Design and fabrication of artificial neural network-digital image-based colorimeter for protein assay in natural rubber latex and medical latex gloves

Abstract

Digital image-based colorimetry (DIC) using a complementary metal oxide semiconductor (CMOS) camera as a detector coupled with artificial neural network (ANN) was developed for protein assay in natural rubber (NR) latex and medical latex gloves. This method was based on the RGB values (red, green and blue) of different color intensities from the reaction of protein complexes with the modified Lowry reagent. Protein standard solutions and protein in the samples were captured by a single image in the DIC light box. The data was processed by an ANN program to evaluate the RGB value of each digital image. Under the optimum conditions, the amount of protein could be determined in the concentration range of 0-10μgmL-1. When comparing the results obtained from the DIC-ANN with the spectrophotometric method, there was no statistical difference at 95% confidence level by applying t-test at unequal variance. The average mean squared error (MSE) for the protein assay was 0.037. The limit of quantitation by the proposed method (defined as the concentration that could be photographed and processed by an ANN program) was 1μgmL-1. The proposed method was successfully applied to the determination of extractable protein in NR latex and medical latex gloves and proved to be a convenient, rapid and inexpensive method. © 2012 Elsevier B.V.