Near- and mid-infrared determination of some quality parameters of cheese manufactured from the mixture of different milk species

AYVAZ H., Mortas M., Dogan M. A., Atan M., YILDIZ TİRYAKİ G., Karagul Yuceer Y.

JOURNAL OF FOOD SCIENCE AND TECHNOLOGY-MYSORE, vol.58, no.10, pp.3981-3992, 2021 (SCI-Expanded) identifier identifier

  • Publication Type: Article / Article
  • Volume: 58 Issue: 10
  • Publication Date: 2021
  • Doi Number: 10.1007/s13197-020-04861-0
  • Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED), Scopus, ABI/INFORM, Agricultural & Environmental Science Database, Analytical Abstracts, Aquatic Science & Fisheries Abstracts (ASFA), BIOSIS, Biotechnology Research Abstracts, CAB Abstracts, Compendex, Food Science & Technology Abstracts, INSPEC, Veterinary Science Database
  • Page Numbers: pp.3981-3992
  • Keywords: Cheese, Chemometrics, Dairy, Ezine, Mid-infrared, Near-infrared, Quality
  • Çanakkale Onsekiz Mart University Affiliated: Yes


This study aimed to evaluate the performance of both near-infrared (NIR) diffuse reflectance and mid-infrared-attenuated total reflectance (MIR-ATR) in determining some quality parameters of a commercial white cheese made of unknown ratios of various milk species. For this purpose, 81 commercial Ezine cheese samples, a special ripened cheese produced in Turkey, containing unknown ratios of bovine, caprine, and ovine milk, were used. Reference analyses, including textural properties, protein content, nitrogen fractions, ripening index coefficients, fat, salt, dry matter-moisture, and ash contents as well as pH and titratable acidity levels, were conducted in the samples following the traditional gold standards. For NIR applications, the spectra of both intact cubes and hand-crushed cheese samples were collected, whereas the spectra of only hand-crushed cheese samples were collected for MIR-ATR. PLSR (Partial Least Squares Regression) calibration models were developed for each parameter (n = 61) and then validated using both cross-validation (leave-one-out approach) and an external validation set (n = 20). Overall, PLSR models developed for total protein, fat, salt, dry matter, moisture, and ash content, as well as pH and titratable acidity, yielded satisfactory performance statistics in the complementary use of NIR and MIR spectroscopy. However, PLSR models of the other parameters, including textural properties, nitrogen fractions, and the ripening index, could only separate high and low values and were not able to make accurate quantitative predictions. NIR spectroscopy was found to be more accurate than that of MIR-ATR spectroscopy for almost all the parameters except for pH and titratable acidity, for which MIR-ATR spectroscopy was superior.