Three-dimensional ultrasound in the diagnosis of Müllerian duct anomalies and concordance with magnetic resonance imaging

Transcription

1 Ultrasound Obstet Gynecol 2010; 35: Published online in Wiley InterScience ( DOI: /uog.7551 Three-dimensional ultrasound in the diagnosis of Müllerian duct anomalies and concordance with magnetic resonance imaging C. BERMEJO*, P. MARTÍNEZ TEN, R. CANTARERO*, D. DIAZ*, J. PÉREZ PEDREGOSA, E. BARRÓN, E. LABRADOR and L. RUIZ LÓPEZ *Gabinete Médico Velazquez, DELTA-Ultrasound Diagnostic Center in Obstetrics and Gynecology, Department of Obstetrics and Gynecology, Hospital La Zarzuela and Centro de Resonancia Magnética de Alto Campo Abirem, Madrid and SESCAM, Spain KEYWORDS: 3D US; congenital uterine malformations; magnetic resonance; MRI; Mullerian; Müllerian anomalies; Müllerian duct anomalies; three-dimensional ultrasonography; UM; uterine malformations ABSTRACT Objectives To demonstrate the value of three-dimensional (3D) ultrasound in the diagnosis of uterine malformations and its concordance with magnetic resonance imaging (MRI). Methods This study included 286 women diagnosed with uterine malformation by 3D ultrasound, having been referred to our clinics on suspicion of uterine malformation following clinical and/or conventional twodimensional ultrasound examination. With the exception of three with intact hymen, patients underwent both bimanual examination and speculoscopy before and/or after sonography. MRI was performed in 65 cases. We analyzed the diagnostic concordance between the techniques in the study of uterine malformations. Results Using 3D ultrasound we diagnosed: one case with uterine agenesis; 10 with unicornuate uterus, four of which also underwent MRI; six with didelphic uterus, one of which had MRI; 45 with bicornuate uterus, 12 of which had MRI; 125 with septate uterus (18 with two cervices), 42 of which had MRI (six with two cervices); 96 with arcuate uterus, three of which had MRI; and three with diethylstilbestrol (DES) iatrogenic uterine malformations, all of which had MRI. Among the 65 which underwent MRI, the diagnosis was: four cases with unicornuate uterus, 10 with bicornuate uterus (two with two cervices), 45 with septate uterus (five with two cervices), three with arcuate uterus and three with DES-related uterine malformations. The concordance between 3D ultrasound and MRI was very good (kappa index, (95% CI, )). Discrepancies in diagnosis between the two techniques occurred in four cases. There was very good concordance in the diagnosis of associated findings (kappa index, (95% CI, )), this analysis identifying differences in two cases. Conclusions There is a high degree of concordance between 3D ultrasound and MRI in the diagnosis of uterine malformations, the relationship between cavity and fundus being visualized equally well with both techniques. 3D ultrasound should be complemented by careful gynecological exploration in order to identify any alterations in the cervix. Copyright 2010 ISUOG. Published by John Wiley & Sons, Ltd. INTRODUCTION Uterine malformations make up a heterogeneous group of congenital anomalies that can result from the underdevelopment of the Müllerian ducts, disorders in their fusion and/or alterations in septum resorption. The prevalence of uterine malformations is difficult to establish. They are estimated to occur in 0.4% (0.1 3%) 1,2 of the general population and in 4% of infertile women (some authors do not distinguish between these two groups 3 ), and in patients with repeated spontaneous miscarriages the figures fluctuate between 3 and 38% 4 8. The discrepancy among different publications stems from their use of different diagnostic techniques, heterogeneous population samples and the clinical diversity of Müllerian anomalies. There are several classifications of uterine malformation, but the most widely accepted is that established in Correspondence to: Dr C. Bermejo, Gabinete Médico Velázquez, Velazquez 25, 1, Madrid, Spain ( carminabermejo@yahoo.es) Accepted: 28 August 2009 Copyright 2010 ISUOG. Published by John Wiley & Sons, Ltd. ORIGINAL PAPER

