Testing milk : statistical approach reveals fat content through spectra

Testing times in the dairy

A partial least-squares regression models based on either selected regions or full Fourier-transform Raman spectra can reveal the fat content and profile of milk samples quickly and easily. To make appropriate nutritional and management decisions on a dairy farm, additional tools assessing the nutritional status of a dairy herd, such as milk urea or beta-hydroxybutyrate estimates can help guide management.

Now, I. Stefanov, E. Colman, B. Vlaeminck and V. Fievez of the Laboratory for Animal Nutrition and Animal Product Quality (LANUPRO) at Ghent University, in Melle, with V. Baeten and O. Abbas of the Walloon Agricultural Research Centre, in Gembloux, and B. De Baets of the Faculty of Bioscience Engineering, at Ghent University, Ghent, Belgium, have developed a chemometrics approach to analysing Fourier transform infra-red (FTIR) spectroscopic data.

Milk assessment

Previously, the team had demonstrated that the potential of odd- and branched-chain fatty acids (OBCFAs), i.e. milk secretion of OBCFAs, is related to duodenal flow of microbial biomass. Moreover, an ability to detect individual saturated odd-, iso-, and anteiso-branched chain fatty acids (FAs), has the potential to monitor the nutrients produced during digestive processes. The implication is that rapid determination of milk OBCFAs and FAs via spectroscopic testing could be a powerful way to management dairy herd nutrition.

The team points out that given the potential of spectroscopy in this area a new analytical approach to extract this data from spectra is needed. Their approach uses two pre-treatment methods, namely multiplicative scatter correction (MSC, using raw spectra of milk fat only) and modified MSC (MMSC, a combination of pure FAMESs and milk fat spectra), with cross-validation to evaluate the different types of milk fat in samples. It can then be used to predict the presence of individual odd- and branched-chain fatty acids (OBCFAs) and their sums in milk samples.

They explain that various spectroscopic methods have been proposed for the analysis of the FA content of milk, or milk fat. These are infrared (IR), including mid-infrared (MIR) and near-infrared (NIR), Raman, and ultraviolet−visible (UV−vis) spectroscopy. Raman has recently demonstrated its prowess in the identification and even the quantification of molecular constituents of various types of sample. It is of particular interest because Raman spectra can show well-resolved fundamental vibrational transitions. As such, the team says, Raman has been used to analyse fats and oils, in olive and olive pomace, for oil maturation, authentication and quantitative determination of saturated, monounsaturated, and polyunsaturated components of lard.

"The aim of this work was to investigate the potential of Raman spectroscopy to determine OBCFAs in milk fat. In addition, the study investigates the importance of the spectra acquisition temperature conditions and the physical state of the sample during the analysis, as well as its effect on the FA prediction models," the researchers say.

A hot point of contention for fat testing

Determination of FA profiles has improved as spectroscopic instrumentation improves and as spectroscopists improve their statistical methodology but experimental conditions have often been overlooked. Given that temperature as an analytical condition largely ignored in fat and oil analyses the team has also looked at this parameter in building their model.

Fievez and colleagues have demonstrated that Raman spectroscopic analysis in combination with a fast milk fat extraction procedure could be used to address direct routine applications for the quantification of several OBCFAs. They explain that tight control of experimental conditions would make the analyses reproducible as well as allowing statistical interpretation.

"While the questions of what is the exact pathway under which temperature conditions influence Raman scattering intensities and why it affects only specific Raman shifts still remain and need to be addressed to further increase prediction performance, FT-Raman spectra collected at different temperatures have shown interest to improve predictions of low concentrations of saturated OBCFAs in milk fat," the team concludes.