Rapid and quantitative detection of the microbial spoilage in milk using Fourier transform infrared spectroscopy and chemometrics

Abstract

Microbiological safety plays a very significant part in the quality control of milk and dairy products worldwide. Current methods used in the detection and enumeration of spoilage bacteria in pasteurized milk in the dairy industry, although accurate and sensitive, are time-consuming. FT-IR spectroscopy is a metabolic fingerprinting technique that can potentially be used to deliver results with the same accuracy and sensitivity, within minutes after minimal sample preparation. We tested this hypothesis using attenuated total reflectance (ATR), and high throughput (HT) FT-IR techniques. Three main types of pasteurized milk – whole, semi-skimmed and skimmed – were used and milk was allowed to spoil naturally by incubation at 15 °C. Samples for FT-IR were obtained at frequent, fixed time intervals and pH and total viable counts were also recorded. Multivariate statistical methods, including principal components-discriminant function analysis and partial least squares regression (PLSR), were then used to investigate the relationship between metabolic fingerprints and the total viable counts. FT-IR ATR data for all milks showed reasonable results for bacterial loads above 10⁵ cfu ml⁻¹. By contrast, FT-IR HT provided more accurate results for lower viable bacterial counts down to 10³ cfu ml⁻¹ for whole milk and, 4 × 10² cfu ml⁻¹ for semi-skimmed and skimmed milk. Using FT-IR with PLSR we were able to acquire a metabolic fingerprint rapidly and quantify the microbial load of milk samples accurately, with very little sample preparation. We believe that metabolic fingerprinting using FT-IR has very good potential for future use in the dairy industry as a rapid method of detection and enumeration.