|Liu, Yongliang - UNIV OF GEORGIA|
Submitted to: Poultry Science
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: April 24, 2004
Publication Date: July 28, 2004
Citation: Liu, Y., Lyon, B.G., Windham, W.R., Lyon, C.E., Savage, E.M. 2004. Prediction of physical, color, and sensory characteristics of chicken breasts by visible/near-infrared spectroscopy. Poultry Science. 83:1467-1474. Interpretive Summary: Although poultry processors need fast, non-destructive techniques to understand how processing techniques affect eating quality of their products, most of the current instrumental and sensory techniques are destructive, time consuming and unsuitable for use directly on-line in processing plants. In this study, researchers combined a procedure called near-infrared (NIR) spectroscopy with several existing meat quality assessment techniques to determine the feasibility of NIR as a fast, non-destructive technique to predict color, tenderness and sensory attributes of poultry. By investigating ranges of poultry breast meat quality achieved through varying the debone times, researchers found several combinations of instrumental and sensory attributes with strong and positive correlations that could be used to develop quality predictions when combined with NIR. For example, one statistical model, based on predicted shear values, gave a correct classification of 74% into "tender" and "tough" groups when 7.5 kg was used as a boundary. This information is useful to poultry processors, retailers, and researchers who are interested in applying visible/NIR spectroscopy-based systems for on-line quality grading or classifying or for modeling the quality predictions of cooked poultry muscle.
Technical Abstract: The feasibility of predicting pH, color, shear force and sensory characteristics of chicken breasts deboned at 2, 4, 6, and 24 h post-mortem by visible/near infrared reflectance (NIR) spectroscopy in 400 - 1850 nm region was determined. Prediction of physical attributes of CIE color values (L*, a*, and b*), pH, and shear force had better accuracies than those of individual sensory attributes. Calibration and validation statistics for shear force and sensory traits indicated that visible/NIR models were not significantly improved for cooked muscles compared to predictions based on raw muscle characteristics. On the basis of predicted shear values from the partial least squares (PLS) model, breast samples were classified into "tender" and "tough" classes with a correct classification of 74.0% if the boundary was set to be 7.5 kg. The model developed from measured shears using soft independent modeling of class analogy/principal components analysis (SIMCA/PCA) showed nearly the same classification success.