Feasibility of the direct application of near-infrared reflectance spectroscopy on intact chicken breasts to predict meat color and physical traits

Physical and color characteristics of chicken meat were investigated on 193 animals by directly applying a fiberoptic probe to the breast muscle and using the visible-near-infrared (NIR) spectral range from 350 to 1,800 nm. Data on pH was recorded 48 h postmortem (pH); lightness (L*), redness (a*), and yellowness (b*) 48 h postmortem; thawing and cooking losses and shear force after freezing. Partial least squares regressions were performed using untreated data, raw absorbance data (log(1/R)), and multiplicative scatter correction plus first or second derivative spectra. Models were validated using full cross-validation, and their predictive ability was determined by root mean square error of cross-validation (RMSE(CV)) and correlation coefficient of cross-validation (r(cv)). Means (±SD) of pH, L*, a*, b*, thawing loss, cooking loss, and shear force were 5.83 ± 0.13, 44.54 ± 2.42, -1.90 ± 0.62, 3.21 ± 3.28, 4.84 ± 2.44%, 19.39 ± 2.95%, and 16.08 ± 3.83 N, respectively. The best prediction models were developed using log(1/R) spectra for b* (r(cv) = 0.93; RMSE(CV) = 1.16) and a* (r(cv) = 0.88; RMSE(CV) = 0.29), while a medium predictive ability of NIR was obtained for pH, L*, and thawing and cooking losses (r(cv) from 0.69 to 0.76; RMSE(CV) from 0.01 to 1.73). Finally, predicted model for shear force (r(cv) = 0.41; RMSE(CV) = 3.18) was unsatisfactory. Results suggest that NIR is a feasible technique for the assessment of several quality traits of intact breast muscle.