OC19.03: Contrast‐enhanced sonography in the preoperative diagnosis of ovarian tumors considered difficult to classify by ultrasound examiners

Abstract

Methods: 109 women with adnexal masses underwent transvaginal gray scale and 3D power Doppler ultrasound examination using a GE Voluson 730 Expert system. After 12–18 months the volumes were analysed twice by two independent experienced sonologists who used IOTA terms and definitions to describe the mass. The assessors also had to classify the mass as benign or malignant using pattern recognition and to suggest a likely histological diagnosis. The time between the two assessments by the same sonologist was 1 to 6 months. Results: There were 90 benign, 8 borderline and 11 malignant adnexal masses. There were no systematic differences in measurement results within or between observers. Differences in measurement results did not change with the magnitude of the measurements. Intra-observer agreement in terms of Kappa value was 0.90 (agreement 92%) for specific histological diagnosis, 0.90 (96%) for malignancy, 0.84 (88%) for type of mass, 0.79 (85%) for tumor size (categorized), 0.71 (86%) for irregular walls and 0.58 (72%) for number of papillary projections. Kappa values for inter-observer agreement were 0.78 (85%) for specific histological diagnosis, 0.81 (93%) for malignancy, 0.76 (83%) for type of mass, 0.89 (92%) for tumor size, 0.53 (80%) for irregular walls and 0.35 (60%) for number of papillary projections. Conclusions: Intraand inter-observer agreement was good with regard to size and type of mass, specific histological diagnosis, and malignancy. Inter-observer agreement was poorest for irregularity and the number of papillary projections. This study shows that repeated analyses of 3D ultrasound volumes of adnexal masses by the same or different observers may yield different results, but that the diagnosis suggested using pattern recognition will often be the same.