OBJECTIVE: To compare biometric measures of pelvic floor function obtained using three-dimensional (3D) ultrasound and magnetic resonance imaging (MRI) in a group of nulliparous asymptomatic young women.
METHODS: Twenty-seven asymptomatic nulliparous volunteers were assessed prospectively, using translabial 3D ultrasound and multiplanar 3D MRI. Levator hiatal dimensions were measured in the axial plane in both modalities. All participants were imaged supine, after voiding with data acquired at rest, on maximum Valsalva and maximum pelvic floor contraction. Interobserver variability was determined for both methods. Normally distributed continuous ultrasound data were compared with equivalent MRI parameters, and intraclass correlation coefficients were used to estimate correlation between the two methods. Bland-Altman analysis was also used to estimate agreement between methods.
RESULTS: Interobserver repeatability was fair to excellent for all parameters measured with both methods. Moderate-to-substantial agreement between methods was shown for all tested parameters (intraclass correlation coefficients 0.587–0.783). There was a systematic but nonsignificant difference between methods, in that measurements on Valsalva tended to be larger for MRI, and the poorest agreement (intraclass correlation coefficient 0.587) was found for hiatal area on Valsalva.
CONCLUSION: Agreement between the two methods was moderate to substantial for all parameters except for hiatal area on Valsalva. Magnetic resonance imaging yielded higher area measurements on Valsalva, which may indicate difficulties in identifying the plane of minimal dimensions due to poorer temporal resolution compared with ultrasound imaging.
LEVEL OF EVIDENCE: II
This study in 27 nulliparous volunteers found moderate-to-substantial agreement between three-dimensional ultrasound and magnetic resonance imaging for measures of levator hiatal biometry.
From the 1Department of Sport & Exercise Science, University of Auckland, Auckland, New Zealand; and the 2University of Sydney, Sydney, Australia.
Corresponding author: J. A. Kruger, Department of Sport & Exercise Science, University of Auckland, P.O. Box 92019 Auckland, New Zealand; e-mail: firstname.lastname@example.org.
Financial Disclosure The authors have no potential conflicts of interest to disclose.
Imaging of the pelvic floor muscles to further determine their role in support of the pelvic organs, continence, and childbirth has been the focus of research in recent years. Changes in the morphology of the pelvic floor after vaginal delivery, incontinence, and prolapse have been demonstrated using magnetic resonance imaging (MRI) and, more recently, three-dimensional (3D) ultrasonography.1–4 To date, MRI has been the imaging method of choice due to its superior spatial resolution capabilities and, as shown most recently,5 its ability to identify different muscle groups of the pelvic floor. However, recent technological advances in 3/4D ultrasonography have allowed access to the axial plane and direct imaging of the entire levator hiatus, previously the domain of MRI. Ultrasonography is clearly more clinically convenient, easily accessible, and can be safely used throughout pregnancy. The technique to acquire and interpret the ultrasound data can be readily learned by interested clinicians, as opposed to MRI, which requires a referral to a radiologist. The dynamic imaging capabilities of MRI are limited at present because spatial resolution is lost with faster acquisition times, and temporal resolution is limited by the physical characteristics of MRI systems. Other limitations include the fact that it cannot be used in early pregnancy, and there are cost and access issues. Rendered 3D volumes and 4D cine loop capabilities of ultrasonography enable the assessment of the functional anatomy with good spatial and superior temporal resolution, with multiple volume data sets obtained per second.6 Several studies have recently defined pelvic floor function using MRI and 3D ultrasonography in relation to the measurement of descent of the pelvic organs during a voluntary Valsalva maneuver, the diameter and area measures of the levator hiatus at rest, during maximum pelvic floor contraction and during a voluntary valsalva.7–10
Levator ani is the collective term used to describe the deep muscles of the pelvic floor. Levator ani is a laminar muscle that spans the space between the obturator internus muscle laterally, the pubis anteriorly, and the coccyx posteriorly. The urethra, vagina, and anal canal exit the abdomen through a funnel-shaped cleft in the muscle, termed the levator hiatus. The levator ani consists primarily of the striated muscles, m. pubococcygeus, m. puborectalis, and m. iliococcygeus, although pubococcygeus and puborectalis are at times linked together in the clinical literature and termed “pubovisceralis.”11 With MRI, these muscles may be separately identifiable in the posterior compartment, but because of their proximity, they are often indistinguishable anteriorly. On ultrasonography the distinction between the two muscles is less apparent, and therefore the term “pubovisceralis” is highly useful.6,11 Although hiatal area and diameter measurements are easily obtained on 3D ultrasound,9,12,13 area measurement of the levator hiatus using MRI is time consuming and not easily amenable to dynamic imaging at present.14
