Long afterglow nanoprobes labeled image enhancement using deep learning in rapid and sensitive lateral flow immunoassay

Point-of-care testing (POCT) is currently the predominant method of in vitro diagnostics, with lateral flow immunoassay(LFIA) being commonly utilized in POCT. The performance of the probe in LFIA will directly affect the results of the detection. Therefore, the selection of probes is particularly important. However, the current commercially produced probes generally suffer from short luminescence lifetime, high detection cost, and poor anti-interference capability. Meanwhile, the traditional detection system has difficulties in detecting weakly positive samples. These ultimately lead to the problem of poor detection sensitivity. In order to address the above issues, we have synthesized an ultralong organic phosphorescence nanoprobe (UOPN). The UOPNs exhibit a long afterglow phosphorescence signal of around 300 ms after being excited, which can be used to eliminate endogenous interference. Meanwhile, focusing on the problem of poor detection of weakly positive samples in conventional detection systems, and the weak inherent signal of UOPNs, resulting in a low signal-to-noise ratio, we have developed a detection system that combines image processing and neural network-driven analysis for long afterglow image analysis. Through the analysis of the long afterglow images of serum amyloid A (SAA) in human blood, a detection limit of 0.01 μg/mL and a linear range of 0.1–251 μg/mL were achieved. The whole analysis was completed within 5 minutes. After pre-processing with the denoising algorithm, the system demonstrated an impressive classification accuracy of 99.24 %, validating the reliability of the detection system. This approach has further enhanced the sensitivity of immunoassay, contributing to the advancement of point-of-care testing. [Display omitted] •A low-cost and low-noise long afterglow nanoprobe was synthesized and applied in lateral flow immunoassay.•A deep learning model based on long afterglow images is proposed to enhance sensitivity.•A rapid and highly sensitive portable long afterglow image reader is proposed for processing long afterglow images.•Clinical validation indicates that biosensors using deep learning are highly reliable and contribute to enhanced sensitivity.