Identification of broiler chicken meat using a visible/near-infrared spectroscopic technique

Abstract

A near-infrared spectroscopic method was developed with a dummy regression technique to differentiate meat originating from broilers and Chinese local chickens. Best classification accuracies of 100%, 92%, 96% and 92% were achieved for minced thigh meat, minced breast meat, breast cut without skin and breast cut with skin respectively. Comparison among the regression models of MLR, PCR, PLS and mPLS did not show obvious differences in classification accuracy. Scatter correction and derivative treatment of the spectral data before discriminant analysis often improved the classification accuracy for minced meat, while for meat cuts, spectra without pretreatment produced better classification. In general, using the full spectrum of 400-2500 nm produced satisfactory classification. The spectrum in the visible region of 400-750 nm, the short-wavelength NIR region of 750-1100 nm or the long-wavelength NIR region of 1100-2500 nm can also produce satisfactory classification depending on sample presentation methods and regression models. The spectroscopic classification was supported by physical and chemical properties of meat samples, which showed significant differences in collagen and fat contents and pH and chromatic values between the two groups of chickens. The results of the present study indicate that NIR spectroscopy can be used to identify broiler meat or carcass from those of local chickens.