Historically, using a true axial image, an area termed the levator hiatus has been measured.7,15 This was bounded by the pubis anteriorly and the levator ani on its lateral and posterior margins, but because of the funnel-shaped configuration of the levator ani, this measurement in the true axial plane was unsatisfactory. With 3D ultrasonography, more consistent measurements can be obtained using a plane defined by the lower margin of the pubis and the inner aspect of the puborectalis muscle where it rounds the anorectal junction. The state-of-the-art software used with 3D ultrasonography allows for manipulation of all three orthogonal planes to any orientation to display a particular anatomical characteristic more clearly. The capability to display data in the C plane or the axial view is one on the main advantages of 3D ultrasound systems when used for pelvic floor ultrasonography because it enables structures to be accessed that previously were the domain of MRI. The axial plane representing the plane of minimal hiatal dimensions9 is the true levator hiatus, which is defined as the area bounded by the pubococcygeus and puborectalis muscle, laterally and dorsally, and the inferior pubic rami and the symphysis pubis and its ligaments, anteriorly. Figure 1 shows these planes using a high-resolution MR sagittal image. These parameters were specifically chosen because it has been previously shown that levator hiatal biometry (linear diameters and area measures of the hiatus) are highly repeatable and reliable12,13,16 and clinically have an association with pelvic floor dysfunction (De Leon J, Steensma AB, Shek C, Dietz HP. Ballooning: can we define pathological distensibility of the levator hiatus [abstract]? Ultrasound Obstet Gynecol 2007;30:447).17,18
This study aimed to validate pelvic floor 3D ultrasonography by comparing biometric measures of the levator hiatus obtained on MRI and on ultrasound, as defined above, in a group of nulliparous young women.
MATERIALS AND METHODS
In this prospective study, translabial 3D ultrasonography and multiplanar MRI was used to assess the pelvic floor anatomy and function in nulliparous female volunteers. To ensure a large enough sample size to reliably assess the correlation between the two modalities, we based the calculations on a correlation of 0.8, which is considered high, an α of 0.05, and a power of 0.80. Using the formula and tables of Kraemer and Thiemann (1987)19 to calculate sample size for reliability studies, this yielded a sample size of 22. We sampled 27 subjects to allow for any technical difficulties with the data acquisition, and/or any subjects not being able to return for the MRI scanning.
Recruitment was primarily from advertisement at local university campuses, where the participants were matched according to age and body mass index (BMI). The enrollment period was approximately 6–9 months, from June 2005 to March 2006, which allowed for acquisition of the ultrasound imaging and recalling the participants and organizing MRI scans. Before imaging, all participants were questioned regarding pelvic floor dysfunction, including symptoms of urinary incontinence, prolapse, or fecal incontinence, using a standardized questionnaire. Additionally the participants were asked if they were aware of pelvic floor muscle exercises and if they practiced them. All participants were imaged after voiding, in the supine position, for both modalities.
The participants underwent ultrasound imaging using GE Kretz Voluson 730/730 Expert systems with 7-4 and 8-4 MHz volume transducers (GE Kretztechnik GmbH, Zipf, Austria). A previously published U.S. imaging protocol was used whereby images were obtained in the midsagittal plane with the angle of acquisition set at 85°. Volume data sets were acquired at rest on maximal pelvic floor muscle contraction and maximal Valsalva after visual biofeedback teaching. The most effective of at least three maneuvers was used for evaluation at a later date with the software 4D View V 2.1 (GE Medical Kretztechnik, Zipf, Austria). The entire levator hiatus was measured at rest, on Valsalva, and on maximum pelvic floor contraction. Figure 2 shows the plane used to determine levator hiatal area. The plane of minimal hiatal dimensions is identified in the midsagittal plane as the minimal distance between the inferoposterior margin of the pubis to the posterior aspect of the anorectal angle where it is encircled by the puborectalis muscle (see left image in Fig. 2). Both of these structures are highly echogenic and act as landmarks for the ultrasound images. Once the plane was defined in the midsagittal plane during postprocessing, this cross-section of the volume was displayed in the axial plane. A more detailed description of the methodology is given elsewhere.9 These ultrasound parameters were duplicated as closely as possible when obtaining and analyzing magnetic resonance images.
