ABSTRACT
BACKGROUND: The aim was to evaluate and compare inter-observer reproducibility by magnetic resonance imaging (MRI), transvaginal ultrasonography (TVS), hysterosonographic examination (HSE) and hysteroscopy (HY). METHODS: Different observers consecutively evaluated MRI, TVS, HSE and HY independently in 51 pre-menopausal women, who underwent hysterectomy for benign diseases. RESULTS: Inter-observer agreement (kappa) was as follows: Exclusion of uterine cavity abnormalities: MRI 0.97, TVS 0.68, HSE 0.48 and HY 0.63; submucous myomas: MRI 0.97, TVS 0.59, HSE 0.60 and HY 0.67; polyps: MRI 0.49, TVS 0.48, HSE 0.35 and HY 0.50; identification of myometrial myomas: MRI 0.97, TVS 0.74; adenomyosis: MRI 0.73 and TVS 0.38. Mean difference between observers in number of observed myomas was (absolute values) MRI 0.58, TVS 0.93. Agreement on evaluation of abnormalities in the uterine cavity, submucous myomas, number of myomas and adenomyosis was significantly greater by MRI than by any of the other techniques, whereas agreement was in line by TVS, HSE and HY. CONCLUSIONS: Inter-observer disagreement reached substantial levels only for exclusion of uterine cavity benign abnormalities by HY, TVS and HSE. Strategies should be adopted to reduce observer variation of common gynaecological imaging techniques or the less observer-dependent MRI technique could be favoured.
