Three-dimensional ultrasound and magnetic resonance imaging assessment of cervix and vagina in women with uterine malformations

Objectives: To investigate the accuracy of three-dimensional ultrasound (3D-US) with respect to magnetic resonance imaging (MRI), and compared to clinical examination, in the assessment of cervix and vagina in women with uterine malformations.

Methods: In this prospective study, 16 patients diagnosed with uterine malformation with cervical involvement underwent 3D-US examination. The acquisition of cervical volumes was transvaginal, with four cases repeated in the peri-ovulation period, while vaginal volumes were acquired by transperineal imaging following filling of the vagina with gel. MRI was performed in 13 patients using endovaginal gel. All cases underwent clinical examination, comprising bimanual gynecological examination and speculoscopy. Diagnostic concordance of each of the methods with the gold standard was calculated.

Results: 3D-US cervical examinations revealed 12 cases of duplicate cervix, two of complete septate cervix and two of incomplete septate cervix. Images of the cervical canal in the peri-ovulation period were judged subjectively to be better in quality, but did not lead us to change any diagnosis. 3D-US vaginal examinations revealed four cases with a vaginal dividing wall and two with a blind hemivagina. None of the 3D-US findings contradicted the clinical findings of the cervix; however, clinically we observed two cases with vaginal dividing wall that had not been diagnosed with 3D-US. MRI diagnosed nine cases of duplicate cervix, three of complete septate cervix, one of incomplete septate cervix, five of vaginal dividing wall and two of blind hemivagina. One case diagnosed as complete septate cervix was in fact a duplicate cervix on 3D-US and on clinical examination. Compared with the gold standard, both 3D-US and MRI were highly efficient in the diagnosis of anomalies of the cervix and vagina. The overall diagnostic concordance of 3D-US with clinical examination (kappa, 0.84; 95% CI, 0.62-1) was slightly inferior to that of MRI with clinical examination (kappa, 0.9; 95% CI, 0.72-1), but this difference was not statistically significant.

Conclusions: The acquisition of isolated cervical volumes, without including the uterus, defines the extent of the ectocervix and the limits of the cervical canal in uterine malformations. The use of endovaginal gel makes possible the diagnosis of associated vaginal anomalies with 3D-US.