--- title: "China Tried a Third Route for Brain Implants — Through the Blood Vessels, Not the Skull" date: 2026-10-21 category: Brain–Computer Interface site: NeuroAI canonical: https://neuroai.site/a/na-brain-interventional-bci-nankai language: en --- # China Tried a Third Route for Brain Implants — Through the Blood Vessels, Not the Skull > Nankai University's Duan Feng (段峰) team ran the first human interventional brain-computer interface (介入式脑机接口) in 2025, threading a 2mm stent-electrode via neck veins to restore a stroke patient's grip — with no craniotomy at all. The surgeon never opened the skull. Instead, a thin electrode支架 slipped up through a blood vessel in the neck and parked against the wall of a vessel beside the motor cortex. Weeks later, a man who had not moved his left arm in half a year picked up a cup. This is the "third route" of brain-computer interface (脑机接口) — neither the deep invasive implant nor the skull-cap EEG, but a stent placed through the vasculature. China just put it in a human. ## The route that avoids the bone Brain interfaces are usually sorted into two camps. Invasive (侵入式) systems push electrodes into brain tissue for the sharpest signal, at the cost of craniotomy and long-term tissue risk. Non-invasive (非侵入式) systems read blurred signal through the skull, safe but low-resolution. The interventional route (介入式脑机接口) sits between them: a catheter carries a stent-electrode to a vessel next to the target brain region, where it records local field potentials through the vessel wall. No opening of the skull; the procedure resembles placing a vascular stent. That single property reshapes who can do it. Interventional radiologists already place stents every day. The skill transfers. A brain interface that rides on a mature, widespread clinical procedure is, in principle, far easier to scale than one that needs a neurosurgeon's hands. ## The first human case In 2025, a team led by Professor Duan Feng (段峰) at Nankai University ran what Chinese state media described as the first human interventional brain-computer interface trial to assist motor-function recovery. The subject was a **67-year-old man** whose left side had been paralyzed for half a year after a cerebral infarction (脑梗), with conventional treatment offering little hope. Under high-precision DSA imaging guidance, the team delivered a **2-millimetre stent-electrode** through a minimally invasive neck-vessel approach to the wall of the relevant intracranial vessel, and implanted a wireless power-and-transmission module under the skin. The result was a closed "brain–machine–limb" loop. The reported outcomes: - Within **two weeks**, the patient could grasp by intention. - Signal transmission stability reached **99.7%**. - Grip strength recovered to **82%** of the healthy side. To strengthen the effect, the team paired the interface with functional electrical stimulation in a "central–peripheral–central (中枢-外周-中枢)" closed loop, aiming to drive neuroplasticity rather than merely route around the break. The procedure produced no reported infection or thrombosis, and the system ran stably. ## The long runway behind one patient The 2025 human case was the visible tip of a multi-year animal program, laid out in People's Daily's coverage: - **2022:** sheep-brain interventional experiment. - **2023:** non-human-primate experiment. - **2024:** demonstration of complete retrievable removal of an in-brain sensor — the ability to take the device back out, a key safety property for any implant. In July 2025, Duan's team established an Interventional Brain-Computer Information Technology Research Institute (介入脑机信息技术研究所) with the Yangtze River Delta National Innovation Center and Nantong Innovation District, to push from single experiments toward clinical validation and translation. The collaborators on the human work included Professor Lin Zhixiong (林志雄) of Sanbo (三博) neurosurgery and Professor Wu Chenghan (吴成翰) of Fujian Provincial Hospital. The CAS academician Pu Muming (蒲慕明) publicly noted the technique not only controls the limb but may activate latent neuroplasticity. ## Why a vessel matters more than a channel count Interventional brain-computer interface captures lower-resolution signal than a penetrating electrode — it reads the slow field potential through a vessel wall, not single neurons. Its advantage is the opposite axis: safety, reversibility, and fit with existing clinical practice. For the large populations who need help but would never qualify for, or accept, open-brain surgery — stroke survivors, some spinal-cord and ALS (渐冻症) patients — a no-craniotomy route may be the only one that reaches them. It is also not uniquely Chinese in concept; a U.S. company, Synchron, has run endovascular trials since 2020. The Chinese program's notable points are its independently developed stent-electrode and wireless platform, and the explicit "central–peripheral–central" rehabilitation framing that aims to rebuild function, not just bypass it. ## Honest limitations This is a single-subject case report, not a controlled trial; the 82% grip-strength figure describes one patient and should not be read as a population average. Interventional BCI captures lower-resolution signal than invasive electrodes, limiting the complexity of what can be decoded. Long-term vascular questions — endothelialization, thrombosis risk over years, signal decay as the vessel wall remodels — are unanswered. The "first human" and "first in China" claims are as reported by People's Daily and Science and Technology Daily; similar endovascular programs exist elsewhere and at other Chinese centres (including a Xuanwu Hospital line reported in 2026), so the milestone is one of several parallel efforts rather than a sole national first. The technique still requires multicenter validation before any regulatory approval. ## What readers can do now - If a family member has post-stroke paralysis, ask clinicians about **interventional and non-invasive neuro-rehab trials** — the no-craniotomy routes may be the only realistic options for many patients. - If you follow the field, track **retrievability and vascular safety data**, not channel counts; for a stent-based device, "can it come back out" is the question that decides approval. - If you evaluate neurotech, separate the three routes by **who can perform the procedure** — a device that needs only interventional radiology scales very differently from one needing neurosurgery. --- Published by NeuroAI (https://neuroai.site/) — https://neuroai.site/a/na-brain-interventional-bci-nankai Free to quote with attribution and a link to the original.