Blind Patients Poised for Visual Restoration: Chinese Brain-Computer Interface Breakthrough Advances Toward Clinical Application

Wen Wei Po, Hong Kong

Paralyzed patients have regained control over their limbs, enabling basic movements such as raising arms and grasping objects; individuals with aphasia have regained the ability to “speak,” breaking down communication barriers; and patients with neurological disorders are alleviating their symptoms through brain-computer interface (BCI) technology. As a significant innovation in human-machine integration, BCI technology is currently demonstrating its value in fields such as medical rehabilitation and assistive technologies for daily living.

As a cross-disciplinary frontier technology enabling information exchange between the brain and external devices, brain-computer interfaces (BCIs) have seen significant regulatory milestones. The National Medical Products Administration (NMPA) has previously approved the innovative product registration application for the implantable BCI hand motor function compensation system developed by Neuracle Technology (Shanghai) Co., Ltd., marking the world’s first market launch of a BCI medical device and signifying that the first invasive BCI medical device globally has entered the clinical application stage. Publicly available data indicate that China’s BCI industry is developing rapidly; multiple leading enterprises have announced that their invasive BCI products have entered clinical trial stages, while some developers of non-invasive BCIs are also exploring consumer-grade application scenarios.

First Case of Grayscale Information Visual Reconstruction Completed at the Beginning of the Year

In February this year, Mindtrix announced the world’s first grayscale visual reconstruction based on a brain-computer interface (BCI). “Our research team has successfully enabled subjects to identify numbers and simple geometric shapes by delivering precise electrical stimulation to the visual cortex via implanted electrodes. Currently, the company has initiated preparations for an Investigator-Initiated Trial (IIT) involving totally blind volunteers. The goal is to help patients blinded by optic nerve damage regain contour perception, achieve basic obstacle avoidance, and acquire functional vision capable of recognizing large-font text and grayscale levels,” said Chen Wenkai, Chief Commercial Officer of Mindtrix, in an interview.

Chen Wenkai pointed out that although China’s brain-computer interface (BCI) industry is currently expanding across multiple scenarios—covering areas such as motor function rehabilitation, neuromodulation, and speech rehabilitation—it still predominantly focuses on unidirectional data decoding. In contrast, “we specialize in implantable BCI technology for visual reconstruction, the core of which lies in ‘writing’ encoded information into the brain, rather than merely reading signals.” Currently, the company’s independently developed “1024-channel large-scale stimulation implantable visual reconstruction system” encodes environmental information captured by an external camera into spatiotemporal electrical stimulation signals recognizable by the brain. These signals act directly on the visual cortex, simulating natural visual imaging mechanisms. This technology holds promise for repairing visual impairments caused by optic nerve damage—conditions that are difficult to treat with existing clinical methods—thereby restoring “functional vision” for individuals with blindness.

Expected to complete registration and initiate clinical application in 2031

According to the technological roadmap, Chen Wenkai revealed that Mindtrix will advance product finalization and clinical validation in phases. The company plans to launch this year’s first Investigator-Initiated Trial (IIT) in China—and potentially globally—for totally blind patients using a mid-to-low channel count implantable system, with clinical registration and application expected to be completed by 2031. Once the system is formally implemented, patients are anticipated to clearly recognize object contours following rehabilitation training, thereby meeting the needs for independent mobility, daily living, and basic social interaction, thus achieving true social integration. Furthermore, leveraging neural plasticity and incorporating a “brain-computer dual learning” mechanism, the system may enable personalized adaptive optimization over the long term, further enhancing visual perception quality. “We aim to become a global leader in the niche field of implantable visual reconstruction and drive advanced brain-computer interface technologies from the laboratory into widespread clinical accessibility.”

Meanwhile, in the field of sports and rehabilitation medicine, Jieti Medical has completed multiple clinical validations of invasive brain-computer interfaces (BCIs), successfully helping numerous patients achieve precise brain-controlled operations. The precision and stability of these systems have met the standards for clinical application, offering a novel rehabilitation solution for individuals with paralysis and motor function impairments. In an interview with reporters, a representative from Jieti Medical stated that the team is expanding the application scenarios of BCIs: “We will accelerate investigator-initiated trials (IITs) on neuromodulation for neurological disorders such as Parkinson’s disease and depression. Leveraging closed-loop stimulation technology, we aim to achieve precise intervention and treatment for neurological conditions, thereby broadening the disease applicability boundaries of BCI technology. Furthermore, building on our mature motor function reconstruction techniques, we will gradually tackle the challenge of language function reconstruction to address communication difficulties faced by individuals with aphasia, further filling the gaps in domestic clinical BCI applications.”

