Ankara Universitesi Veteriner Fakultesi Dergisi, cilt.60, sa.1, ss.59-65, 2012 (SCI-Expanded)
ROC curve is a commonly used method for performance evaluation and comparison of diagnostic tests. ROC curve, by using specifity and sensitivity values, determines best cut-off points that categorize experimental groups. The aim of the study is, to introduce the roc analysis method for determination of new diagnostic test performance, accuracy and discrimination of healthy and unhealthy animals and disease types. A new diagnostic method, computer based echostructure software and analyse values of ultrasonographic images of cows with chronic endometritis has been used as application data. Diagnostic rates were calculated for obtained echostructure variables (mean gradient, homogenity, contrast, mean grey value) in types of endometritis (E1, E2, E3). 11,44 and 58,60 threshold values of mean gradient and mean grey value give 75% sensitivity and 70% specificity at E1, 0,07 value of homogenity gives 91% sensitivity and 30% specificity at E2, mean gradient, homogenity and mean grey values are important parameters at E3 were determined. It was shown that statistical analysis by using ROC curve can be a complementary calculation method for discrimination of severity of chronic endometritis.