Convolutional Neural Network-Assisted Ultrasensitive Immunochromatographic Strips of Salmonella typhimurium through Bright Luminescence and Nano-Biointerfacial Affinity Leveraging Schiff-Base Chemistry-Confined Mechanism.

Abstract

The development of a signal probe that simultaneously possesses high luminescence and superior nano-biointerfacial affinity remains challenging for advancing immunochromatographic assay (ICA) strips to detect foodborne pathogens. Herein, we propose convolutional neural network-assisted ICA strips to detect Salmonella typhimurium in various agri-foods through anchoring aggregation-induced emission luminogens, ETT, into aminophenol-formaldehyde resin nanobowls (AFRNBs) via Schiff-base chemistry, yielding high-performance fluorescent AFRNBs/ETT with enhanced nano-biointerfacial affinity. The rigid nanobowl effectively restricts the intramolecular motion of ETT, bringing a markedly enhanced quantum yield and a prolonged fluorescence lifetime. Concurrently, the abundant groups on the AFRNBs surface enable entropy-driven efficient conjugation with antibodies, endowing AFRNBs/ETT with exceptional nano-biointerfacial affinity. Therefore, the developed AFRNBs/ETT-based ICA strips for detecting Salmonella typhimurium demonstrate good analytical performance, with a detection limit of 78 CFU mL-1, alongside high specificity, stability, reproducibility, and feasibility in various agri-foods. The convolutional neural network model further enables automated and objective readout, boosting detection reliability.