The plane of minimal dimensions used for the MR sequence was drawn from the inferior aspect of the pubis to the inner aspect of the puborectalis muscle, where it rounded the anorectal angle as demonstrated on a high-resolution sagittal image, and was comparable to that used in the ultrasound protocol (Fig. 3). These structures are easily identifiable on the static MR images, although there is some loss of resolution with the dynamic imaging. The images were obtained using a Siemens MAGNETOM Avanto 1.5T scanner (Siemens AG, Munich, Germany) by a radiologist (S.W.H.) with expertise in pelvic floor imaging. Multiplanar images were acquired using a 3D T2 TSE Siemens SPACE sequence 1 mm×1 mm×1 mm over 8 minutes. The static sagittal image used was T2-weighted (slice thickness 3 mm; slice gap 2 mm TR/TE 3400/102), 20 images in approximately 4 minutes. Dynamic images were acquired in the midsagittal, axial, and coronal plane using a T2 TRUFISP sequence TR/TE 4.77/2.39, using 12 measurements per slice. Using the midsagittal slice at rest, on maximum pelvic floor contraction and on Valsalva, the anterior-posterior diameter of the hiatus was measured from the inferior aspect of the symphysis pubis to the inner aspect of the puborectalis muscle (Fig. 4). Using the appropriate axial slice from the static 3D multiplanar sequence, the left-right distance through the vagina between the inner aspects of pubococcygeus was measured plus the anterior-posterior diameter and area measurements made, as on ultrasonography (Fig. 5). The appropriate angled axial plane on Valsalva was defined using the dynamic sagittal Valsalva images. Anterior-posterior diameters and transverse diameters of the levator hiatus and the area measurement were made using the angled axial plane (Fig. 6).
Data acquisition and analysis of the ultrasound images was undertaken by two of the authors (J.A.K. and H.P.D.). These authors have previously published test retest data for a series of 46 ultrasound volumes showing intraclass correlation coefficients values of between 0.57 and 0.81, with the best agreement for levator hiatal area on Valsalva.20 Images were saved on DVD, which allowed for later analysis at remote sites. Single plane MR images were saved onto CD and analyzed using eFilm software (Merge Healthcare, Milwaukee, WI) on a personal computer. The multiplanar images were analyzed using postprocessing software Syngo Leonardo systems (Siemens, Erlangen, Germany), which allowed all three planes—axial, coronal, and sagittal—to be analyzed at the same coordinate. Interobserver variability was also determined for two authors (J.A.K. and S.W.H.) using intraclass correlation coefficients in the analysis of the MR images. Ethical approval was granted by the University of Auckland Human Participants Ethics committee (reference 2004/250).
Statistical analysis was performed with SPSS 15 software (Systat Software, Chicago, IL). Kolmogorov-Smirnov analysis was used to check for normal distribution of the data. For all normally distributed data, intraclass correlation coefficients were used to compare continuous ultrasound data with comparable MR data from the same subject, where the end points were clearly definable and in the same plane. In addition, intraclass correlation coefficients were used to measure reliability between the examiners for both the ultrasound and the MR data. Finally, Bland-Altman analysis was used to evaluate the difference between the ultrasound measurement and the MRI measurement against their mean, and the 95% confidence interval for each parameter was calculated.21 Bland-Altman plots were used in addition to intraclass correlation coefficients because intraclass correlation coefficients measure correlation and this can be high, even when one modality is systematically measuring larger values. Bland-Altman analysis and plots, on the other hand, can identify systematic bias, from the mean difference, with the limits of agreement calculated by the mean difference±2×standard deviations of the differences (±2SDd). This gives an interval in which 95% of the differences can be expected to lie.
None of the participants had symptoms of pelvic floor dysfunction, and all were nulliparous, hence all met the inclusion/exclusion criteria. The general demographic characteristics of the group are presented in Table 1. For the MRI data, the measurements of the two examiners, J.A.K. and S.W.H., showed substantial-to-excellent agreement, with all intraclass correlation coefficients being greater than 0.7. Thus, interobserver repeatability was fair to excellent for all parameters measured with both methods.
