Role of three-dimensional transvaginal sonography compared with magnetic resonance imaging in diagnosis of Mullerian duct anomalies

Mullerian duct anomalies are a heterogenous group of congenital anomalies due to abnormalities during the ductal development, fusion, or alteration of septal resorption. These lead to the occurrence of different uterine anomalies. The present study aimed to assess the value of three-dimensional transvaginal sonography (3DTVS) in comparison with pelvic MRI to diagnose uterine anomalies. We prospectively evaluated 30 female patients, from January 2016 to May 2017. Their ages ranged from 18 to 40 years. Cases were referred from obstetrics and gynecology clinic with suspected uterine anomalies by 2DUS or with HSG done for infertility workup. 3DTVS and MRI examination were done for all patients. The final diagnosis of patients by MRI according to The European Society of Human Reproduction and Embryology–European Society for Gynaecological Endoscopy consensus classification included 2/30 (6.7%) patients classified as class U0, 4/30 (13.3%) patients classified as class U1, 17/30 (56.7%) cases classified as class U2, 2/30 (6.7%) patients classified as class U3, 3/30 (10%) patients classified as class U4, and 2/30 (6.7%) patients were classified as class U5. There was a significant strong agreement between the diagnosis by 3D-transvaginal ultrasound and MRI (P < 0.01). 3DTVS proved to be highly accurate in the diagnosis of uterine anomalies and showed strong agreement with MRI, as both can provide valuable information about both the internal details and the external contour of the uterus.


Background
Mullerian duct anomalies are heterogenous spectrum of congenital anomalies that result from abnormalities during the ductal development, ductal fusion, or alteration of septal resorption. These abnormalities lead to the occurrence of different uterine anomalies [1].
Early detection and proper diagnosis of uterine anomalies are needed to distinguish operable from inoperable cases. Most of the patients present with infertility, repeated first-trimester abortion, fetal intrauterine growth restriction, and obstetric complications [1].
Several trials have been made for proper and accurate classification of Mullerian duct anomalies. The American Society for Reproductive Medicine (ASRM) classification is considered the most commonly used one ( Fig.  1) [3]. The European Society of Human Reproduction and Embryology-European Society for Gynaecological Endoscopy (ESHRE-ESGE) is used to limit the subjective diagnosis of the American classification. It helps in the differentiation of septate uterus from similar anomalies, regardless of the absolute morphological characteristics [4]. The arcuate uterus entity is not present in the ESHRE-ESGE classification. This classification includes uterine anomaly main classes and subclasses as well as cervical and vaginal anomalies as coexistent subclasses, as shown in Tables 1 and 2 [6,7].
There are different imaging modalities that can be used for the diagnosis of uterine malformations. Hysterosalpingography (HSG) and hysteroscopy are considered good modalities to assess the uterine cavity. Laparotomy and laparoscopy are also used for providing information about the external contour and uterine cavity [7].
Pelvic magnetic resonance imaging (MRI) has also been proven to be excellent in the diagnosis of Mullerian duct anomalies due to high soft tissue resolution. MRI examination is more expensive and less available than other imaging modalities [8].
Three-dimensional ultrasound represents a valid alternative or adjunct to pelvic MRI. It is less expensive than MRI and being better tolerated by patients. Threedimensional transvaginal sonography (3DTVS) provides image quality like those provided by magnetic resonance imaging [9].
So, the aim of our study was to assess the value of 3DTVS in comparison with that of pelvic MRI to diagnose cases with uterine anomalies.

Patients
The study included 30 married female patients. They were referred to the Radiology Department from the obstetrics and gynecology clinic of a university hospital in the period from January 2016 to May 2017. The study was approved by the ethical committee of our institution. All participants signed informed consent prior to the examinations. Most of the cases were presented with infertility and some cases with repeated abortion.
They were initially suspected to have uterine anomalies by 2D US or with HSG done for infertility workup, and then 3DTVS and MRI examinations were done for all patients. Exclusion criteria included cases with general contraindications to MRI examination and cases with prior gynecological operations.

3DTVS
We used (GE Voluson S6) ultrasound machine, equipped with endocavitary 3D probe RAB2-6 5-9 MHz. The uterus was first examined by a two-dimensional ultrasound in a strict mid-sagittal view showing the uterus from the fundus to the cervix, with the In any uterus with a large transverse diameter, another acquisition was made from a transverse section, where we used a 90°sweep angle to obtain a multiplanar view of the uterus. The volume data were observed in three orthogonal planes and processed to do the needed reconstructions. The time of examination required for the diagnosis of uterine anomalies using a three-dimensional US, varied from 4 min to maybe up to 15 min in some complex cases.

