Diagnosis of the most frequent benign ovarian cysts: is ultrasonography accurate and reproducible?

OBJECTIVE: To evaluate the reproducibility and the accuracy of B-mode ultrasonographic features of three different kinds of benign ovarian cysts: ovarian endometrioma, mature teratoma, and serous cyst.
METHODS: Digitally stored B-mode sonographic images of 98 women submitted to surgery for the presence of an adnexal mass were evaluated by five different examiners with different degrees of experience. The histological type of each mass was predicted on the basis of the B-mode typical benign findings, as in the case of endometrioma (groundglass endocystic pattern), cystic teratoma (echogenic pattern with or without acoustic shadow), and serous cyst (anechoic cyst without endocystic vegetations). To assess the reproducibility of the B-mode findings, intraobserver and interobserver agreements were calculated using the kappa index.
RESULTS: The intraobserver agreement was good or very good for all examiners and for all patterns (kappa = 0.71-1) except for the dermoid cyst, which showed moderate agreement (kappa = 0.42) for the highly experienced operator. The interobserver agreement was good for all experts for endometrioma (kappa = 0.66-0.78) and for serous cyst (kappa = 0.82-1), whereas it was moderate or good for cystic teratoma (kappa = 0.51-0.72). Interobserver agreement between experts and highly experienced operators was fair (kappa = 0.33-0.36) for teratoma and good or very good for endometrioma (kappa = 0.70-0.83) and serous cyst (kappa = 0.76-0.82). For different kinds of cysts, the accuracy was comparable among different operators.
CONCLUSIONS: Typical features of benign masses using grayscale transvaginal ultrasonography are reproducible even in moderately experienced examiners, although more experience was associated with better interobserver agreement. The diagnostic performance of different operators with different degrees of experience is similar.