Deriving Neurological Connectomes to Better Understand Complex Brain Functionality

The quest to unlock the mysteries of the brain has stimulated vast initiatives to map brain activity and create neuromorphic chips capable of emulating large neural networks. While promises of sophisticated artificial intelligence systems and cures for neurological diseases excite the imagination of many, the reality is that we struggle to understand the dynamics and functionality of even small organisms with a few hundred neurons. Simply put, acquiring and analyzing the data needed to understand neurological function is time-consuming and difficult. We focus on one avenue to partially characterize a neurological system by efficiently determining a neurological connectome, i.e., a set of neurons and their connections, from microscopic image data. In particular, we will highlight the imaging and computational techniques used to extract a meaningful connectome in the fly visual system and a resulting circuit motif that leads to a better understanding of how motion is detected. Further advances in imaging, computer vision, and circuit analysis promise to make much larger systems tractable and, in the near future, provide templates for certain types of behavior that might manifest in low-power, robust VLSI implementations. The presentation will cover material published in A visual motion detection circuit suggested by Drosophila connectomics. Nature 2013, 500: 175-181.

Stephen Plaza graduated Summa Cum Laude with a BSE in computer engineering in 2003 and received his PhD in 2008 in computer science and engineering at the University of Michigan. His primary area of focus was transistor and wiring optimization in computer chips. After completing his degrees at Michigan, he spent over a year working at Synopsys Inc in its Advanced Technology Group and Implementation Group devising and implementing algorithms used to design computer circuitry.

In 2010, Stephen decided to combine his experience in computer programming, electrical circuits, logic simulation, and parallel computation and algorithms at Janelia by working under Lou Scheffer and for the Fly EM project. His primary area of research is introducing and deploying strategies that enable high-throughput verification and correction of the neurological connectome in an EM dataset using both automatic computation and semi-automatic procedures. In early 2012, Stephen became the project manager for Fly EM. His main goals are to explicitly define and monitor all project requirements and encourage greater synergy between the large, inter-disciplinary team that composes FlyEM. To aid in these goals, Stephen has completed certificates at Georgetown for Project Management and Budget and Finance. Steve continues to conduct new research in addition to managing and acts as a technical lead for several software developers and researchers.