| dc.description.abstract | The purpose of this study was to examine if Near infrared (NIR) spectroscopy which is a rapid and non-destructive method, can be used to determine primary and secondary lipid oxidation products in salmon oils. PV (peroxide value) and TBARS (thiobarbituric acid reactive substances) were used as chemical measures of lipid oxidation. PV measures the degree of primary oxidation products, while TBARS measures the degree of secondary oxidation products formed from the oil. NIR spectroscopy with transectance method was used to obtain the NIR spectra for 108 salmon fat extracts and 9 crude salmon oils.
The results from this study, showed that NIR spectroscopy can be used for PV determination. The calibration model for TBARS had a poorer result, but several other factors needs to be tested, to see if this model can be improved.
A good calibration model needs between 50-100 samples, and also a sucient calibration range. These goals have been achieved for salmon fat extracts models (PV and TBARS) and crude salmon oils (but only 9 samples are included in the model). The PV measured for salmon fat extracts were in the range between 0 and 20.57 meq/kg oils, and for TBARS the values varied from 0 to 1.316 μmol malonaldehyde/g oil. The PV measured for crude salmon oils were in the range between 0.61 and 43.32 meq/kg oil.
The best PV calibration model for salmon fat extracts (108 samples) was obtained from the non-preprocessed data, resulting in r=0.862, RMSEP=2.799, and PCs=21. And for TBARS the best calibration model for salmon fat extracts (86 samples) was obtained from the 1st derivative Savitzky-Golay preprocessed data, resulting in r=0.690, RMSEP=0.230, and PCs=16. For the crude salmon oils (9 samples) the best PV calibration model was obtained from the MSC (mulitplicative scatter correction) preprocessed data, yielding r=0.820, RMSEP=8.748, and PCs=2.
PV calibration models were also made for salmon fat extracts and crude salmon oils together ((117 samples), the best result was obtained from the MSC preprocessed data, with r=0.914, RMSEP=2.896, and PCs=18. This good result was obtained from the large PV range from 0-43.32 meq/kg oil, but the salmon fat extracts are not represented above PV=20.57 meq/kg oil, so higher PV would probably be needed in this model for salmon fat extracts.
PV and TBARS calibration models for selected wavelength regions were also made for salmon fat extracts, and for crude salmon oils. The PV and TBARS calibration models for selected wavelength regions for salmon fat extracts, resulted in a much poorer calibration model than for the non selected wavelength regions. However, for crude salmon oils the PV calibration model gave good result for regions between 2000 and 2200nm, with r=0.978, RMSEP=3.099, and PCs=3. This region can be used for PV determination in crude salmon oils. | nb_NO |
