Predicting intramuscular fat in beef longissimus muscle from speed of sound

Abstract
The ultrasonic A-mode method was investigated for intramuscular fat measurement of beef. Cubes of meat samples that had different marbling scores were used as specimens. Ultrasonic speed decreased with increasing fat concentration. The correlation coefficient was −.82 (P < .05). Conversely, the correlation coefficient between visual marbling score and fat concentration was .70 (P < .05). Quantitatively, from the first derivative of a nonlinear function, it was concluded that the speed decreased at a rate of 2.69 m/(s-% fat) as a function of intramuscular fat. A nonlinear regression model to predict intramuscular fat content in beef was developed as follows: percentage of fat concentration = 7132.1574 − 9.1222 × (speed) + 2.91803 × 10−3 × (speed)2 (R2 = .81). The nonlinear model was capable of predicting intramuscular fat concentration with 90% accuracy (> 8% fat) and 76.4% accuracy (< 8% fat), respectively (P < .001).