Automated Image Processing Algorithm for the Quantitative Evaluation of Multi-Line Strip-Type Biosensors

Lateral flow assays (LFAs) are valuable tools in point-of-care (POC) diagnostics owing to their simplicity, portability, rapid analysis, and cost-effectiveness. However, conventional LFAs typically provide qualitative or semi-quantitative results, which are susceptible to subjective interpretation and false results. To overcome these limitations, objective, automated evaluation tools are required. This work presents a novel, Python-based automated image processing algorithm developed for the simultaneous qualitative and quantitative analysis of multiplexed LFAs. The algorithm's workflow encompasses (a) automated image capture and cassette detection, (b) extraction of the region of interest (ROI), (c) test line segmentation and boundaries detection and (d) optical density quantification and subsequent conversion to analyte concentration. To validate the platform, two multiplexed assay configurations were developed: a 4-plex diagnostic panel for infectious disease pathogens (Escherichia coli, Streptococcus pneumoniae, Haemophilus influenzae, and SARS-CoV-2) and a 3-plex liquid biopsy panel for microRNA detection in urine. The automated system demonstrated high specificity for both LFAs. Compared to traditional, semi-automated analysis using ImageJ software, the proposed algorithm achieved better accuracy, while similar performance was reached compared to a commercial strip reader but with enhanced automation. For the 4-plex assay, the system achieved an overall accuracy, sensitivity, and specificity of 93.5%, 100%, and 84.4%, respectively. For the 3-plex assay, the corresponding metrics were 92.3%, 88.4%, and 96.7%. This integrated approach provides a low-cost, portable, and automated multi-analyte diagnostic solution highly suited for decentralized clinical healthcare, environmental monitoring, and resource-limited settings.