Pattern Detection at Low Signal-To-Noise Ratio

Robust detection of patterns at low signal-to-noise ratios (SNR) is a fundamental challenge of analyzing imaging data, particularly in biological imaging. Harvard researchers have developed innovative computational algorithms for detecting and characterizing multiple patterns of interest (e.g. different colored fluorophores) in low SNR data sets. Typical imaging methods often require a high SNR, which can result in photobleaching or cytotoxic effects and therefore limit the duration of an experiment. This invention allows for detection of multiple fluorophores from images taken at a low SNR with high spatial resolution, thus allowing for longer experiments in 3D over biologically-relevant time scales. These methods could also be applied towards improving the maximum frequency of imaging in sequencing reactions. By enabling fluorescent spots to overlap, the maximum density of measurements (e.g. DNA reactions) per image is increased, thus improving throughput. These methods have the potential to greatly improve the capabilities of current imaging hardware and be highly impactful for biological research.

U.S. Patent(s) Issued: WO2017/040669