Quality assessment of intact chicken breast fillets using factor analysis with Vis/NIR spectroscopy

Authors: YANG, YI - China Agricultural University; Zhuang, Hong; Yoon, Seung; WANG, WEI - China Agricultural University; JIANG, HONGZHE - China Agricultural University; JIA, BEIBEI - China Agricultural University; LI, CHUNYANG - Jiangsu Academy Agricultural Sciences

Submitted to: Journal of Food Analytical Methods
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 11/14/2017
Publication Date: 12/4/2017
Citation: Yang, Y., Zhuang, H., Yoon, S.C., Wang, W., Jiang, H., Jia, B., Li, C. 2018. Quality assessment of intact chicken breast fillets using factor analysis with Vis/NIR spectroscopy. Journal of Food Analytical Methods. https://doi.org/10.1007/s12161-017-1102-0.
DOI: https://doi.org/10.1007/s12161-017-1102-0

Interpretive Summary: This study is concerned with non-destructively and rapidly predicting important quality attributes of fresh chicken breast fillets, including color (CIE L*, a*, b*), pH, moisture, drip loss, expressible fluid, and salt-induced water gain. The visible and near-infrared spectral responses between 400 and 2,500 nm were measured and analyzed by factor analysis (FA) and partial least squares regression (PLSR). The PLSR models using three important factors that were obtained from full spectra (all wavelengths in a spectrum) by FA were developed, called FA-PLSR models. The prediction models based on the FA-PSLR using only three key factors outperformed the models using the individual quality traits using full spectra. The PLSR models using a small number of key wavelengths in each spectrum also showed the comparable predictive ability with the corresponding FA-PSLR models. The study finding demonstrated the potential for FA and Vis/NIR spectroscopy as a useful method to assess the quality of intact chicken breast fillets.

Technical Abstract: Factor analysis (FA) method was tested to assess quality of chicken breast fillets with the visible/near-infrared (Vis/NIR) spectroscopy with wavelength range between 400 and 2500 nm. According to inherent correlation, three factors were extracted from the measured eight quality traits (L*, a*, b*, pH, moisture, drip loss, expressible fluid, and salt-induced water gain). The extracted “grade factor” (F1), “color factor” (F2), and “moisture factor” (F3) could respectively represent the characteristics and the variation tendency of the corresponding quality traits and were defined as three new quality assessment indexes. Furthermore, partial least squares regression (PLSR) models were established to quantitatively relate spectral information to eight individual quality traits and three factors. The results indicated that the models for predicting each factor performed better than those for individual quality traits. Key wavelengths of each quality trait were then selected, and the corresponding spectra were taken to build new PLSR prediction models. The selected key wavelengths showed obvious practical significance, and the new models had comparable predictive performance to those models developed based on the full spectra, among which the new models of F1 and F2 had acceptable and robust predictive abilities (R2p'='0.73, RPD'='1.91; R2p'='0.74, RPD'='1.97). Our results in the present study demonstrate the potential for FA and Vis/NIR spectroscopy as a useful method to assess the quality of chicken breast fillets.