UTILIZATION OF NIR SPECTROSCOPY TO ESTIMATE THE PROXIMATE COMPOSITION OF TROUT MUSCLE WITH MINIMAL SAMPLE PRETREATMENT

Reflectance spectroscopy in the near-infrared (NIR) region 900-1800 nm) was used to predict the proximate composition of either fresh or frozen rainbow trout muscle. The standard error of prediction (RMSEP) for moisture was 1.1% (over the range 71.5-76.7% moisture), for crude lipid, 3.1% [range 8.8-17.0% dry weight basis (dwb)], and for protein nitrogen (Kjeldahl, % N x 6.25), 5.4% [over the range 69.3-87.4% protein (dwb)]. Two calibration methods, partial least squares (PLS) and multiple linear regression (MLR), were used to develop models that related spectral absorbances to moisture, lipid, and protein concentrations. This NIR reflectance spectroscopic method could be useful for making a rough estimate of the moisture and protein content of fish muscle for quality control or animal husbandry purposes. Sample preparation is minimal, and the method does not require that the muscle sample be homogenized, dried, or extracted prior to analysis.