The spatial composition of porcine adipose tissue investigated by multivariate curve resolution of near infrared spectra: Relationships between fat, the degree of unsaturation and water

Abstract

This study represents a continuation of a previously performed study by Sørensen, et al. (2012) about the use of and development of an online method for determining the iodine number in porcine adipose tissue by the use of near infrared spectroscopy. In the original study, the fat tissue was measured as a function of the depth from the skin and thus represent spatial or 2D near infrared data, and if measured across a porcine carcass it represents 3D data of the pig. These data structures, which were largely ignored in the original publication, are here explored by multivariate curve resolution. The massive amount of near infrared data (11,329 spectra) was unfolded from the true five-way array (carcass × horizontal position × vertical position × depth × spectral data points, 30 × 6 × 15 × 6 × 204) into a two-way matrix (sample × spectra, 11329 × 204) (maintaining spatial data in a separate array) and multivariate curve resolution applied. A two-component multivariate curve resolution model proved to fit the data nearly perfectly (99.69% X variance explained). By concentration the highest component was found to be fat, and the second component was found to be water. Surprisingly, the latter was found to co-vary strongly with the iodine value of the fat ( r = −0.83). The resulting multivariate curve resolution model was folded back into its original spatial dimension to produce maps showing distribution of the constituent concentrations in the adipose tissue.