An illustration of nerves in the brain based on data from magnetic resonance imaging.Credit: Thom Leach/Science Photo Library
Brain implants are beginning to help people with severe disabilities to speak and even sing in near-real time. Now, a company wants to read people’s minds and treat mental conditions without implanting electrodes deep into the brain by using ultrasound — high-frequency sound waves above the range of human hearing.
Merge Labs, which launched last month with only a vague description of its goals, is one of many companies in a booming brain–computer interface (BCI) market. What makes it stand out is US$252 million in investment from funders that include artificial-intelligence firm OpenAI, based in San Francisco, California. The start-up is being billed as a rival to Elon Musk’s Neuralink, which makes devices that detect and manipulate electrical activity in the brain and are already being trialled in patients.
Nature asked researchers who the people are behind Merge Labs and whether the company’s approach is based on solid science.
What is Merge Labs?
The for-profit company bills itself a research laboratory, rather than a firm focused on a rapid return on investment. It was spun out of the non-profit research organization Forest Neurotech, based in Los Angeles, California.
Merge Lab’s co-founders include three researchers: Forest Neurotech’s chief scientific officer Tyson Aflalo and chief executive Sumner Norman and Mikhail Shapiro, a BCI researcher at the California Institute of Technology in Pasadena and Forest Neurotech adviser. Other co-founders are tech entrepreneurs Alex Blania and Sandro Herbig and OpenAI’s chief executive Sam Altman.
How will the company’s approach to BCIs differ from Neuralink’s?
Merge Labs seems to be developing ultrasound techniques to both image and modulate brain activity. The approach aims to be much less invasive than Neuralink-style devices, because it inserts sensors either just underneath the skull or operates through a window in the bone, rather than deep in the brain. Whereas electrical devices are fixed and can interface only where electrodes are implanted, ultrasound waves can monitor very large areas of the brain and stimulate multiple sites, which could help to treat multifaceted disorders such as depression, says Elsa Fouragnan, a neuroscientist at the University of Plymouth, UK, who collaborates with Forest Neurotech. Like Neuralink, Merge Labs looks set to use AI to decode brain activity.
How does ultrasound interact with the brain?
Conventional ultrasound imaging works like sonar, bouncing waves off tissue to build up an internal picture of the body. Functional ultrasound is more complex: it analyses how the frequency and amplitude of the returning ultrasound change as the waves scatter off moving objects, to detect the movement of blood cells and estimate flow volume. When neurons are very active, they need more oxygen. This drives changes in blood flow that reveal brain activity, says Fouragnan. “It will create a map that looks like red when there is activity and nothing if there is no activity,” she says.
Ultrasound can also be used to stimulate neurons. When multiple beams are focused on one spot, the waves change the pressure around the neurons, altering their firing rate. Merge Labs has hinted at the possibility of combining this focal ultrasound with a more speculative approach known as sonogenetics, which uses genetic engineering to make specific cells even more responsive to the waves.
What are the disadvantages of using ultrasound?
Although less invasive than in-brain devices, the method still requires surgery to get beneath the skull. And although ultrasound can read the brain with a high spatial resolution (of around 0.2 millimetres), the method is relatively slow, because blood flow is an indirect measure of brain activity and comes with a lag, says Giacomo Valle, a neurotechnology researcher at Chalmers University of Technology in Gothenburg, Sweden.
“If the goal is interactive BCI, where the system responds fast enough to feel connected to intention” of a patient — for example to decode their speech — then blood-flow-based methods face “a fundamental constraint”, says Dimitrios Adamos, a neuroinformatician at Imperial College London and a co-founder of BCI firm Cogitat.
For what kinds of uses does the technique show potential?
Despite the challenges of using ultrasound to interact with the brain in real time, Merge Labs is pursuing the approach to build BCIs. Researchers at the company have used an ultrasound device to interpret the intended movements of monkeys1 and to detect the brain activity behind human actions such as strumming a guitar and playing a video game2.
Ultrasound holds promise for other forms of therapy beyond BCIs. Such devices could be a less invasive and more flexible alternative to electrical deep-brain stimulation to treat epilepsy and conditions that involve multiple sites in the brain, such as tinnitus, severe depression, addiction and eating disorders, says Fouragnan. Under-skull devices would also offer substantial advantages over existing efforts to use focused ultrasound to treat neurological devices from outside the head.
