Kemometrik Yaklaşımlarla Gıda Tağşişlerinin Belirlenmesinde Spektroskopik Yöntemlerin Kullanılması
![](/themes/Mirage2//images/mime2.png)
View/ Open
Date
2019Author
Temiz , Havva Tümay
xmlui.dri2xhtml.METS-1.0.item-emb
Acik erisimxmlui.mirage2.itemSummaryView.MetaData
Show full item recordAbstract
Food integrity can be actualised by providing monitorable, safe, high quality and authentic food. The provision of these components reveals the need to develop rapid and precise analysis methods. In recent years, synchronous fluorescence spectroscopy (SFS), laser induced plasma spectroscopy (LIBS) and Raman spectroscopy (RS) have been extensively investigated due to their high sensitivity and the detailed information provided. Since data obtained from these techniques are of a high dimension, different chemometric methods are used to make the data meaningful. Within the scope of this thesis; SFS, LIBS and RS techniques, and Principal Component Analysis (PCA), Partial Least Squares-Discriminant Analysis (PLS-DA), Partial Least Squares Regression (PLSR), Principal Component Regression (PCR) and Multiple Linear Regression (MLR) were used to develop the detection methods for three different food adulterations. The common point of these methods is the use of information on the lipid composition of food. The potential of some of the major and minor molecules in the lipid phase of foods to detect different food adulterations was revealed. Addition of sunflower oil to the tahini oil was determined by using SFS. In the PCA model, the cumulative variances of PC1 and PC2 were 81.50% and 99.45%. Using the wavelength selection mode (WSM), the wavelength interval (Δλ) of the most successful PCA model was determined to be the data collected at 60 nm. The separation between the groups resulted from the difference between the phenolic and tocopherol contents. PLSR, PCR and MLR models were developed for quantitative determination of adulteration. The best results were obtained by using PLSR. The root mean square error of calibration (RMSEC), cross validation (RMSECV) and prediction (RMSEP) models at different Δλ are between 0.26-0.84, 0.74-1.40 and 1.24-1.82. The relative error of prediction (REP) and relative standard deviation (RSD) values are 0.87-6.71 to 13.98-20.64. The limit of detection (LOD) and limit of quantification (LOQ) values are 0.34-0.86% and 1.14-2.88%. WSM application generally increased the error values, REP, RSD, LOD and LOQ. SFS and PLS techniques has enabled the development of a detection method with high sensitivity and reproducebility. The fatty acid composition of the oils was determined by GC. Using LIBS, addition of margarine to butter was determined. In the PCA model, the cumulative variance of PC1 and PC2 were 88.76% and 97.92%. The classification was caused by the differences in Na, Ca, K, Fe, Mg and Zn contents. PLSR resulted with R2cal=0.999 and R2pred=0.984. RMSEC, RMSEP, RSD and REP were 2.02, 3.37, 11.3 and 13.0. LOD and LOQ were 3.9% and 13.0%. The elemental composition of the oils was determined by inductively coupled plasma-mass spectrometry (ICP-MS) and flame atomic absorption spectrometry (FAAS). RS was used to monitor the changes in lipid composition due to the application of repeated freezing and thawing of the chicken samples. Lipid was extracted by hexane and Folch extractions. The fresh and frozen-thawed samples were separated by using their Raman spectra. The metabolomics analysis of the lipid phase was performed with quadrapol-time of flight liquid chrmatography mass spectrometry (Q-TOF LCMS).