There are moderate-to-substantial correlations between the two modalities, particularly when measuring parameters at rest. The linear anterior-posterior MR measures in the midsagittal plane were easily identifiable and demonstrated good agreement between the two modalities at rest, where the mean difference was only 0.07 cm, with narrow limits of agreement (–0.97 to 1.11). The anterior-posterior diameter measured during dynamic imaging, in the midsagittal plane and the transverse diameters measured in the axial plane, also demonstrated good agreement, with all parameters showing a mean difference of less than 1 cm, using Bland-Altman analysis, with intraclass correlation coefficients ranging from 0.775 to 0.595 for all these parameters. Measurements of hiatal area at rest demonstrated substantial correlation (intraclass correlation coefficient 0.783), while measurements of hiatal area on Valsalva were less strongly correlated (intraclass correlation coefficient 0.587). Bland-Altman analysis showed the limits of agreement were most narrow for hiatal areas at rest (–0.97 to 1.11), with a mean difference of 0.07, and widest for area on Valsalva (–10.87 to 16.69), with a mean difference of 2.91, in the axial plane. Representative plots are shown in Figure 7. The midsagittal MR slice proved the most reliable for identifying the anterior-posterior hiatal diameter, both on static and dynamic images.
All intraclass correlations between parameters measured on the ultrasound and MR images, the limits of agreement, mean differences, and standard deviation of the difference (Bland-Altman analysis) are presented in Table 2.
This study compared hiatal biometry obtained by 3D pelvic floor ultrasonography and pelvic MRI in an attempt to validate the former method as a means of assessing pelvic floor anatomy and function. It appears that, although hiatal area on Valsalva was not significantly different between ultrasonography and MRI, ultrasonography measured consistently smaller hiatal areas on maximum Valsalva (ultrasonography 17.46 cm2, MRI 20.37 cm2). These results are verified by the Bland-Altman analysis, where the mean difference between measurements for the hiatal area on Valsalva was 2.91 cm with the largest limits of agreement, whereas the mean difference for the measurements of the hiatal area at rest was only 0.07 cm with smaller limits of agreement. An explanation may be that it is difficult on MRI to accurately predict the end point of the Valsalva maneuver, even when using the sagittal image as a reference. Magnetic resonance imaging is not “real time” scanning, and the Valsalva maneuver needed to be repeated after acquisition of the sagittal reference slice (as shown in Fig. 6). Therefore, the resulting axial image may not always be obtained in exactly the same plane. Moreover, due to the funnel shape of the levator hiatus and the fact that the hiatal plane is non-Euclidian or “warped,” the end point for measurement of hiatal area from the MR images may in fact be more cranial than ultrasound images. Assessment of dynamic changes in levator hiatal area describes functional properties of the pelvic floor musculature. Because 3D ultrasonography is real time imaging, the levator hiatus can be followed during maneuvers and provide qualitative and quantitative information on muscle function. Regardless of any remaining methodological differences, it appears that both ultrasonography and MRI are highly repeatable, which agrees with literature data on the subject.4,9,12,13,22
There is little comparative research between 3D translabial ultrasonography and MRI of the pelvic floor, although both static and dynamic MRI have been used extensively to evaluate the anatomy and function of the levator ani complex in both normal women and in women who suffer from pathology.1,7,8,23 Translabial 3D ultrasound has been used to define a range of normal pelvic floor anatomy and function as well as for demonstrating dysfunction and postdelivery soft tissue trauma.17,24 One previous small study used 3D ultrasonography and MRI in a comparative analysis of hiatal dimensions in women with prolapse.18 The authors found a moderate-to-poor correlation between the hiatal area measures at rest (intraclass correlation coefficient 0.485) and on Valsalva (intraclass correlation coefficient 0.13). Barry et al18 did not use the same plane for MRI and ultrasonography, whereas the present study did. This may have contributed to the poorer reliability observed in that study.
Additionally, it is probable that any difficulties in defining the plane of minimal dimensions during a voluntary Valsalva are more marked in women with pelvic organ prolapse. Consequently, one would expect MRI to perform less well in such women, resulting in poorer correlations between ultrasound and MR findings, which is exactly what was observed by Barry et al.18
Pelvic floor 3D ultrasonography is becoming more widespread, due to the increasing availability of 3D-capable ultrasound systems. The ability of ultrasonography to reflect the functional anatomy of the pelvic floor muscles has added another dimension to the information already gained by MRI. In the clinical setting, changes in the levator ani muscle postdelivery are easily demonstrable on ultrasonography,24 and it seems that such alterations are associated with an increased risk of significant pelvic organ prolapse.25,26 This confirmed earlier clinical work which found a relationship between pelvic organ prolapse and the size of the levator hiatus.27 Furthermore, a recent study has shown an inverse association between hiatal area on contraction and length of second-stage labor. This raises the possibility of using 3D ultrasonography as an easy, reliable diagnostic tool, not only for pelvic floor dysfunction, but also for risk assessment before delivery.28
Magnetic resonance imaging has superior discriminatory capabilities and may be able to define the different muscle components of the levator complex.5 In certain contexts MR imaging may be preferable because each muscle component is likely to have a unique mechanical effect. Defining exactly which aspect of the muscle is damaged may aid in correlating clinical with anatomical findings and, in the future, allow for more informed surgical correction.
