Quantitation of lamb content in mixtures with raw minced beef using visible, near and mid-infrared spectroscopy

Samples of beef (32), lamb (33) and 5%, 10% and 20% (w/w) lamb-in-beef mixtures (33 each) were minced and reflectance scanned in the visible, near and mid-infrared spectral regions. Partial least squares (PLS) regression models were developed to predict percentage lamb content using each spectral region alone and combinations of all three,The most accurate models combined mid-IR (800-2000 cm(-1)) and near IR (1100-2498 nm) spectral data following 2(nd) derivatization; standard errors of prediction of 0.91% (0-20% range in lamb content) and 4.1% (0-100% range in lamb) were obtained. This technique may be useful for screening such meat mixtures.