MRI imaging
MRI was done by Siemens Magnetom Aera 1.5 T using senes body coil 18. Before the examination, all patients were instructed to have a full bladder. Preliminary scout localizers in axial, coronal, and sagittal planes were done. Routine conventional MR study including axial T1 and T2, sagittal T2, and coronal T2 was done for all patients. Sequences were oriented according to the axis of the uterus. Coronal T2-weighted FSE was

Evaluation of findings
The uterine morphology was interpreted in the coronal plane of the uterus with the interstitial portion of both Fallopian tubes as points of reference. We measured the internal indentation in the midcoronal plane, and according to its length, we classified cases into different classes. The external cleft also was measured in the midcoronal plane. The wall thickness was evaluated in the mid-sagittal plane (Fig. 2). The findings were analyzed using the ESHRE-ESGE consensus classification as shown in Table 1 and Fig. 3.

Statistical analysis
Scale data were presented as mean and standard deviation. Categorical data were presented as number (frequency) and percent. The agreement between both three-dimensional transvaginal ultrasound and MRI was presented as crosstab and kappa index of agreement. All significant P values were less than 0.05. The accuracy was calculated in the form of sensitivity, specificity, negative predicted value, and positive predicted value.
The diagnostic accuracy and agreement of 3DTVS compared to MRI were illustrated in (Tables 4 and 5). There was a significant strong agreement between the diagnosis by 3D transvaginal ultrasound and MRI examination.

Discussion
Uterine anomalies are different groups of malformations with a broad spectrum of presentations [11]. Mullerian anomalies are initially suspected by 2DUS and/or with HSG done for infertility workup [8].
MRI is the most accurate imaging study for uterine anomalies; it reached a sensitivity and specificity of 92% and 100%, respectively, compared to the results of combined hysteroscopy and laparoscopy [12].
In this study, we compared the efficiency of 3DTVS in diagnosing uterine anomalies to the MRI study, considering MRI as the standard reference. We used the ESHRE-ESGE consensus classification which provides objective parameters to classify uterine anomalies. This classification allows an accurate distinction between the different types of uterine anomalies with 3DTVS and MRI. This agrees with the study done by Graupera et al. [13] who reported that the ESHRE-ESGE classification provides objective parameters for diagnosing Mullerian duct anomalies (MDA).
The ESHRE-ESGE classification is more valuable than the ASRM classification. It shows uterine malformation main classes and subclasses. Cervical and vaginal anomalies are included as separate subclasses. So, ESHRE-ESGE classification is more helpful in the diagnosis of complex anomalies of the female genital system [5].
We agreed with Robins et al. [14] who reported that septate uterus is the commonest uterine anomalies. In the present study, we found that septate uterus represented 17/30 (56.7%) of our study population.
Septate uterus shows different morphological changes in both classifications. The ratio of internal fundal indentation to myometrial thickness was significantly lower in the ESHRE-ESGE classification than in the ASRM classification. Diagnosis of septate uterus by ESHRE-ESGE classification resembles the arcuate or normal uterus in the ASRM system [5]. The most important point in these classifications is to limit the overdiagnosis and improper treatment in cases of uterine anomalies. ESHRE-ESGE classification can guide the gynecologist to do the appropriate surgical treatment [10].
The limited positive predictive value of 3DTVS in our study was met in class U3 where four cases were included: 2/4 (50%) were true bicorporeal and 2/4 (50%) were wrongly diagnosed by 3DTVS as bicorporeal and diagnosed by MRI as septate (class U2), sensitivity = 100%, specificity = 92.8%, PPV = 50%, and NPV = 100%. We attributed this to the limited tissue   resolution of US compared to MRI, and ultrasound is operator dependent. If the midcoronal plane is not ideal, this will give the wrong diagnosis. The problem of misdiagnosing the two cases of septate uterus by 3DUS in our study might be due to improper location of the line of the US beam during 3D reformatting which was at the level of the septum and not at the fundal level. So, it was better to move the line a little bit higher to assess the external contour of the uterus. We found that 7/30 (23.3%) cases had associated renal and other anomalies; this coincides with Li et al. [15] who reported that concomitant renal anomalies are found in 29% of Mullerian duct anomalies cases.
In this study, the overall agreement between 3DTVS and MRI was 86.7% and the agreement coefficient (kappa) was 0.743; this coincided with the work of Bermejo et al. [7] who reported good agreement between 3D transvaginal US and MRI in diagnosing Mullerian duct anomalies, with a kappa index of 0.880.
The limitations in our study were sampling bias, since we did 3DTVS only to patients with suspected uterine anomalies and some uterine anomalies that had a limited number in our study.

Conclusion
3DTVS proved to be highly accurate in the diagnosis of uterine anomalies and showed strong agreement with MRI, as both can provide valuable information about both the internal details and external contour of the uterus.