Optimizing the accuracy of detecting a functional corpus luteum in dairy cows

The objectives were to evaluate the accuracy of detecting a functional CL by transrectal palpation and ultrasonography, and to optimize the accuracy of detecting a functional CL by ultrasonography in Holstein cows. In Experiment 1, four veterinarians performed transrectal palpation in 1250 cows at 37 d in milk (DIM), two veterinarians repeated transrectal palpation in 823 cows at 58 DIM, and one veterinarian performed 206 ultrasonographic examinations at 37 DIM. In Experiment 2, 987 and 983 ultrasonographic examinations were performed at 21 and 24 d after AI by one veterinarian for detection and measurement of CL. Cows with a blood progesterone concentration > or =1ng/mL were assumed to have a functional CL. Sensitivity and specificity were optimized using receiver operating characteristic analysis. In Experiment 1, sensitivity of transrectal palpation for diagnosing a functional CL ranged from 33.3 to 59.9% at 37 DIM and from 48.3 to 68.4% at 58 DIM, whereas specificity ranged from 76.7 to 93.2% at 37 DIM and from 73.3 to 86.7% at 58 DIM. Sensitivity and specificity for ultrasonography were 89.4 and 45.7%, respectively. In Experiment 2, the sensitivity and specificity of ultrasonography were 97.3 and 38.1% at 21 d after AI, and were 97.9 and 51.0% at 24 d after AI. Sensitivity and specificity were optimized using a cutoff diameter of 23mm at 21 d and 22mm at 24 d, which resulted in sensitivity and specificity of 87.2 and 83.0% at 21 d, and 89.5 and 89.4% at 24 d after AI, respectively. Sensitivity was low and specificity was high for transrectal palpation, whereas ultrasonography resulted in high sensitivity and low specificity. Using a cutoff diameter during ultrasonography improved accuracy of detection of a functional CL compared with either ultrasonography without cutoff or transrectal palpation.