2 594 Bermejo et al by the American Fertility Society (AFS) 9,whichis not only based on embryological factors, but also takes into account clinical factors, prognosis and treatment (Figure 1). It classifies uterine malformations into seven groups, providing a very useful, but somewhat incomplete, categorization, as it requires the specification of any associated malformations, when found. There are several techniques available for the evaluation of uterine malformations. When the cavity only is to be assessed, hysterosalpingography (HSG) and hysteroscopy are especially useful. Laparotomy and laparoscopy can be also used for examination of the uterine fundus. There are two techniques, however, that combine the study of both these structures, which is indeed relevant for the diagnosis: magnetic resonance imaging (MRI) and three-dimensional (3D) ultrasound. While MRI is a useful option in the diagnosis of Müllerian anomalies, with numerous studies having proved its excellent efficacy in this field 10 13, 3D ultrasound represents a valid alternative, because, in addition to its lower cost and better tolerance by patients, it provides images of very similar quality to those yielded by MRI 14. There is a lack, however, of studies comparing these two techniques for the diagnosis and categorization of uterine malformations. The objective of this study, therefore, was to demonstrate the value of 3D ultrasound in the diagnosis of uterine malformations and its concordance with MRI. METHODS Included in this study were 286 women diagnosed with uterine malformation by 3D ultrasound between November 2004 and May 2009 in one of two centers (Gabinete Médico Velázquez, Madrid, Spain and Delta Ecografía Madrid, Spain). They had been referred to our clinics on suspicion of uterine malformation following clinical and/or conventional two-dimensional (2D) ultrasound examination. We performed 3D ultrasound in all cases, bimanual examination and speculoscopy in 283 cases and MRI in 65. 3D ultrasound examination of uterine cavity and cervical canal Examinations were performed using a Voluson Pro and three Voluson 730 Expert (GE Medical Systems, Zipf, Austria) ultrasound machines, equipped with convex transabdominal probe RAB4-8L 4 8 MHz 4D and endocavitary probe RIC5-9H 5 9 MHz 4D. In all cases (except for three patients with intact hymen), we obtained transvaginally between one and three static volumes of the uterus, with a quality ranging from medium to maximum. Initially we visualized the uterus on 2D ultrasound in a strict mid-sagittal view, adjusting the capture window to obtain the optimal 3D volume. The volume was then obtained using a sweep angle of 90 from one side of the uterus to the other, bisecting the capture plane (Figure 2). In 25 cases with anomalies resulting in a large transverse uterine diameter (didelphic uterus (n = 5), wide septate uterus (n = 10), bicornuate uterus with ample separation between horns (n = 8) and communicating unicornuate uterus (n = 2)) the volume was obtained from a transverse plane so that both uterine horns could be visualized, in order to allow better estimation of the cavity/fundus relationship in the 3D reconstruction (Figure 3). In seven cases (with optimal ultrasound conditions including low I Hypoplasia/agenesis II Unicornuate III Didelphus (a) Vaginal (b) Cervical (a) Communicating (b) Non-communicating IV Bicornuate (c) Fundal (d) Tubal (e) Combined (c) No cavity (d) No horn (a) Complete (b) Partial V Septate VI Arcuate VII DES drug related (a) Complete (b) Partial Figure 1 Classification of uterine malformations according to the American Fertility Society 9. DES, diethylstilbestrol.

