" finally has a road out of his brain. The road runs through one of his veins. In June 2025, a team led by Professor Feng Duan (段峰) of Nankai University completed what researchers describe as the world's first interventional brain-computer interface (介入式脑机接口) trial to restore movement in a paralyzed human limb. No craniotomy. No electrodes driven into brain tissue. The implant traveled in by the same route a cardiac pacemaker lead does. ## How you install a BCI without touching the brain Most brain-computer interfaces (脑机接口) today sit at two extremes. Invasive systems — think Neuralink — place electrodes inside brain tissue: excellent signals, but they require opening the skull and carry risks of inflammation and long-term damage. Non-invasive caps read signals through hair, scalp and skull: safe, but blurry, like hearing a conversation through a wall. Duan's team built a third path, and China's Ministry of Science and Technology now officially lists interventional BCI as a category alongside the other two. In the trial, performed at Fujian Sanbo Funeng Brain Hospital with neurosurgeon Lin Zhixiong (林志雄) and collaborators from Fujian Second People's Hospital, doctors made a tiny puncture in the patient's neck vein. Under high-precision angiographic imaging, they threaded a stent-style electrode — just 2 millimeters across — into a blood vessel lying against the brain's motor cortex. A wireless power and data unit, roughly the size of a biscuit, was implanted under the skin near the chest. The vessel wall does the protecting. The electrode listens from inside it, close enough to the cortex to capture clean signals, while never penetrating brain tissue itself. ## Closing the loop from brain to muscle Reading the brain is only half the job. In a stroke patient, the "command center" may be intact while the line to the muscles is severed. The system pairs the implant with functional electrical stimulation (功能性电刺激) — external current that makes the paralyzed muscles contract on command. The patient imagines moving; the implant decodes the intent; the stimulator fires the muscles at the right moment. Researchers call this a "central–peripheral–central" closed loop, and it matters beyond the immediate trick of moving a hand: pairing intent with movement is what drives neuroplasticity, encouraging the brain to rebuild its own damaged circuits. According to an August 2025 report in People's Daily, the patient's signal transmission stability reached 99.7 percent, and his grip strength recovered to 82 percent of his healthy side within weeks — enough to grasp freely and pick up medicine bottles. Per a China Youth Daily report, the device stayed in his body for over ten days with no infection or thrombosis, and was then safely retrieved. ## Why the vein route may scale Three practical reasons: - **Surgery risk drops.** Interventional neurosurgery is a mature discipline; threading catheters through vessels is what stroke doctors already do daily. The learning curve for hospitals is far shallower than for open-skull implantation. - **It is reversible.** In 2024 the team showed a sensor could be completely retrieved from an animal's brain vessels without rejection — something hard to promise with tissue-embedded electrodes. - **Power lasts.** The wireless platform was designed for over five years of operation, reducing replacement surgeries. The route was earned step by step: a sheep trial in 2022, the world's first non-human primate interventional BCI in 2023 — a macaque feeding itself with a robotic arm driven by thought — the sensor-retrieval trial in 2024, then the first human implantation, and finally the restoration trial in June 2025. The closest Western comparison is Synchron, the US company whose "stentrode" also enters via blood vessels. China's team says its system, sensors through decoding algorithms to wireless power, carries fully independent intellectual property, and it has partnered with neuro-interventional device maker Xinwei Medical on hardware. ## What happens next The team has started multi-center clinical trials and, in July 2025, co-founded an Interventional Brain-Machine Information Technology Research Institute in Nantong with the Yangtze Delta National Technology Innovation Center, aimed at turning the technique into a platform rather than a one-off. Target populations extend beyond stroke to spinal cord injury, Parkinson's disease and ALS (渐冻症). The honest significance is this: of the three BCI routes, interventional may be the one that general hospitals can actually deliver. An implant that arrives without craniotomy, works reversibly, and uses tools the cath lab already owns is far easier for a national health system to scale — which is exactly what a country with tens of millions of stroke survivors needs. ## Honest limitations - This was a **first-in-human trial with essentially one reported subject**. Safety and efficacy across many patients, stroke types and recovery windows remain unproven; multi-center trials are only beginning. - The long-term behavior of an electrode held against a vessel wall for **years — tissue response, signal drift, wear** — cannot be known from a ten-plus-day implant window. The five-year battery life is a design figure, not an observed one. - Grip strength "82 percent of the healthy side" reflects **the reported trial window**, not necessarily a permanent recovery; the therapy depends on continued rehabilitation. - The 99.7 percent signal-stability and 82 percent figures come from **team members quoted by state media**, not yet from a peer-reviewed publication. Independent clinical data should settle the question. ## What readers can do now - **Stroke patients and families:** ask rehabilitation departments whether their hospital participates in BCI clinical research — the team notes the best recovery window is within about six months of onset, so timing matters. - **Follow the regulatory trail, not the hype:** China's national health insurer has begun creating dedicated pricing categories for BCI procedures; when a therapy gets a billing code, hospitals can finally procure it. - **Compare the three routes on reversibility and surgical burden,** not bandwidth headlines alone — for a patient, "can it come back out" is as important as "how many channels." --- Published by NeuroAI (https://neuroai.site/) — https://neuroai.site/a/na-nankai-vein-delivered-bci-stroke Free to quote with attribution and a link to the original.