Characterization of nutritional parameters in bovine milk by Raman spectroscopy with least squares modeling

The macronutrient constitution of bovine milk was investigated by Raman spectroscopy involving 830nm excitation at 300 mW from 600-1800cm(-1) with a probe. Bovine milk was analyzed for total proteins, casein, total fat, and lactose by standard assays. The concentrations of these components were estimated by correlating these assays with the fitting coefficients based on least-squares fitting of lactose, casein, triolein, and stearic acid and milk spectra to prepare calibration curves. The model was employed for the analysis of milk from local markets. The spectra of milk were dominated by bands of proteins, lipids, and carbohydrates. There was a strong correlation between the concentrations of total proteins and casein (r=0.75 and 0.73, respectively), and very strong correlation for total fat and lactose (r=0.93 and 0.91, respectively) with the standard errors of prediction of 0.49, 0.38, 0.61, and 0.27mg/dL, respectively. The milk contained lower concentrations of lactose and protein than in the nutritional facts table. The amount of saturated fat was close to the nutritional value with more unsaturated fat. A single Raman spectrum was employed to characterize the composition of milk and may have application for rapid quality control.