Quantitative Interpretation of Lateral Flow Immunoassays via Smartphone Image Processing

Abstract

Modern disease diagnosis often depends on sophisticated laboratory instruments such as polymerase chain reaction (PCR) systems and enzyme-linked immunosorbent assays (ELISAs), which, despite their excellent analytical performance, are expensive, time-consuming, and unsuitable for point-of-care (POC) testing. In contrast, lateral flow immunoassays (LFIAs) provide rapid, low-cost diagnostics but are limited by low sensitivity and reliance on subjective visual interpretation. To address this, this dissertation integrates smartphone imaging and computational analysis into LFIA workflows to establish a quantitative, image-processing-based diagnostic platform. A custom smartphone application was developed to capture images, detect regions of interest, normalize intensity, and calculate test-to-control line ratios, enabling objective quantification without external instruments. In Chapter 2, a light-source-free LFIA for salivary cortisol was developed using a photoluminescent film and platinum-nanoparticle (Pt NP) labels. The assay achieved a limit of detection (LoD) of 139 pg/mL and showed strong correlation with ELISA results (r = 0.935), confirming accurate quantitative cortisol measurement under variable lighting conditions. Chapter 3 introduces a TMB-enhanced LFIA (TELFIA) for influenza A virus (IAV) detection that integrates catalytic amplification with smartphone-based quantification. The TELFIA cartridge employs a pre-dried enhancing pad containing citric acid, hydrogen peroxide, and 3,3′,5,5′-tetramethylbenzidine (TMB), producing strong colorimetric amplification through Pt NP–mediated oxidation. The system achieved an LoD of 11.6 pg/mL and demonstrated 96.8 % sensitivity and 98.4 % specificity relative to RT-PCR, with percentage-intensity ratios inversely correlated with Ct values (R² = 0.832). Collectively, this work establishes smartphone-based image processing as a powerful tool for quantitative LFIA evaluation, bridging the gap between laboratory-grade precision and POC simplicity. The developed framework provides a scalable foundation for AI-driven, connected biosensing systems, enabling reliable, real-time, and instrument-free disease diagnostics.