A Computer Vision Algorithm for the Digitalization of Colorimetric Lateral Flow Assay Readouts

Lateral flow assays (LFAs) are low-cost testing tools widely used for home, point-of-care, or laboratory medical diagnostics. These tests typically use colorimetry to report the presence and the concentration of a certain physical/ biological quantity, showing the result as a color marker. This work presents a computer vision algorithm for the digitalization of LFA readouts, enabling precise and reliable results at low-cost. The algorithm receives as input an image of a sample, identifies the color marker, and computes its average color intensity. In contrast to existing algorithms, the proposed one can detect color markers that are not characterized by a predetermined precise shape, size, and position, since the topology is identified and analyzed by the algorithm itself. The evaluation of the proposed algorithm on a set of LFA strips shows correct functionality and execution time of less than a second.