Computational methods for faster, more accurate, and more versatile segmentation software for serial electron-microscopic analysis

Connectomics—the comprehensive mapping of synaptic circuits—is poised to revolutionize our understanding of the structural basis of neural information processing. Technical advances have supported the generation of serial electron microscope (SEM) images of tissue volumes at nanoscale resolution that span hundreds of microns across.

A comprehensive mapping of these volumes can advance our understanding of neural computations by allowing to identify synaptic associations among specific classes of neurons. Despite its great promise, SEM has yet to gain widespread adoption among neuroscientists. One of the most prohibitive barriers is that full reconstruction of specific microcircuits has only been possible by large teams of human annotators that spend thousands of hours to manually reconstruct individual cells and map their synapses.

A team of Carney-affiliated researchers is developing a novel analysis pipeline for automated segmentation of SEM volume into component neurons and for identification of synapses—allowing us to mine these invaluable but almost unmanageably massive data sets.

Research Leads

  • Thomas Serre

    Associate Director of the Center for Computational Brain Science, Director for the Center for Computation and Visualization, Professor of Cognitive, Linguistic and Psychological Sciences, Professor of Computer Science

    Understanding the neural computations supporting visual perception

  • David Berson

    Sidney A. Fox and Dorothea Doctors Fox Professor of Ophthalmology and Visual Science, Chair of Neuroscience

    Visual information processing in retinal cells and circuits