3 3D ultrasound and MRI in uterine anomalies 595 Figure 2 Procedure to obtain a rendered image of a Müllerian anomaly, in this case an arcuate uterus. (a) The initial plane for obtaining three-dimensional volumes was generally mid-sagittal (Window A). Acquisition was carried out with a sweep angle of 90. The image corresponds to the bisector of the angle made by the volumetric probe as it is moved from one side of the uterus to the other. Window B is the axial plane and Window C the coronal plane. (b) The rendering box was then adjusted (Window A) and the green line adapted to the curved plane of the uterine cavity so it became positioned on the endometrium. The midline was verified (Window B). In the bottom right is the rendered image of the surface of the arcuate uterus. Figure 3 In uterine malformations with a large transverse diameter, such as this bicornuate uterus, three-dimensional (3D) volumes were generally obtained from an axial plane (Window A) so that both horns would fit in the same field of view, enabling the cavity/fundus relationship to be better established in the 3D reconstruction (bottom right). body mass index) and in the three patients with intact hymen, we captured the volume transabdominally, with initial sagittal, coronal or transverse planes, and in 25 cases we obtained two volumes, one to study the fundus and cavity and another to study the cervix and cervical canal (Figure 4). The volumes were manipulated until a satisfactory surface rendered image was obtained of the fundus and uterine cavity as well as the cervical canal. When a mid-sagittal plane was used to capture the volume, we adjusted the rendering box in Window A (capture image) to include the uterine fundus and adjusted the green rendering line (from front to back), tracing the sagittal curve plane of the uterine cavity so that the line was located on the endometrium, and checking in Window B that we were in the midline (Figure 2). When studying the cervix we readjusted the rendering box and the green line. When the volume was obtained in a transverse plane, we included both uterine horns in the rendering box and adjusted the green line so that a good quality image showing both cavity and fundus was obtained in the rendered view. Luminosity and contrast curves were adjusted for both multiplanar and rendered images, as well as for threshold and transparency. The rendering modes used were a mixture of surface/gradient of light of 10/60 to 60/10. Physical examination With the exception of the three patients with intact hymen, patients underwent bimanual examination and

4 596 Bermejo et al. Figure 4 Three-dimensional surface rendered ultrasound images of the uterine cervix in three women with uterine malformations: (a) incomplete cervical septum in a case of septate uterus; (b) complete cervical septum in a case of septate uterus; (c) two diverging cervical canals in a case of septate uterus with two cervices. Table 1 Indications for magnetic resonance imaging (MRI) (n = 65) Indication for MRI (n) Uterine type on 3D ultrasound n Complex malformations Poor reproductive outcome Poor reproductive outcome and complex malformations Unicornuate uterus 4 (1 communicating) 2 (1 communicating) 0 2 Didelphic uterus Bicornuate uterus Septateuterus 42(6with2cervices) 4(3with 2 cervices) 34 4 (3 with 2 cervices) Arcuate uterus Iatrogenic uterine malformations Total D, three-dimensional; MRI, magnetic resonance imaging. speculoscopy before and/or after the ultrasound examination to confirm the ultrasound findings. We looked for associated findings, i.e. vaginal septum, cervical septum or duplicity and abnormal pelvic masses. Magnetic resonance imaging Sixty-five patients underwent MRI after 3D ultrasound and physical examinations. MRI was indicated for patients with previous poor obstetric outcome (two or more spontaneous miscarriages, premature birth or impossibility to accomplish pregnancy) and for those with complex anomalies (Table 1). The examination was performed by the same observer (E.L.) in all cases, using a Siemens Magneton Avanto 1.5 Tesla machine (Siemens Medical Solutions, Mountain View, CA, USA). All studies included coronal high-resolution T2-weighted turbo spinecho imaging with the following parameters: TR/effective TE, 3410/114; refocusing flip angle, 180 ; turbo factor, 25; rectangular field of view, mm; matrix, ; slice thickness, 4 mm; 195 Hz/pixel; 19 slices; 1 3 signal averages; average time of acquisition, 2 min 49 s. Findings were classified according to the AFS 9 (Figures 1 and 5), describing the associated malformations found. In order to distinguish bicornuate from septate uteri using 3D ultrasound we used the formula proposed by Troiano and McCarthy 15 : a line was traced, joining both horns (Figure 6). If this line crossed the fundus or was 5 mm from it, the uterus was considered bicornuate, while if it was > 5 mm from the fundus, the uterus was considered septate, regardless of whether the fundus was dome-shaped, smooth or discretely notched (Figure 6). When differentiating bicornuate from septate uteri using MRI, all cases with an incision > 1cm deep in the fundus were considered to be bicornuate uterus. Differentiation between arcuate and septate uteri was carried out in the coronal plane on both 3D ultrasound and MRI. While both types of uterus have a normal contour, in arcuate uterus the fundal indentation appears as an obtuse angle at the central point 16, with a depth of up to 1.5 cm 17, whereas septate uterus is characterized by a fundal indentation with an acute angle at the central point, with a depth of 1.5 cm or more (Figure 7).

