Abstract Brain-machine interfaces (BMIs) are one of the key motivating applications for the BRAIN Initiative’s drive to develop innovative technologies for large-scale recording of neural activity, benefiting not only BMI, but many other neuroscience studies. The most advanced techniques for neural recording and BMIs are currently invasive, causing local damage to living brain tissue, limiting their applications in human neuroscience research and BMI. On the other hand, noninvasive techniques typically offer relatively low spatial resolution and sensitivity. A minimally invasive BMI would bridge the gap between these extremes, opening a new avenue for neuroscience research and neuroprosthetics. Recently, functional ultrasound (fUS) imaging was introduced as a breakthrough technology for large-scale recording of neural activity – providing highly sensitive imaging of activity-dependent changes in blood flow with a spatiotemporal resolution of ~100 µm and 100 ms at several-cm depth. Importantly, fUS can record from outside the brain and protective dura mater tissue, vastly expanding its potential use in neuroscience applications and BMIs alike. While fUS is a hemodynamic technique, its excellent spatiotemporal performance and single-trial sensitivity offer a substantially closer connection to the underlying neuronal signals than achievable with other hemodynamic methods such as fMRI. In this project, we will push the boundaries of fUS as a technology for large-scale recording of neural activity by developing a fUS-based minimally invasive BMI. This proposal is based on preliminary data acquired by the collaborating investigators showing that ultrafast fUS imaging of the posterior parietal cortex in non-human primates (NHP) provides sufficient information to predict planned movements from single trial fUS recordings. These remarkable findings suggest that it may be possible to use fUS as the basis for a minimally invasive BMI that is implanted in the skull and does not penetrate the dura or brain tissue. Turning this potential into reality requires several fundamental advances in fUS neural imaging technology, which will greatly enhance the utility of this large-scale neural imaging technique across neuroscience applications. These advances include (1) maximizing the speed, data processing and information content extracted from fUS to enable a high- performance, real-time acute BMI; (2) developing a surgically implantable fUS technology for chronic, longitudinal minimally invasive recording of neural activity from a specific brain region; and (3) extending the fUS technology from 2D to 3D to facilitate applications requiring real-time imaging of large brain volumes. This proposal is enabled by two key innovations made by the co-PIs: the invention of fUS by Tanter, and the discovery by the collaborative team of Andersen, Shapiro and Tanter that fUS signals contain information that can be used for BMI. In addition, several new innovations are introduced through the proposed work including techniques to acquire and process fUS data in real time, advances in fUS hardware and surgical techniques for chronic implantation, and advances to enable wide-field and sparse real-time 3D imaging. If successful, this project will significantly advance the capabilities of fUS as a widely useful technology for rapid, sensitive, large-scale neural imaging and enable minimally invasive BMI.

Project Narrative The most advanced techniques for large-scale neural recording and brain-machine interfaces (BMIs) are currently invasive, causing damage to living brain tissue, and inviting significant risk for infection. The proposed work describes the development and implementation of the first BMI based on functional ultrasound (fUS) imaging - a recently developed minimally invasive (epidural) neuroimaging technique combining high spatial and temporal resolution with deep brain coverage. In this project, we will push the limits of fUS imaging as a technology for online large-scale chronic recording of neural activity by developing minimally invasive and implantable ultrasonic BMIs, creating new opportunities for neuroprosthetics and fundamental neuroscientific inquiry.