Direct visual enumeration of viruses under dark-field microscope (DFM) using plasmon resonance probes (PRPs) is fast and convenient; however, it is greatly limited in the assay of real samples because of its inability to accurately identify false positives owing to non-specific adsorption. In this study, we propose an artificial intelligence (AI)-assisted DFM enumeration strategy for the accurate assay of Enterovirus A71 (an ultra-small human virus) using two PRPs; a 40 nm silver nanoparticle probe (SNP) that appears bright blue under DFM, and a 120 nm gold nanorod probe (GNP) that appears red under DFM. The capture chip was prepared by immobilizing the SNPs with antibodies on the glass to capture the target virus and to form dichromatic sandwich structures with the GNPs, followed by imaging under a dark field (DF). Subsequently, the DF images of the capture chip were subjected to a two-step screening: first, using image processing, and thereafter using the AI algorithm screening to eliminate false positive results and background noise. The results revealed that the data from the AI-assisted dual PRPs assay were highly consistent with those of quantitative PCR (qPCR), and that the sensitivity with a minimum detectable concentration of 3 copies/μL was 5 times higher than that of qPCR. The entire analysis was completed within 45 min. Therefore, our AI-assisted virus enumeration strategy with two DF PRPs holds great potential for ultra-sensitive and accurate quantification of viruses in real samples.
Keywords: Accurate enumeration; Artificial intelligence; Dark-field; Plasmon resonance probe; Ultra-small virus.
Copyright © 2021 Elsevier B.V. All rights reserved.