5 3D ultrasound and MRI in uterine anomalies 597 Figure 5 Three-dimensional surface rendered ultrasound images showing different types of uterine malformation using the American Fertility Society 9 classification: (a) normal uterus; (b) unicornuate uterus (Type IId); (c) didelphic uterus (Type III); (d) complete bicornuate uterus (Type IVa); (e) partial bicornuate uterus (Type IVb); (f) septate uterus with two cervices (Type Va); (g) partial septate/subseptate uterus (Type Vb); (h) arcuate uterus (Type VI); and (i) uterus with diethylstilbestrol (DES) drug-related malformations (Type VII). Statistical analysis Two kappa indices were calculated in order to study the diagnostic concordance between 3D ultrasound and MRI for the diagnosis of uterine malformations, one taking into account the diagnosis of the type of anomaly according to AFS classification 9 alone, and the other according to AFS classification in combination with the associated malformations found 18,19. RESULTS There was a high degree of concordance between 3D ultrasound and MRI, with a kappa index of (95% CI, ), for the diagnosis of uterine malformations using the AFS classification, with a discrepancy between sonography and MRI in only four cases (Table 2). On analysis of associated findings there was discordance in two cases: one bicornuate uterus in which MRI revealed the presence of a cervical septum that was not confirmed on hysteroscopy for its resection, and a septate uterus that appeared to have two cervices on 3D ultrasound (in one of the three patients who did not undergo clinical examination) but which MRI showed to have a complete septum with low signal intensity. There was very good concordance (kappa index, (95% CI, ) between 3D ultrasound and MRI for the diagnosis of associated anomalies. Table 2 Diagnosis of uterine malformations on three-dimensional (3D) ultrasound and magnetic resonance imaging (MRI) Uterine type 3D ultrasound diagnosis (n) Underwent MRI MRI diagnosis Agenesis 1 0 Unicornuate uterus 10 (2 communicating) 4 (1 communicating) 4 (1 communicating) Didelphic uterus 6 1* 0 Bicornuate uterus 45 (1 with two cervices) (2 with 2 cervices) Septate uterus 125 (18 with 2 cervices) 42 (6 with 2 cervices) 45 (5 with 2 cervices) Arcuate uterus Iatrogenic uterine malformations Total *One of the three patients who were not examined clinically was diagnosed as having didelphic uterus on 3D ultrasound but was found to have bicornuate uterus with two cervices on MRI. Three uteri diagnosed as bicornuate on 3D ultrasound were found to be septate on MRI.

6 598 Bermejo et al. (a) (b) (c) Figure 6 To distinguish bicornuate uteri from septate uteri with three-dimensional ultrasound we used the formula proposed by Troiano and McCarthy 15 : a line was traced joining both horns of the uterine cavity. If this line crossed the fundus or was 5 mm from it, the uterus was considered bicornuate (a and b); if it was > 5 mm from the fundus it was considered septate, regardless of whether the fundus was dome-shaped (c), smooth or discretely notched. DISCUSSION Figure 7 (a) Three-dimensional surface rendered ultrasound image (coronal view) showing the normal outer uterine contour of a uterus that was identified as arcuate (rather than partial septate) because the fundal indentation appeared as an obtuse angle at the central point 16, < 1.5 cm deep 17. (b) A partial septate uterus characterized by a normal outer uterine contour, which could be differentiated from arcuate uterus because the fundal indentation was an acute angle at the central point, > 1.5 cm deep. In the majority of uterine malformations, especially the less extreme forms (arcuate, septate and bicornuate uteri) two cavities and little further detail are seen on conventional 2D ultrasound. On 3D ultrasound, however, the relationship of the cavity with the fundus is evident, enabling precise diagnosis due to the contribution of the C-plane (coronal), which is impossible to obtain in the majority of cases on 2D ultrasound yet is crucial to the diagnosis of these anomalies. Distinction between these less extreme cases is vital due to their differing prognoses and management. Additionally, 3D ultrasound enables us to make measurements such as the length and thickness of the septum, calculate the volume of the cavity (achieving high intra- and interobserver reproducibility ) and study the vascularization, which can affect fertility prognosis, thus aiding the choice of treatment The success of 3D ultrasound in the diagnosis of uterine malformations is well recorded. In 1995 Jurkovic et al. 23 compared 2D and 3D ultrasound and HSG, finding both

