BENIGN AND MALIGNANT OVARIAN MASSES - SELECTION OF THE MOST DISCRIMINATING GRAY-SCALE AND DOPPLER SONOGRAPHIC FEATURES

PURPOSE: To determine the gray-scale and Doppler sonographic features that best enable discrimination between malignant and benign ovarian m asses and develop a scoring system for accurate diagnosis with these f eatures. MATERIALS AND METHODS: Gray-scale and Doppler sonographic fea tures of 211 adnexal masses were correlated with the final diagnosis; the most discriminating features for malignancy were selected with ste pwise logistic regression. RESULTS: Twenty-eight masses were malignant and 183 benign. All masses with a markedly hyperechoic solid componen t or no solid component were benign. For masses with a nonhyperechoic solid component, additional features that allowed statistically signif icant discrimination of benignity from malignancy were, in decreasing order of importance, (a) location of flow at conventional color Dopple r imaging, (b) amount of free intraperitoneal fluid, and (c) presence and thickness of septations. A scoring formula that made use of values based on the logistic regression equation had an area under the recei ver operating characteristic curve of 0.98 +/- 0.01. The cutoff score with the highest accuracy had a sensitivity of 93% and specificity of 93%. CONCLUSION: A solid component is the most statistically significa nt predictor of a malignant ovarian mass. A multiparameter scoring sys tem that uses three gray-scale and one Doppler feature, developed by m eans of stepwise logistic regression, has high sensitivity and specifi city for predicting malignancy.