Prioritize focusing on high-level paraplegia and amyotrophic lateral sclerosis (ALS), among others.

The person in charge stated that, leveraging their independently controlled core technologies, the team has achieved ultra-high signal acquisition precision at the single-neuron level. This capability not only meets the demands of high-end medical rehabilitation but also holds the potential to revolutionize human-computer interaction models in the future, paving the way for new scenarios involving the deep integration of brain-computer interfaces (BCI) and artificial intelligence. “We will prioritize focusing on patient groups with disabilities, such as those with high-level paraplegia and amyotrophic lateral sclerosis (ALS), striving to address genuine and urgent clinical pain points. This approach represents the core value of BCI technology implementation and serves as a critical pathway for regulatory compliance review and iterative technical validation, thereby accumulating sufficient clinical data and safety experience to support subsequent expansion into consumer-grade scenarios such as digital health and civilian interactive applications.”

[Did You Know]

Acquisition of EEG Signals and Conversion into Machine Commands

Brain-Computer Interface (BCI) is a cutting-edge interdisciplinary technology that enables direct bidirectional communication between the brain and external devices. By bypassing peripheral organs such as the limbs and vocal cords, it acquires electroencephalographic (EEG) signals generated by neuronal discharges, which are then amplified, filtered, and decoded via artificial intelligence into machine commands. Conversely, it can write optoelectronic and tactile signals back into the brain, thereby enabling mind-controlled device operation and neural repair, helping patients with functional impairments restore their ability to communicate externally and control their limbs.

Based on different signal acquisition methods, brain-computer interfaces are mainly divided into three major categories:

1. Non-invasive brain-computer interfaces do not require craniotomy or intrusion into brain tissue; sensors are placed only on the external scalp. This approach is safe, non-invasive, and cost-effective, and is widely used in scientific research and assistive control applications;

2. Semi-invasive brain-computer interfaces implant electrodes between the skull and the cerebral cortex, without direct contact with brain tissue. This approach achieves a balance between signal quality and safety;

3. Invasive brain-computer interfaces involve the direct implantation of electrodes into the cerebral cortex via neurosurgical procedures. This approach enables the acquisition of high-quality neural signals and facilitates more precise control, but it carries surgical risks and poses long-term biocompatibility challenges.

Integrated Embodied Intelligence

Excellent Rehabilitation Training Outcomes

As AI endows robots with enhanced understanding and predictive capabilities, embodied intelligence is accelerating its transition from passive execution to active collaboration, while the integrated exploration of brain-computer interfaces (BCI) and embodied intelligence continues to advance. "The latest clinical consensus indicates that human-robot collaborative training models involving active participation yield better rehabilitation outcomes than passive, mechanical training. We previously initiated internal pre-research projects combining BCI with exoskeleton robots, successfully achieving EEG-driven exoskeleton locomotion and completing early proof-of-concept validation," said Yang Zhihao, Director of Rehabilitation Innovation Projects at Shanghai Fourier Intelligence Co., Ltd. This year, Fourier Intelligence further introduced BCI technology and launched the "BCI Embodied Intelligence Rehabilitation Hub" solution, establishing a closed-loop training mechanism of "intention–execution–sensory feedback," thereby driving a leap in rehabilitation training from passive functional compensation to active mind-controlled intervention.

Yang Zhihao introduced that in the field of rehabilitation medicine, brain-computer interface (BCI) technology can effectively compensate for the limitations of traditional rehabilitation in terms of early intervention, training efficacy, and assessment precision. On one hand, it is suitable for ultra-early rehabilitation training; intention-driven active rehabilitation training can be initiated even when voluntary limb movement is absent, thereby seizing the critical window for functional remodeling. On the other hand, by synchronously monitoring motor cortex activity and electroencephalographic (EEG) spectral changes, and integrating these with biomechanical data, BCI provides objective, continuous, and traceable quantitative evidence for training engagement and rehabilitation outcomes, thus maximizing the effective density of training.

