FAST-ET: Fast Automated Serial Tape-Enabled Electron Tomography

This project aims to develop a solution for high-throughput brain connectomics by integrating GridTape-based automated section collection and handling with advanced serial electron tomography (ET). ET is an indispensable tool for ultra-high resolution imaging, typically used to study very small regions in detail by computationally reconstructing 3D volumes from 2D projections. Using intermediate voltage transmission electron microscopy (IVEM), ET generates volumetric information through thick sections (up to 2 μm). By combining ET of thick sections with automated section collection and handling, we aim to significantly advance whole mouse brain connectomic imaging in terms of robustness and practicality, improving upon capabilities for high- and multi-resolution imaging for mammalian brains, and more efficiently elucidating the organizational principles of brain circuits. Our approach will further facilitate sample-preserving imaging, allowing for multi-scale interrogations without sample destruction, thereby significantly reducing the risks associated with data acquisition and amplifying the value of prepared samples. The efficacy of this solution will be validated using the mouse cerebellum, providing insights into cell-to-cell communication within this brain region crucial for motor control. 

The project will be driven by three complementary aims structured to: 1) Evaluate ET scalability by assessing imaging conditions, lossless sectioning, and GridTape characteristics, 2) Develop tape-enabled serial ET through the design and testing of a reel-to-reel tape system, a tape tilting system, and related software, and 3) Augment tape-enabled ET with advanced machine learning algorithms, energy-filtered TEM strategies (e.g., automated most-probable loss tomography (MPL)), continuous imaging, and modifications for large section tomography. Deliverables from this work will include the development, testing, and troubleshooting of prototype FAST-ET platform, critical reference datasets for the neuroscience community, and scalable metrics for mammalian whole-brain connectomics. Our goal is to significantly advance understanding of brain function and dysfunction with broad implications for neuroscience, engineering, and artificial intelligence.