Predicting aged pork quality using a portable Raman device

• Published in: Meat science 2018-11, Vol.145, p.79-85
• Main Authors: Santos, C.C., Zhao, J., Dong, X., Lonergan, S.M., Huff- Lonergan, E., Outhouse, A., Carlson, K.B., Prusa, K.J., Fedler, C.A., Yu, C., Shackelford, S.D., King, D.A., Wheeler, T.L.
• Format: Article
• Language: English
• Subjects: least squares; On-line data collection; pork; Pork quality; prediction; Raman spectral; Raman spectroscopy; screening; sensory properties; spectral analysis; supply balance; Support vector machine; Tenderness prediction
• ISSN: 0309-1740; 1873-4138
• DOI: 10.1016/j.meatsci.2018.05.021

The utility of Raman spectroscopic signatures of fresh pork loin (1 d & 15 d postmortem) in predicting fresh pork tenderness and slice shear force (SSF) was determined. Partial least square models showed that sensory tenderness and SSF are weakly correlated (R2 = 0.2). Raman spectral data were collected in 6 s using a portable Raman spectrometer (RS). A PLS regression model was developed to predict quantitatively the tenderness scores and SSF values from Raman spectral data, with very limited success. It was discovered that the prediction accuracies for day 15 post mortem samples are significantly greater than that for day 1 postmortem samples. Classification models were developed to predict tenderness at two ends of sensory quality as “poor” vs. “good”. The accuracies of classification into different quality categories (1st to 4th percentile) are also greater for the day 15 postmortem samples for sensory tenderness (93.5% vs 76.3%) and SSF (92.8% vs 76.1%). RS has the potential to become a rapid on-line screening tool for the pork producers to quickly select meats with superior quality and/or cull poor quality to meet market demand/expectations.