Plasmon‐Enhanced Ultrasensitive Digital Imaging Immunoassay for the Quantification of microRNAs Assisted by Convolutional Neural Network Analysis

Abstract

AbstractDigital imaging immunoassays for precisely detecting microRNAs have attracted much attention in early cancer diagnosis. Currently, there is an urgent demand for improving the sensitivity and selectivity in digital immunoassays, which is limited by low imaging signal‐to‐noise ratio and pollutant interference. Herein, a plasmon‐enhanced ultrasensitive and selective digital‐readout imaging immunoassay is reported for microRNA quantification. Through DNA hybridization technology, Au nanoparticles are bound to target microRNAs and selectively placed onto an Au film coated slide, locally forming plasmonic nanocavities that dramatically enhance the scattering imaging signal of nanoparticles in dark‐field microscopy. Additionally, a convolutional neural network algorithm not only automatically recognizes and counts the image spots of the nanoparticles, but also excludes the counting errors inevitably induced by environmental pollutants or other noises. Ultimately, a quantitative determination of the microRNA‐375 cancer biomarker is demonstrated with a dynamic range that spans across 4 orders of magnitude (from 1 fm to 10 pm) with a low limit of detection (1.29 fm). This plasmon‐enhanced digital imaging immunoassay is a promising analytical tool toward point‐of‐care testing for various disease biomarkers at very low concentrations.