High-Throughput Image Alignment for Connectomics using Frugal Snap Judgments
Accurate and computationally efficient image alignment is a vital step in scientific efforts to understand the structure of the brain through electron microscopy images of neurological tissue. Connectomics is an emerging area of neurobiology that uses cutting edge machine learning and image processing algorithms to extract brain connectivity graphs from electron microscopy images.
This poster abstract describes a high-throughput image alignment system that is designed to run on commodity multicore machines. We achieve alignments that differ by no more than 2 pixels from state-of-the-art alignment pipelines while achieving 40x additional throughput. The tools employed to achieve this performance boost include application-specific optimizations and the use of frugal snap judgments: a general purpose technique for optimizing computations that have strict accuracy–performance trade-offs.