7 3D ultrasound and MRI in uterine anomalies 599 Figure 8 Comparison of three-dimensional ultrasound and magnetic resonance imaging in cases of uterine malformation; the two imaging modalities are extremely similar. Images, according to the American Fertility Society 9 classification, show: (a) unicornuate uterus (Type IId); (b) bicornuate bicollis uterus (Type IVb); (c) septate uterus with two cervices (Type Va); (d) partial septate uterus (Type Vb); (e) uterus with diethylstilbestrol (DES) drug-related malformations (Type VII). ultrasound modalities to be more efficient in the diagnosis of arcuate uterus and to have a high predictive value for larger anomalies, especially in the differentiation of bicornuate and partial septate uteri. Raga et al. 24 found 3D ultrasound to have a 91.6% accuracy in the study of the fundus and 100% in that of the cavity, using laparoscopy and HSG, respectively, for reference. Wu et al. 25, comparing the technique with laparoscopy and/or hysteroscopy, found 3D ultrasound to have a 92% accuracy in the diagnosis of septate uterus and 100% for bicornuate uterus. Also comparing it with laparoscopy and hysteroscopy, Mohamed et al. 26 recorded a sensitivity of 97%, specifity of 96%, positive predictive value of 92% and negative predictive value of 99% in the diagnosis of Müllerian anomalies while Ghi et al. recorded both a sensitivity and a specificity of 100% in the diagnosis of uterine malformations and 96% concordance between ultrasound and endoscopy with respect to the type of anomaly diagnosed 27. In this study we have shown that images obtained with 3D ultrasound and MRI are practically equivalent (Figure 8). The relationship between fundus and uterine cavity can be established perfectly either with sonographic reconstructions in the coronal plane or with the coronal sequences obtained by MRI. Their diagnostic accuracy is similar, as demonstrated by the excellent concordance achieved when classifying anomalies according to the AFS. The few differences observed occurred only when the lower part of the uterus was studied. Ultrasonographic studies of the cervix are fundamental for this type of pathology. The probe should be moved outwards gently to study the cervical canal and myometrium, assessing whether there are one or two cervices, and whether there is a complete or incomplete septum in the cervical canal (Figure 4). The presence of two, generally thick canals Figure 9 Septate uterus with an incomplete cervical septum on three-dimensional surface-rendered ultrasound (a) and magnetic resonance imaging (b). which diverge in their inferior section would suggest a double cervix (Figure 4c) rather than a septum (Figure 4a and b). However, although in the majority of cases the ultrasonographic impression will be correct, only confirmation by speculoscopy allows us to be absolutely sure. MRI can discriminate a septum (although the majority of septa contain myometrium in their upper regions) from cervical myometrium by the different signal intensities obtained from each (Figures 9 and 10). In terms of intensity, myometrium is the reference, low signal intensity indicating fibrous tissue and isosignal intensity indicating muscle. MRI can also differentiate vaginal septa as these have a less intense signal than does the vaginal wall. This differentiation cannot, however, be made on 3D or 2D ultrasound, necessitating careful bimanual examination. If bimanual examination is always performed in conjunction with 3D ultrasound, the two imaging techniques are comparable in terms of