Build Datasets and Refine the Technical Framework

Discussing the synergy between brain-computer interfaces (BCIs) and embodied intelligence, Yang Zhihao explained that both fundamentally rely on electrical signals for control, and the key to their integration lies in building specialized datasets. “Currently, Fourier Intelligence is collaborating with several rehabilitation centers and partner organizations to jointly construct a BCI dataset based on embodied intelligence. The core objective is to explore how to precisely analyze and decode human motor intentions, and integrate this with robotic data acquisition and intent recognition, thereby bridging the critical link in their synergy. This represents the core breakthrough point in the convergence of BCIs and embodied intelligence.” It is reported that Fourier Intelligence has previously open-sourced relevant robotic datasets, laying a solid foundation for technological R&D. The newly developed BCI dataset will further refine the technical framework for integrating these two domains.

Rehabilitation Medicine: From Experience to Data

It is reported that Fourier Intelligence has established over 300 rehabilitation departments across China. This collaborative exploration of brain-computer interfaces (BCI) and embodied intelligence will further expand the perceptual dimensions of the Embodied Intelligence Rehabilitation Hub, driving rehabilitation medicine from experience-based to data-driven practices. Fourier Intelligence, in conjunction with multiple research universities and institutions, has launched the “BCI-Embodied Intelligence Data Engine Joint Innovation Program.” Yang Zhihao stated, “We will focus on the development of embodied intelligence hardware, undertake interface adaptation between BCI and embodied intelligence, and provide underlying technical support such as accompanying toolchains. Through division of labor and collaboration with partners, we aim to explore the deep integration of brain-computer interfaces with embodied intelligent agents, facilitating a complete pathway from intent recognition to embodied execution.”

Fully Implantable and Fully Wireless

Patients Enjoy Greater Mobility

Reporters recently learned from BrainCo that, following the clinical implantation of its first independently developed fully implanted, fully wireless, and fully functional (“three-all”) brain-computer interface (BCI) in late 2025, the second clinical trial has achieved a key breakthrough. By deeply integrating 50-millisecond ultra-low latency across the entire link with BCI-driven functional electrical stimulation (BCI-FES) technology, the system enabled a 29-year-old art teacher with high-level paraplegia to independently eat and create paintings. This marks the official entry of this technology into a new stage of clinical practicality and replicable promotion, laying the foundation for large-scale application.

Latency < 50 ms

Reportedly, as the first brain-computer interface (BCI) system with a built-in battery in mainland China and the second globally, the "Sanquan" system achieves truly independent energy storage and operation, eliminating external cable constraints on the body surface. It can stably support daily use with short-duration wireless charging. Compared to traditional devices that rely on external power supplies, are constrained by wired connections, and fail immediately upon disconnection, the "Sanquan" BCI enables patients to move freely, aligning more closely with real-life scenarios. The system’s end-to-end latency is under 50 milliseconds, truly realizing “thought-to-action” immediacy: the moment a patient generates the intention to grasp a pen, their hand begins moving in synchrony. This performance metric has reached the international leading level for latency control in implantable brain-computer interfaces.

Achievement of Independent Mobility One Month Postoperatively

A representative from TaoHu Technology introduced that this clinical trial marks the first adoption of brain-computer interface-driven functional electrical stimulation (BCI-FES) technology. By deeply integrating the “Sanquan” system with functional electrical stimulation (FES) technology, a complete closed-loop pathway of “intention–decoding–stimulation–action” was established. This system precisely acquires and real-time decodes motor intentions via cortical electrodes, bypasses the damaged spinal cord, and delivers precise stimulation to peripheral muscles, thereby enabling natural and coordinated finger movements. Clinical validation demonstrated that one month post-surgery, the patient achieved precise, active grasping of the affected hand through intention-based control, and could independently perform high-precision tasks such as writing and drawing, truly returning the control target to the patient’s own limbs.

Tao Hu, Founder and Chief Scientist of NeuroXess, stated, “In the first clinical case, we demonstrated feasibility; in the second, we proved efficacy and scalability. From controlling external devices to driving one’s own limbs, each iteration of our technology has been closely aligned with urgent clinical needs. Our goal is not merely to restore movement for patients, but to enable them to return to life with quality and dignity.”

Reporter: Ni Mengjing

Newspaper Editor:Zhuo Xian

Online Editor: Lu Yongjun

Past Issues Review

REVIEW

Recommended Follows