LAUSR

Nucleic acid amplification assays including loop-mediated isothermal amplification (LAMP) are routinely used in diagnosing diseases and monitoring water and food quality. The results of amplification in these assays are commonly measured with an analog fluorescence readout, which requires specialized optical equipment and can lack quantitative precision. Digital analysis of amplification in small fluid compartments based on exceeding a threshold fluorescence level can enhance the quantitative precision of nucleic acid assays (i.e. digital nucleic acid amplification assays), but still requires specialized optical systems for fluorescence readout and the inclusion of a fluorescent dye. Here we report Fractal-LAMP, an automated method to detect amplified DNA in sub-nanoliter scale droplets following LAMP in a label-free manner. Our computer vision algorithm achieves high accuracy detecting DNA amplification in droplets by identifying LAMP byproducts that form fractal structures observable in brightfield microscopy. The capabilities of Fractal-LAMP are further realized by developing a Bayesian model to estimate DNA concentrations for unknown samples and a bootstrapping method to estimate the number of droplets required to achieve target limits of detection. This digital, label-free assay has the potential to lower reagent and reader cost for nucleic acid measurement while maintaining high quantitative accuracy over three orders of magnitude of concentration.