A two-stage signal enhancement method integrating the Gaussian mixture model and adaptive rolling ball technique for ultrasensitive fluorescent immunochromatographic detection

Abstract

Lateral flow immunochromatographic assays (LFIAs) with fluorescent labels have emerged as powerful analytical tools for point-of-care diagnostics, offering superior sensitivity over conventional colorimetric methods. However, quantitative analysis at low analyte concentrations remains challenging due to insufficient signal contrast and background interference. To address this challenge, this study developed a two-stage signal-enhancement method integrating the Gaussian mixture model (GMM) and adaptive rolling ball (ARB) technique, achieving the ultrasensitive detection and precise quantification of weak fluorescent signals. The method employed a “coarse classification-fine refinement” collaborative strategy and combined the Hill equation to establish quantitative relationships between signal intensity and target concentration. Validation using a quantum dot fluorescent labeling system demonstrated a detection sensitivity of 10⁻¹⁰ mol L⁻¹, representing 1–2 orders of magnitude improvement over conventional methods. Under limited concentration conditions, the method achieved a signal-to-noise ratio of 27.3 ± 2.64 dB, contrast-to-noise ratio of 12.62 ± 2.87, peak-to-valley ratio of 151.65 ± 30.2, and background suppression rate of 73% ± 3.4%, which were significantly superior to those of control methods. In Escherichia coli detection, the detection limit improved from 10³ to 10² CFU mL⁻¹, with a Pearson correlation coefficient of 0.998 compared with the PCR gold standard. The method exhibited excellent performance in high-noise environments and multi-target detection (Staphylococcus aureus/Klebsiella pneumoniae), with R² > 0.99, providing a practical solution for ultrasensitive point-of-care diagnostics and pathogen screening in resource-limited settings.