Henrik Toft Pedersen

Application of chemometrics to low-field 1H NMR relaxation data of intact fish flesh

The possibilities for application of low-field 1H nuclear magnetic resonance (NMR) as a rapid method for simultaneous assessment of basic quality parameters in fish were explored. In a first experiment, 200 salmon (Salmo salar) samples mapping the variation over an entire fish were measured by NMR and subsequently analysed for oil or water content by standard chemical methods. In a second experiment, 58 differently thawed cod (Gadus morhua) samples were measured by NMR and subsequently analysed for water-holding capacity. Correlations between chemical data and NMR data were evaluated using partial least squares (PLS) regression on complete relaxation curves and compared with conventional regression models on exponential fitting parameters. Predictions on an independent test set were superior for the PLS regression models, with optimal prediction errors of 12 g kg−1, 6 g kg−1 and 3.9% for oil and water content in fresh salmon flesh and water-holding capacity in thawed cod flesh respectively. Thus rapid, non-invasive low-field NMR can be used to simultaneously determine both oil and water content of fish flesh. Furthermore, it can predict water-holding capacity of cod flesh, with an R2 of 0.9 over the range 30–90%, as determined by a centrifuge test. © 1999 Society of Chemical Industry

Application of the NMR-MOUSE to food emulsions

The application of the NMR-MObile Universal Surface Explorer (NMR-MOUSE) to study food systems is evaluated using oil-in-water emulsions, and the results are compared to those obtained using a conventional low-field NMR (LF-NMR) instrument. The NMR-MOUSE is a small and portable LF-NMR system with a one-sided magnet layout that is used to replace the conventional magnet and probe on a LF-NMR instrument. The high magnetic field gradients associated with the one-sided MOUSE magnet result in NMR signal decays being dominated by molecular diffusion effects, which makes it possible to discriminate between the NMR signals from oil and water. Different data acquisition parameters as well as different approaches to the analysis of the NMR data from a range of oil-in-water emulsions are evaluated, and it is demonstrated how the concentration of oil and water can be determined from the NMR-MOUSE signals. From these model systems it is concluded that the NMR-MOUSE has good potential for the quantitative analysis of intact food products.

Comparative Chemometric Analysis of Transverse Low-field 1H NMR Relaxation Data

1 Abstract Transverse relaxation data from low-field (23.2 MHz) 1H NMR was investigated by different data analytical approaches. The quantitative information in the relaxation data of 200 intact salmon samples with respect to overall water and fat content was evaluated by four different data analytical methods: (a) bi-exponential curve fitting followed by linear regression, (b) forward selection of variables followed by multiple linear regression (FS-MLR), (c) partial least squares regression (PLS) and (d) non-negative alternating least squares regression (NN-ALSR). The investigation demonstrates that the quantitative prediction performance is significantly enhanced (reduction of the prediction error from 14% to 34%) by the use of multivariate chemometric procedures such as PLS. While PLS is an extraordinarily robust and efficient algorithm its strictly orthogonal latent variables suffer from a difficult qualitative interpretation. NN-ALSR and FS-MLR which exhibit a quantitative performance comparable to PLS do not suffer from this problem, but are more unstable and ineffective data analytical techniques.

Application of Multi-Way Analysis to 2D NMR Data

Two- and three-dimensional or even higher-dimensional NMR spectroscopy is changing from specialised techniques to more commonly used ones. As the complexity of the acquired NMR data increases, the task of analysing these data constantly becomes more and more demanding and new methods are required to facilitate the analysis. With one-dimensional NMR data multivariate data analysis has proven to be a strong tool, but how should one analyse higher-dimensional NMR data in order to extract as much relevant information as possible without having to break data down into smaller dimensions and thus lose the inherent structure? A class of multivariate data analytical techniques called multi-way analysis encompass techniques that have been designed to handle and analyse such data structures directly. In this paper, the theory of some of the most commonly used multi-way methods will be described and examples of their application to three-way arrays of NMR data reported in the literature will be given. The focus will be on the generic principles of multi-way analysis using three-way data in order to gently introduce the new concepts.