Highly Flexible Deep-Learning-Based Automatic Analysis for Graphically Encoded Hydrogel Microparticles

초록

Graphically encoded hydrogel microparticle (HMP)-basedbioassayis a diagnostic tool characterized by exceptional multiplex detectabilityand robust sensitivity and specificity. Specifically, deep learningenables highly fast and accurate analyses of HMPs with diverse graphicalcodes. However, previous related studies have found the use of plainparticles as data to be disadvantageous for accurate analyses of HMPsloaded with functional nanomaterials. Furthermore, the manual dataannotation method used in existing approaches is highly labor-intensiveand time-consuming. In this study, we present an efficient deep-learning-basedanalysis of encoded HMPs with diverse graphical codes and functionalnanomaterials, utilizing the auto-annotation and synthetic data mixingmethods for model training. The auto-annotation enhanced the throughputof dataset preparation up to 0.11 s/image. Using synthetic data mixing,a mean average precision of 0.88 was achieved in the analysis of encodedHMPs with magnetic nanoparticles, representing an approximately twofoldimprovement over the standard method. To evaluate the practical applicabilityof the proposed automatic analysis strategy, a single-image analysiswas performed after the triplex immunoassay for the preeclampsia-relatedprotein biomarkers. Finally, we accomplished a processing throughputof 0.353 s per sample for analyzing the result image.

키워드