Leitora de testes rápidos do tipo fluxo lateral utilizando machine learning

Abstract

Lateral flow immunoassays are simple, fast-resulting tests that allow large-scale applications without the need for complex infrastructure or laboratory processes. This type of test came to prominence with the recent Covid-19 pandemic, but its scope is much broader and can range from detecting contaminants in food to veterinary medicine. Characterized by the result indicated through stripes that appear when depositing a sample in the immunoassays, the result is given in a quantitative way (where the intensity of the stripes indicates the concentration of an antigen or contaminant in the sample being analyzed) or qualitatively (where a threshold is established, and the result can be reactive or non-reactive). Analyzing this type of test with the naked eye can be challenging, in some cases leading to human errors, in addition to being a time-consuming task for health professionals when used on a large scale. In this undergraduate thesis, a tool was developed capable of displaying the results of such tests without the need for human interference, through a reader that uses a camera to generate images and, through a previously trained artificial neural network, classifies the rapid immunoassays according to their respective test disease and displays the results through a graphical user interface.