Tissue characterization for beef grading using texture analysis of ultrasonic images

The primary factors in determining beef quality grades are the amount and distribution (or marbling) of intramuscular fat. Under the current USDA Grading system, the quality is subjectively graded by certified inspectors. There is a growing demand in the meat industry for an objective system of evaluating the quality of beef carcasses as well as live animals. A real-time ultrasound technique was used to characterize intramuscular fat in the longissimus dorsi (rib-eye) muscle of 126 live animals. Two approaches of B-mode image texture analysis, namely spacial gray-level dependence matrices and gray-level run-length matrices, were used. The significant texture parameters were used to develop models for predicting intramuscular %fat. With validation testing of these models, from more than 75 percent of the images, the %fat values were predicted to within 1.5 percentage and from about 68 percent of the images, the %fat was predicted to within one percent. The preliminary results showed a good potential of ultrasound and texture analysis for tissue characterization and objectively evaluating beef quality