The added value of ADC histogram in characterization of intrauterine masses

Purpose to investigate the added value of ADC histogram parameters combined with the MRI features to differentiate among the uterine sarcomas (US), endometrial carcinomas (EC) and endometrial polyps (EP). The differences of MRI features of 31 cases of US, 51 cases of EC and 27 cases of EP were retrospectively analyzed. The binary logistic regression model was adopted to evaluate the discriminating value of variables. ADC histograms of all lesions were reconstructed and measured, the sensitivity and specificity were calculated, and the diagnostic efficacy was evaluated by AUC value. Binary logistic regression showed that low intensity area on T2WI and presence of cystic degeneration for the characterization between US and EC, the maximum tumor diameter and the muscular invasion for the characterization between US and EP were the most valuable predictive variables (P < 0.001). The ADCmean, ADCmax, ADCmin, Q10, Q25, Q50, Q75, Q90 of US were all significant higher than that of EC, (P < 0.05). Among of them, the ADCmax had the best diagnostic efficiency (AUC = 0.789), the sensitivity and specificity were 77.4