FocA: A deep learning tool for reliable, near-real-time imaging focus analysis in automated cell assay pipelines

• DOI: 10.1016/j.slasd.2023.08.004
• Publication Date: 2023-08-11T01:48:30Z
• Authors: Jeff Winchell; Gabriel Comolet; Geoff Buckley-Herd; Dillion Hutson; Neeloy Bose; Daniel Paull; Bianca Migliori
• Keywords: Automated method

The increasing use of automation in cellular assays and cell culture presents significant opportunities to enhance the scale throughput imaging assays, but do so, reliable data quality consistency are critical. Realizing full potential will thus require design robust analysis pipelines that span entire workflow question. Here we present FocA, a deep learning tool that, near real-time, identifies in-focus out-of-focus images generated on fully automated biology research platform, NYSCF Global Stem Cell Array®. is trained small patches downsampled maximize computational efficiency without compromising accuracy, optimized make sure no sub-quality stored used downstream analyses. automatically generates balanced maximally diverse training sets avoid bias. resulting model correctly 100% 98% under 4 s per 96-well plate, achieves this result even heavily (∼30 times smaller than native resolution). Integrating into workflows minimizes need for human verification as well collection usage low-quality data. FocA offers solution ensure image hygiene improve using minimal resources.