What is the diagnostic accuracy of 3D ultrasound in comparison to MRI for uterine anomalies?

Sally-Anne Jones1, Jane Arezina2, 1Medical Imaging Medical Physics, Sheffield Teaching Hospitals, 2School of Medicine, University of Leeds,


Congenital uterine anomalies (UAs) have a higher incidence in women with infertility or recurrent miscarriage with septate uteri associated with the poorest outcomes (Grimbizis et al., 2016). The lack of a standardised approach for diagnosing and classifying UAs has been widely recognised with examinations including two-dimensional ultrasound (2D US), three-dimensional ultrasound (3D US), hysterosalpingography (HSG), magnetic resonance imaging (MRI) and laparoscopy/hysteroscopy (Saravelos et al., 2008). The aim of the review was to determine the accuracy of 3D US in comparison with MRI, the current imaging gold standard, in the characterisation of uterine anomalies.


Studies performed after 2006 were identified using a comprehensive Ovid search.


Comparison of 3D US and MRI with hysteroscopy/laparoscopy as the reference standard identifies the most accurate procedure for the characterisation of UAs is 3D US with an accuracy of 97.2% to 96% compared with 91.6% to 79% achieved by MRI using the American Fertility Society (1988) classification. Comparison of 3D US against MRI diagnosis alone demonstrates good strength of agreement (0.878 95% CI, 0.775-0.980) using the AFS classification and associated anomalies. Using the European Society of Human Reproduction and Embryology-European Society for Gynaecological Endoscopy (ESHRE-ESGE) (Grimbizis et al., 2013) reported sensitivity of 3D US ranged from 83.3 to 100%, specificities of 88.9 to 100%, positive predictive value (PPV) of 95.5 to 100% and negative predictive value (NPV) of 98.2 and 100%.


The accuracy of 3D US in the characterisation of UAs is superior to MRI. Findings of this review support the recommendation that 3D US should be a necessary step to investigate UAs to achieve a definitive diagnosis, eliminating the need for further investigation in the majority of cases. However, a universally agreed classification system incorporating all variations and offers clear descriptions to facilitate diagnosis and treatment planning is urgently required.

View the Poster Here