8 600 Bermejo et al. Figure 10 Septate uterus with two cervices on three-dimensional surface-rendered ultrasound (a and d) and magnetic resonance imaging (b, c, e and f). Axial plane showing the two cervices can be observed in (e) and sagittal planes of the two hemiuteri, each with a cervix, can be observed in (c) and (f). accuracy/efficacy, giving 3D ultrasound the advantage over MRI as it is cheaper and better tolerated by patients. Furthermore, the gynecological examination is simple and is part of our usual work-up. 3D ultrasound was of most use when distinguishing between arcuate, septate and bicornuate uteri. Alcázar 28 confirmed this, also noting its lower accuracy in diagnosing didelphic uteri. However, the differential diagnosis between these three it is not always easy for 3D ultrasound or MRI. There are intermediate and incomplete forms of bicornuate and septate uteri, due to simultaneous lack of fusion and reabsorption of Müllerian ducts. For example, septate uteri with very wide septa have a large separation between the horns, and the structure of the septum on ultrasound is similar to that of myometrium. In these cases, the morphology of the cavity and the type of signal obtained from the septum on MRI, indicating the presence of myometrium, which is theoretically only present in the wall of bicornuate uteri, would lead to an incorrect diagnosis of bicornuate uterus. Applying Troiano and McCarthy s 15 formula to distinguish between these two types would allow the correct diagnosis of septate uterus to be made (Figure 11). We found it difficult to decide whether some intermediate forms should be assigned to Group V or Group VI, as some deeply arcuate uteri could be partially septate with a short, thick septum. Markedly arcuate uteri have a worse reproductive prognosis than do those with a minor cavitary incision, as Troiano and McCarthy 15 and Salim et al. 29 found. Our study was limited by the fact that the radiologist who carried out the MRI examination was not blinded to the 3D ultrasound diagnosis, and only those cases found to be positive for uterine malformations by 3D ultrasound were later analyzed by MRI. Thus, we remain ignorant as to whether any of the negative cases on 3D ultrasound would have proved positive on MRI. We believe that 3D ultrasound is a useful complement to 2D ultrasound on many occasions in gynecology, but that it is in the diagnosis of uterine malformations that it is fundamental. We propose that, on suspicion of Müllerian anomalies, 3D ultrasound be carried out, accompanied by complete gynecological examination. In doubtful or complex cases, MRI should be performed, particularly for the assessment of the cervix and vagina. Surgery should be reserved for malformations that could benefit from this treatment.

9 3D ultrasound and MRI in uterine anomalies 601 Figure 11 Septate uteri with a very wide septum have a large separation between the horns and the three-dimensional sonographic structure of the septum is similar to that of the myometrium. The morphology of the cavity and the type of signal obtained from the septum on magnetic resonance imaging indicates the presence of myometrium, leading to an incorrect diagnosis of bicornuate uterus, but applying Troiano and McCarthy s 15 formula leads to the correct diagnosis of septate uterus. ACKNOWLEDGMENTS We thank Maria Luisa Bermejo for her illustrations and Manuel Recio for his participation and his iconographic collaboration. REFERENCES 1. Byrne J, Nussbaum-Blask A, Taylor WS, Rubin A, Hill M, O Donnell R, Shulman S. Prevalence of Mullerian duct anomalies detected at ultrasound. Am J Med Genet 2000; 94: Ashton D, Amin HK, Richart RM, Neuwirth RS. The incidence of asymptomatic uterine anomalies in women undergoing transcervical tubal sterilization. Obstet Gynecol 1988; 72: Acien P. Incidence of Müllerian defects in fertile and infertile women. Hum Reprod 1997; 12: Stampe Sorensen S. Estimated prevalence of Mullerian duct anomalies. Acta Obstet Gynecol Scand 1988; 67: Stray-Pedersen B, Stray-Pedersen S. Etiologic factors