Measuring intramuscular fat in beef with ultrasonic frequency analysis.

Frequency analysis of Fourier spectra from ultrasonic signals was used for predicting intramuscular fat content of beef tissue. The most significant parameter in the frequency domain for predicting intramuscular fat concentration in beef was the number of local maxima. It represents the discontinuity of the Fourier spectrum caused by inhomogeneous fat concentrations in the longissimus muscle, which had the correlation coefficient .89 (P < .05) when a 2.25-MHz shear probe was used. The optimum frequency for predicting the amount of intramuscular fat content in the longissimus muscle was found to be 1.92 MHz. A multivariate regression model was developed using parameters in the frequency domain as follows: percentage of fat concentration = 1.790 - 2.373x (lower frequency) + .049x (bandwidth) + 1.178x (local maxima) (R2 = .82). Validation demonstrated that the multivariate model in the frequency domain was capable of predicting intramuscular fat concentration with an average of 1.17 percentage of fat error (P < .05). The multivariate model was most appropriate for predicting intramuscular fat below 4%. The mean accuracy of the model in the frequency domain was approximately 79%.