萨姆·奥尔特曼（Sam Altman）支持的合并实验室（Merge Labs）在构建由超声波驱动的脑机接口的竞赛中有了一个新的竞争对手。Bridge Neurotech将获得1350万美元的资金，其中包括风险投资公司General Catalys，该公司还支持Anduril、Anthropic、Airbnb和Stripe等主要科技公司。
Sam Altman-backed Merge Labs has a new competitor in the race to build a brain-computer interface powered by ultrasound. Bridge Neurotech is launching with $13.5 million in funding, including from venture firm General Catalyst, which has also backed major tech companies such as Anduril, Anthropic, Airbnb, and Stripe.
该公司是一个不断发展的初创公司领域中的最新一家，旨在将来自大脑的信号转化为有用的行动。埃隆·马斯克（Elon Musk）的Neuralink以及Synchron、Paradromics和Precise Neuroscience都在研究脑机接口，以帮助失去行动或说话能力的人。
The company is the latest in a growing field of startups that aim to translate signals from the brain into useful actions. Elon Musk’s Neuralink, plus Synchron, Paradromics, and Precision Neuroscience are all working on brain-computer interfaces to help people who have lost the ability to move or speak.
这些方法都需要进行某种手术，这就是为什么Bridge创始人兼首席执行官威尔·比德曼（Will Biederman）表示，他是自己尝试的。“这从根本上来说取决于愿景的差异，”他说。
These approaches all require surgery of some kind, which is why Bridge founder and CEO Will Biederman says he struck out on his own. “It fundamentally comes down to a difference in vision,” he says.
作为他博士学位工作的一部分，Biederman开发了Neuralink获准在其第一批设备中使用的芯片，但他对Bridge的目标是建造一种可穿戴设备，可以供比大脑植入物更多的人使用。作为临床试验的一部分，世界各地只有大约100人接受了植入式脑机接口，其中至少有21名志愿者接受了Neuralink的设备。
Biederman developed the chip Neuralink licensed to use in its first devices as part of his PhD work, but his goal with Bridge is to build a wearable device that could be used by far more people than a brain implant. Only 100 or so people around the world have received an implanted brain-computer interface as part of clinical trials, including at least 21 volunteers who have gotten Neuralink’s device.
比德曼说：“神经外科手术的风险确实无法最小化或忽视，因此这些设备必须提供的安全性和益处水平确实相当高。”“没有人真的希望自己的头上有一个洞。”但可穿戴脑机接口的信号质量较差。骨骼和皮肤在电信号到达外部传感器之前会削弱电信号，从而影响设备的性能。这就是为什么大多数公司直接从大脑神经元记录电脉冲。Neuralink和其他人已经证明，这些电气系统可以让瘫痪的人能够控制电子设备、移动机器人手臂并仅通过他们的想法进行实时沟通。
“The risks of neurosurgery really cannot be minimized or dismissed, and so the level of safety and benefit that these devices must deliver really is quite high,” Biederman says. “Nobody really wants a hole in their head.”But wearable brain-computer interfaces have suffered from poor signal quality. Bone and skin weaken electrical signals before they reach external sensors, which affects the performance of a device. It’s why most companies are recording electrical impulses directly from neurons in the brain. Neuralink and others have shown that these electrical systems can allow paralyzed people the ability to control electronic devices, move robotic arms, and communicate in real time with just their thoughts.
Bridge Neurotech认为超声波提供了当前植入物和外部传感器的替代方案。超声波使用高频声波从移动的红细胞上反弹，并测量血流的速度和方向。在大脑中，血流的波动密切跟随神经活动。
Bridge Neurotech thinks ultrasound offers an alternative to current implants and external sensors. Ultrasound uses high-frequency sound waves to bounce off moving red blood cells and measure the speed and direction of blood flow. In the brain, fluctuations in blood flow closely follow neural activity.
Bridge Neurotech和Merge Labs都与Arbor Neuroscience有联系，Arbor Neuroscience以前被称为Forest Neurotech，这是一家非营利组织，由Biederman及其同事Sumner Norman和Tyson Aflalo于2023年创立，旨在从事“对于单一学术实验室来说过于综合性和长期性的工作，而对于一家公司来说则为时过早”。该组织建造了一种使用超声波对大脑功能进行成像的仪器，目前正在英国进行试验。
Both Bridge Neurotech and Merge Labs have ties to Arbor Neuroscience, which was previously known as Forest Neurotech, a nonprofit founded in 2023 by Biederman and fellow researchers Sumner Norman and Tyson Aflalo to pursue “work that is too integrated and long-horizon for a single academic lab, and too early for a company.”The organization built an instrument that uses ultrasound to image brain function that is in trials in the UK.
