Abstract
Generalized two-dimensional (2D) correlation spectroscopy has been applied to analyze near-infrared (NIR) spectra of milk with different protein and fat concentrations. The NIR spectra of milk show rather poor signal-to-noise ratios compared with those of a protein or fat solution and have changing baselines from one spectrum to another. Poor signal-to-noise ratio and variations in baseline are common problems for NIR spectra of real-world samples. This study aims at expanding the utility of generalized 2D correlation spectroscopy to complicated multicomponent biological systems. In order to overcome the above two problems, we have employed multiplicative scatter correction (MSC) and smoothing as pretreatment procedures of the milk spectra selected for the calculation of 2D NIR correlation. 2D synchronous correlation spectra in the 2000–2400 nm region constructed from protein or fat concentration-dependent spectral changes of milk sharply enhance bands assignable to proteins or fats, respectively. It has been found that a power spectrum along the diagonal line in a synchronous spectrum very effectively shows the contribution of a particular component to the NIR spectra of milk. In fact, for example, the power spectrum for the fat concentration-dependent spectral changes of milk is very close to an NIR spectrum of fat itself. Two-dimensional asynchronous correlation spectra demonstrate the existence of bands that cannot be identified even by calculation of second derivatives and chemometrics analysis of the spectra. The asynchronous spectra also elucidate interaction between fats, proteins, and water.
Get full access to this article
View all access options for this article.