پديدآورندگان :
Mohandes Dastgerdy Eisa Department of Toxicology and Pharmacology, Shahid Beheshti University of Medical Sciences, Tehran, IR Iran , Dashti Abolfazl Department of Toxicology and Pharmacology, Shahid BehesFood Safety Research Center, Shahid Beheshti University of Medical Sciences, Tehran, IR Iranhti University of Medical Sciences, Tehran, IR Iran## , Parastar Hadi Department of Chemistry, Sharif University of Technology, Tehran, IR Iran , Kobarfard Farzad Department of Medicinal Chemistry, Shahid Beheshti University of Medical Sciences, Tehran, IR Iran , Yazdanpanah Hassan yazdanpanah@sbmu.ac.ir Department of Toxicology and Pharmacology, Shahid Beheshti University of Medical Sciences, Tehran, IR Iran
كليدواژه :
Milk , Handheld infrared spectrometer , Chemometrics , Adulteration , FT , IR.
چكيده فارسي :
Food authenticity has been a major concern of consumers, producers and regulators all over the world [1]. Milk is wholesome nutritious dairy product, which has been confronted with various adulterations. Water is the most commonly added adulterant used in milk. This will not only reduce the nutritious quality of milk but also poses a threat to the health of consumers [2]. The presented study evaluates the performance of two handheld sensors (Wavelength range of 400-1000 nm for Linksquare and 900-1700 nm for Tellspec) benchtop attenuated total reflectance-Fourier transform infrared spectroscopy (ATR-FTIR) instrument combined with chemometrics methods to determine water adulteration in raw bovine milk. The adulterated samples were prepared in concentrations ranging from 0% to 30% water. No further preparation was required prior to analysis. All of the models developed were cross-validated and tested on a different set of data. Principal component analysis (PCA) was used to investigate and visualize the differences between authentic and adulterated milk samples and it could successfully discriminate authentic and adulterated ones. For quantification, two different models of partial least square regression (PLSR) and radial basis function-artificial neural network (RBF-ANN) [3] were applied on data and the best-established model was obtained from samples measured with the benchtop FTIR with the prediction regression coefficient (R2p) of 0.999 and the minimum root mean square of prediction (RMSEP) of 0.196%. Furthermore, the handheld sensors gained reliable results as Tellspec offered a R2p of 0.994 and a RMSEP of 1.116% and Linksquare showed a R2p of 0.920 and a RMSEP of 1.742%. To evaluate the applicability of the built models, ratio of performance to deviation (RPD) and range error ratio (RER) parameters were assessed. The optimal model based on FTIR instrument data obtained RPD=31.62 and RER=153.06, whereas the best models resulted in RPD=12.91 and RER=26.88 for Tellspec and RPD=4.34 and RER=6.21 for Linksquare. Taking into account the results, the studied handheld sensors revealed excellent capability for simple, economical, non-destructive and rapid measurements of water adulterated milk as screening and/or quality control technologies in the dairy industry.