Merge Labs于一月份从Forest Neurotech分拆出来，诺曼和阿弗拉洛是该公司的联合创始人，其中还包括奥尔特曼、研究员米哈伊尔·夏皮罗以及科技企业家亚历克斯·布兰尼亚和桑德罗·赫比格。
Merge Labs spun out of Forest Neurotech in January, with Norman and Aflalo among the company’s cofounders, which also include Altman, researcher Mikhail Shapiro, and tech entrepreneurs Alex Blania and Sandro Herbig.
南加州大学神经工程师玛丽亚姆·沙内奇（Maryam Shanechi）表示，由于超声波通过血液间接测量大脑活动，而不是直接从神经元测量大脑活动，因此在神经元放电和随后的血流量和体积变化之间存在几秒的自然延迟。
Maryam Shanechi, a neuroengineer at the University of Southern California, says because ultrasound measures brain activity indirectly from blood instead of directly from neurons, there is a natural delay of a few seconds between when neurons fire and the subsequent change in blood flow and volume.
“它捕捉到神经活动中较慢的变化，”沙内奇说。“这对于很多应用程序来说都可以，但这取决于您想要构建什么。”对于实时、快节奏的任务，例如立即移动机器人肢体或快速打字，这种滞后时间可能会对基于超声的脑机接口构成挑战。Bridge计划使用预测人工智能模型来减少滞后时间并提高其界面的性能。比德曼表示，该公司有兴趣开发消费者和医疗应用，但没有对具体计划发表评论。
“It captures a slower variation in neural activity,” Shanechi says. “That’s fine for a lot of applications, but it depends what you want to build.”For real-time, fast-paced tasks such as moving a robotic limb instantly or rapid typing, this lag time could pose a challenge for ultrasound-based brain-computer interfaces. Bridge is planning to use predictive AI models to cut down on that lag time and boost the performance of its interface. Biederman says the company is interested in pursuing both consumer and medical applications but didn’t comment on specific plans.
该公司正在加利福尼亚州雷德伍德城开始对志愿者进行首次临床研究。这是一项观察性研究，不涉及治疗或诊断，其中将一个小型超声波探头放置在健康成年人以及接受过颅骨手术的人的头部侧面。布里奇的探头使用便携式超声公司Butterfly Network生产的半导体硅芯片，将在参与者听、看、动和说话时记录大脑活动。
The company is starting its first clinical study in volunteers in Redwood City, California. It’s an observational study, with no treatments or diagnoses involved, in which a small ultrasound probe will be placed on the side of the head of healthy adults, as well as those who have undergone skull surgery. Bridge’s probes, which use semiconductor silicon chips made by the portable ultrasound company Butterfly Network, will record brain activity while participants listen, watch, move, and speak.
Bridge尚未发布其设备的图像，但Biederman表示，它是“接近耳机的东西”，将位于位于耳朵上方的颞窗上。作为头骨最薄的部分，“它是最容易成像的地方，”比德曼说。这与超声波相关，因为骨骼会散射和吸收声音，这给通过头骨进行无创成像带来了技术挑战。
Bridge isn’t releasing images of its device yet, but Biederman says it is “something close to headphones” that will sit on the temporal window, which is located slightly above the ear. As the thinnest part of the skull, “it's the easiest place to image through,” Biederman says. That’s relevant for ultrasound, because bone scatters and absorbs sound, which presents technical challenges for imaging noninvasively through the skull.
在2024年的一项研究中，加州理工学院和南加州大学的研究人员证明，基于超声的脑机接口可以读取和解码人脑活动，但它需要切除头骨的一部分并用一个“窗口”取代超声波可以通过的“窗口”。这是Bridge必须解决的最大技术障碍之一。
In a 2024 study, researchers from Caltech and the University of Southern California demonstrated that an ultrasound-based brain-computer interface could read and decode human brain activity, but it required a portion of the skull to be removed and replaced with a “window” through which ultrasound waves can pass. That is one of the biggest technical hurdles that Bridge will have to address.
“我的愿景专注于非侵入性界面，”比德曼说。“我认为这是帮助人们的最大机会。”
“My vision is focused on noninvasive interfaces,” Biederman says. “I think that's where the biggest opportunity to help people is.”