Research Infrastructure: Mid-scale RI-1 (M1:DA): National Connectomics and Ultrastructure Volume Observatory (NCUVO)

Understanding how living systems process, store, and use information is one of the fundamental challenges in biology. Progress in this area requires new ways to observe biological structures across scales and to integrate many types of data. Recent advances in high-throughput volume electron microscopy (htvEM) have made it possible to create detailed three-dimensional maps of brain circuits, known as "connectomes," that reveal the connections (called synapses) between individual neurons. To date, connectomes have been generated for the complete brains of small animals like the fruit fly and for small portions of mammalian brains, including human brain. This work has had a major impact on neuroscience and may lead to more flexible, capable, and energy-efficient artificial intelligence (AI) architectures. However, htvEM technology has only been accessible to a handful of laboratories, which have mostly focused on generating ever larger image volumes from single samples. This is due to the highly specialized expertise, infrastructure, and cost required to prepare and image high-quality samples and analyze the enormous datasets they produce, often containing millions of image volumes. This project will develop and validate the design for a National Connectomics and Ultrastructure Volume Observatory (NCUVO), national research infrastructure that will make advanced htvEM imaging capabilities broadly accessible to the U.S. scientific community. By enabling researchers across the country to study many samples in parallel rather than focusing only on single large datasets, NCUVO will support comparisons across species, brain regions, developmental stages, experimental conditions, and time points. The Observatory will enable a broad range of research applications in biological information processing, including connectomics as well as the multiscale structural basis of information processing at tissue, cellular, and subcellular levels. This national resource will strengthen the U.S. research enterprise and accelerate discovery in the life sciences. The NCUVO design will span the complete htvEM pipeline, from sample preparation and high-resolution imaging to volume alignment, AI-based automated image segmentation, and software for scalable annotation, analysis, and visualization. The design will be validated using several pilot projects representing diverse biological research applications and domains. Data processing workflows will leverage national data, networking, and compute infrastructure, including the NSF National Data Platform and collaboration with the San Diego Supercomputer Center. Sample preparation and multimodal imaging workflows will also be distributed, leveraging the investigators' complementary expertise as well as that of pilot project collaborating laboratories. Pilot data, protocols, and workflow documentation will be shared according to FAIR (Findable, Accessible, Interoperable, Reusable) principles. The project will also develop training resources that will be incorporated into undergraduate and high-school curricula, and for graduate workforce training in htvEM and data management. The primary outcome will be a validated and scalable pipeline design for a national htvEM observatory, providing the technical framework for a subsequent implementation phase, enabling broad community access to multimodal, multiscale biological imaging capabilities. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.