and subsequent reproductive performance in 195 couples with a prior history of habitual abortion. Am J Obstet Gynecol 1984; 148: Raga F, Bauset C, Remohí J, Bonilla-Musoles F, Simón C, Pellicer A. Reproductive impact of congenital Mullerian anomalies. Hum Reprod 1997; 12: Makino T, Hara T, Oka C, Toyoshima K, Sugi T, Iwasaki K, Umeuchi M, Iizuka R. Survey of 1120 Japanese women with a history of recurrent spontaneous abortions. Eur J Obstet Gynecol Reprod Biol 1992; 44: Clifford K, Rai R, Watson H, Reagan L. An informative protocol for the investigation of recurrent miscarriage: preliminary experience of 500 consecutive cases. Hum Reprod 1994; 9: The American Fertility Society classifications of adnexal adhesions, distal tubal obstruction, tubal occlusion secondary to tubal ligation, tubal pregnancies, Mullerian anomalies and intrauterine adhesions. Fertil Steril 1988; 49: Fedele L, Dorta M, Brioschi D, Massari C, Candiani GB. Magnetic resonance evaluation of double uteri. Obstet Gynecol 1989; 74: Carrington BM, Hricak H, Nuruddin RN, Secaf E, Laros RK Jr, Hill EC. Müllerian duct anomalies: MR imaging evaluation. Radiology 1990; 176: Pellerito JS, McCarthy SM, Doyle MB, Glickman MG, DeCherney AH. Diagnosis of uterine anomalies: relative accuracy of MR imaging, endovaginal sonography and hysterosalpingography. Radiology 1992; 183: Fischetti SG, Politi G, Lomeo E, Garozzo G. Magnetic resonance in the evaluation of Mullerian duct anomalies. Radiol Med 1995; 89: Deutch TD, Abuhamad AZ. The role of 3-dimensional ultrasonography and magnetic resonance imaging in the diagnosis of Mullerian duct anomalies: a review of the literature. JUltrasound Med 2008; 27: Troiano R, McCarthy S. Müllerian duct anomalies: imaging and clinical issues. Radiology 2004; 233: Woelfer B, Salim R, Banerjee S, Elson J, Regan L, Jurkovic D. Reproductive outcomes in women with congenital uterine anomalies detected by three-dimensional ultrasound screening. Obstet Gynecol 2001; 98: Syed I, Hussain H, Weadock W, Ellis J. Uterus, Mullerian duct abnormalities. emedicine. article/ overview [Accessed 20 February 2009]. 18. Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics 1977; 33: Kundel HL, Polansky M. Measurement of observer agreement. Radiology 2003; 228: Raine-Fenning N, Fleischer AC. Clarifying the role of threedimensional transvaginal sonography in reproductive medicine: an evidence-based appraisal. J Exp Clin Assist Reprod 2005; 2: Salim R, Woelfer B, Backos M, Regan L, Jurkovic D. Reproducibility of three-dimensional ultrasound diagnosis of congenital uterine anomalies. Ultrasound Obstet Gynecol 2003; 21: Timor-Tritsch IE, Monteagudo A, Tsymbal T, Strok I. Threedimensional inversion rendering: a new sonographic technique and its use in gynecology. J Ultrasound Med 2005; 24: Jurkovic D, Geipel A, Gruboeck K, Jauniaux E, Natucci M, Campbell S. Three-dimensional ultrasound for the assessment of uterine anatomy and detection of congenital anomalies: a comparison with hysterosalpingography and two-dimensional sonography. Ultrasound Obstet Gynecol 1995; 5: Raga F, Bonilla-Musoles F, Blanes J, Osborne NG. Congenital Mullerian anomalies: diagnostic accuracy of three-dimensional ultrasound. Fertil Steril 1996; 65: Wu MH, Hsu CC, Huang KE. Detection of congenital müllerian duct anomalies using three-dimensional ultrasound. J Clin Ultrasound 1997; 25: Mohamed M, Momtaz MD, Alaa N, Ebrashy MD, Ayman A, Marzouk MD. Three-dimensional ultrasonography in the evaluation of the uterine cavity. MEFS Journal 2007; 12: Ghi T, Casadio P, Kuleva M, Perrone AM, Savelli L, Gianchi S, Pelusi C, Pelusi G. Accuracy of three-dimensional ultrasound in diagnosis and classification of congenital uterine anomalies. Fertil Steril 2009; 92: Alcázar JL. Three-dimensional ultrasound in gynecology: current status and future perspectives. Curr Womens Health Rev 2005; 1: Salim R, Regan B, Woelfer B, Backos M, Jurkovic D. A comparative study of the morphology of congenital uterine anomalies in women with and without a history of recurrent first trimester miscarriage. Hum Reprod 2003; 18: