Prediction of raw meat texture and myopathic severity of broiler breast meat with the wooden breast condition by hyperspectral imaging.

Abstract

1. This research explored the potential of hyperspectral imaging (HSI) to predict meat texture and the wooden breast (WB) condition in raw chicken breast fillets, categorised as normal, moderate WB and severe WB. The Meullenet-Owens Razor Shear (MORS) measurement was employed to characterise raw meat texture traits, including force, energy and peak count.2. Significant differences in MORS force, energy and peak count were observed between normal and severe WB fillets. However, no significant differences in these traits were found between normal and moderate WB fillets.3. Partial least square regression (PLSR) models, using the full wavelength range of visible and near-infrared (Vis-NIR) spectra, successfully predicted meat texture traits, with MORS peak counts exhibiting the highest predictive ability (Rp = 0.915 and RMSEp = 2.26). Key wavelengths were identified using the regression coefficient (RC) method, highlighting their significance in characterising meat texture.4. A linear discriminant analysis (LDA) model, incorporating all key wavelengths, achieved accurate predictions of WB severity, with 84.72% in the calibration set and 77.78% in the prediction set. This model demonstrated the potential of HSI in distinguishing WB fillets from normal ones, with an accuracy of 97.22%in the calibration set and 91.67% in the prediction set. Distribution maps generated using key wavelengths visually depicted variations in meat texture traits and WB severity.5. This study underscored the efficacy of HSI technology in predicting meat texture and WB severity in raw chicken breast fillets.