Smartphone-based plasmonic biosensing via asymmetric droplets and deep generative networks

Low-cost biosensing methods are crucial for the early detection of diseases. The work presents a smartphone-based detection scheme based on two asymmetrically evaporated droplets on a nanofibrous membrane with the specimen sample and plasmonic nanoparticles, respectively. Leveraging deep learning algorithms, we achieve automatic detection and quantification of a range of proteins based on distinct droplet patterns to differentiate positive/negative cases, including N-protein, PCT, CEA, and PSA at concentrations as low as 10 pg/ml in 12 minutes. The method enhances the droplet pre-concentration and the generation of optically structural patterns via plasmonic particles for improved sensing sensitivity.