Pixel-Based Machine Learning and Image Reconstitution for Dot-ELISA Pathogen Diagnosis in Biological Samples

Serological methods serve as a direct or indirect means of pathogen infection diagnosis in plant and animal species, including humans. Dot-ELISA (DE) is an inexpensive sensitive, solid-state version the microplate enzyme-linked immunosorbent assay, with broad range applications epidemiology. Yet, its applicability limited by uncertainties qualitative output assay due to overlapping dot colorations positive negative samples, stemming mainly from inherent color discrimination thresholds human eye. Here, we report novel approach for unambiguous DE evaluation applying machine learning-based pattern recognition image pixels blot using impartial predictive model rather than judgment. Supervised learning was used train classifier algorithm through built multivariate logistic regression based on RGB (“Red,” “Green,” “Blue”) pixel attributes scanned samples known status ( Lettuce big-vein associated virus ). Based trained cross-validated algorithm, probabilities unknown could be predicted images, which would then reconstituted having above cutoff. The cutoff may selected at will yield desirable false rates depending question hand, thus allowing proper classification and, hence, accurate diagnosis. Potential improvements diagnostic proposed versatile method that translates unique antigens universal basic language are discussed.