Determination of moisture and total protein and phosphorus contents in powdered chicken egg samples using digital images, NIR spectra, data fusion, and multivariate calibration

This work uses digital images, near-infrared (NIR) spectra, data fusion, and multivariate calibration to develop fast, non-laborious, and cheap methods for determining moisture and total protein and phosphorus contents in powdered chicken egg samples. These analyzed samples were obtained after breaking, homogenizing, lyophilizing, and pulverizing whole chicken eggs. Digital images were captured employing a scanner, while NIR spectra were recorded using benchtop and portable spectrometers. The proposed multivariate calibration methods were better in terms of correlation coefficient (r) and ratio performance deviation of prediction (RPD) for predicting total protein content using digital images (r = 0.995 and RPD = 9.55) than those obtained by benchtop and portable NIR spectra. On the other hand, in predicting moisture (r = 0.974 and RPD = 4.35) and total phosphorus content (r = 0.98 and RPD = 4.3), the benchtop NIR produced better results. The analytical performance of data fusion methods was slightly poorer than the best methods using preprocessed NIR spectra or digital images separately. Therefore, these proposed methods proved to be a good alternative tool for destructive chemical analysis, given that powdered chicken egg samples could be analyzed without using chemicals and without generating waste harmful to health and the environment, following the basic principles of Green Chemistry.