Ultrasonic elastography to predict beef tenderness

Abstract

Ultrasonic elastography was used to predict chemical and mechanical measures of beef tenderness. Beef carcasses (n=30) were obtained from animals of known genetic background of three breed types (3/4 Angus (A) x 1/4 Brahman (B), 1/4 A x 3/4 B, and F2A x B crosses). The right side of each carcass was electrically stimulated. Four muscles (semimembranosus (Sm), semitendinosus (St), biceps femoris (Bf, and triceps brachii (Tb)) were obtained from both the electrically stimulated (ES) and non-electrically stimulated (NS) side. Additionally, the longissimus muscle (Ld) was obtained from the electrically stimulated side of each carcass. Warner-Bratzler shear force (kg) of steaks aged in refrigerated storage for 2, 14, 28 and 42 days postmortem, calpastatin activity (mg/g) at 24 hours postmortem, sarcomere length (mm), total collagen (mg/g), soluble collagen (%), moisture (%), and lipid (%) were determined as measures of beef tenderness. Two gray-scale images, called elastograms, were obtained for each sample. Fourteen textural features (f 1-f 14) were extracted from each elastogram at four distances (1, 2, 5, and 1 0 neighboring resolution cell pairs) and five angles (0, 45, 90, 135 [degrees] and average). Stepwise regression (P<. 15) was used to determine the ability of the textural features to predict the tenderness measurements. Correlations between selected textural features and Warner-Bratzler shear force also were analyzed. Textural features within the Ld muscles predicted shear force at 2 days (R 2 =.47), shear force at 14 days (R 2 =.40), shear force at 28 days (R 2 =.26), fat (R 2 =.25), moisture (R 2 =.23) and collagen solubility (R 2 =.12). Regression equations for other muscles had low predictability (R 2 =.04 to .39). The textural features f 12 and f 1 3 were significantly correlated to each other across distances, therefore accounting for similar variation in elastograms. The parameters f I and f 12 were extremely variable across distances as indicated by residual standard deviations that were larger than the values for that parameter. These results indicated that ultrasonic elastography has potential to predict beef tenderness, but further research into feature extraction techniques are needed.