The impossible made possible: providing a detailed topological map of chemical classification parameters for pork carcasses with on-line slaughter line instrumentation

The authors have previously demonstrated an apparatus and method for on-line evaluation of the quality of fat on a pig slaughter carcass at the abattoir using NIR spectroscopy and multivariate modeling (Sørensen et al., Applied Spectroscopy, 2012). The methodology developed provides precise estimation of the iodine value of the subcutaneous tissue in its entire thickness at full production speed, a parameter that ultimately dictates the commercial value of the carcass. The instrument is the first of its kind, providing a worlds-first paradigm shift in on-line carcass classification by bringing true chemical on-line measurements of carcass traits directly to the slaughter floor of the abattoir.
We have furthermore produced a method for recording a topological map of the fat quality of the pork carcass (Sørensen et al. NIR News, 2013). Based on correlations, it provides an understanding of how the fat quality is distributed across the carcass and depends on factors such as feeding regimes. The abattoir can use the map to predict the iodine value in all primal cuts of the carcass from one single measurement in the neck region. Thus, a single 3 second on-line measurement can elucidate the quality profile for the entire carcass.
In the present work, we present the results of deriving further chemical traits of the carcass, again set in a topological framework, from NIR spectral data acquired on-line. The analyses are conducted in two levels: first, the abundance of individual fatty acids that comprises the subcutaneous tissue is modeled from the NIR spectra using GC-MS reference analyses. From this analysis, the individual fatty acid distributions and correlations are presented as a topological function of the carcass. Secondly, multivariate curve resolution is applied on a large number of spectra and it is demonstrated how the ratios of specific chemical groups of compounds known to be present in the fat can be modeled and the predictions represented topologically.
30 carcasses were selected from the production stock at an abattoir.. Each carcass had a 5 cm x 5 cm grid of measurement points projected onto one side, and the NitFom measurement instrument (Carometec A/S, Herlev, Denmark) was applied at each grid point, acquiring measurements at 6 depth locations. In average, a carcass side would yield 15 horizontal and 5 vertical grid points, and thus, the experiment, a five-way data matrix (carcass × vertical position × horizontal position × depth × wavelength) providing a total of 15,906 NIR spectra recorded in the range from 1100 nm to 2200 nm. Fatty acid reference analysis was performed on a subgroup of selected fat samples by transesterification and subsequent quantification on GC-MS.
The work presents prediction correlations of 0.60-0.85 for 14 fatty acids. In addition, it is found that changes over depth is highly significant and depends on carcass location. MCR modeling of the spectra isolates at least two groups of fat constituents – water and fat. Within the fat group, the MCR-provided concentrations are shown to partially correlate with the individual fatty acid concentration and thus represent varying fat properties.