Prediction of intramuscular fat by impedance spectroscopy

Multiple multi-frequency impedance measurements during the computer controlled passage of a probe through the M. longissimus dorsi in pork and beef were applied. It was expected that the variability in impedance would correlate with the intramuscular fat (IMF) due to the inhomogeneous distribution of electrolytes and fat. We conducted our experiments in pig carcasses at different, well defined times post-mortem in varying directions of puncture (Experiment 1) and predicted the intramuscular fat in pork and beef by regression (Experiment 2). The highest correlations were obtained in experiment 1 for parameters characterising the variability of the impedance 24h post-mortem and insertion of the probe through the back fat and muscle towards the body cavity (r=0.54-0.79, P<0.001). Both of these were chosen for the measurements in experiment 2. Regression for the prediction of IMF in pork and beef in experiment 2 resulted in R(2) values of 0.12 and 0.48, respectively; and in RMSE values of 0.67 and 0.64, respectively. The correlation between the predicted and the IMF analysed by n-hexane extraction or Near Infrared Transmission varied from 0.28 to 0.69 (P<0.001) depending on species and breed. A selection of the carcasses for high IMF (above a certain threshold) using the impedance measurements agreed poorly with the analysed IMF. Depending on the level of IMF within a breed, low IMF contents were often over-predicted (3.4-92.7%) or high IMF contents were estimated as too low (0-80.9%). Breed specific regression equations could improve the accuracy. These data indicate that the selectivity of the impedance method in the configuration presented here is not yet sufficient for practical use.