Undoubtedly, ultrasonography is more clinically convenient and as such will generally be the method of choice in a clinical setting. Further work is needed to define the relative roles of ultrasonography and MR in the evaluation of symptomatic women and in those with abnormal levator morphology. For example, both imaging methods may perform differently in symptomatic or obese women. This current study used only young, asymptomatic, nulliparous women who, therefore, had relatively low BMIs (mean 22 kg/m2). Although this may affect the ability to generalize the findings to a symptomatic group, imaging quality is more likely affected by muscular atrophy than obesity in both modalities.
Levator hiatal dimensions can be determined by 3D pelvic floor ultrasonography and pelvic MRI, and this study validates the use of 3D ultrasonography as an interchangeable modality with MRI for this purpose. In a group of 27 young nulliparous asymptomatic women, intraclass correlations were substantial except for hiatal dimensions on Valsalva, for which correlations were only moderate. This is most likely explained by the fact that, due to physical limitations, MRI cannot follow the true plane of minimal dimensions to the degree to which this is possible on ultrasonography. This study also contributes normative data on pelvic floor function in nulliparous, asymptomatic women.
1. DeLancey JO, Kearney R, Chou Q, Speights S, Binno S. The appearance of levator ani muscle abnormalities in magnetic resonance images after vaginal delivery. Obstet Gynecol 2003;101:46–53.
2. Dietz H, Haylen B, Broome J. Ultrasound in the quantification of female pelvic organ prolapse. Ultrasound Obstet Gynecol 2001;18:511–4.
3. Dietz HP. Levator function before and after childbirth. Aust N Z J Obstet Gynaecol 2004;44:19–23.
4. Fielding JR. Practical MR imaging of female pelvic floor weakness. Radiographics 2002;22:295–304.
5. Margulies RU, Hsu Y, Kearney R, Stein T, Umek WH, DeLancey JO. Appearance of the levator ani muscle subdivisions in magnetic resonance images. Obstet Gynecol 2006;107:1064–9.
6. Dietz H. Pelvic floor ultrasound. Curr Med Imaging Rev 2006;2:271–290.
7. Singh K, Reid W, Berger L. Magnetic resonance imaging of normal levator ani anatomy and function. Obstet Gynecol 2002;99:433–8.
8. Aukee P, Usenius J, Kirkinen P. An evaluation of pelvic floor anatomy and function by MRI. Eur J Obstet Gynecol Reprod Biol 2004;112:84–8.
9. Dietz H, Shek C, Clarke B. Biometry of the pubovisceral muscle and levator hiatus by three-dimensional pelvic floor ultrasound. Ultrasound Obstet Gynecol 2005;25:580–5.
10. Hoyte L, Schierlitz L, Zou K, Flesh G, Fielding JR. Two- and 3-dimensional MRI comparison of levator ani structure, volume, and integrity in women with stress incontinence and prolapse. Am J Obstet Gynecol 2001;185:11–9.
11. DeLancey JO. The anatomy of the pelvic floor. Curr Opin Obstet Gynecol 1994;6:313–6.
12. Yang JM, Yang SH, Huang WC. Biometry of the pubovisceral muscle and levator hiatus in nulliparous Chinese women. Ultrasound Obstet Gynecol 2006;28:710–6.
13. Guaderrama NM, Liu J, Nager CW, Pretorius DH, Sheean G, Kassab G, et al. Evidence for the innervation of pelvic floor muscles by the pudendal nerve. Obstet Gynecol 2005;106:774–81.
14. Kruger J, Murphy B, Heap S. Alterations in levator ani in elite nulliparous athletes: a pilot study. Aust N Z J Obstet Gynaecol 2005;45:42–7.
15. Fielding J, Dumanli H, Schreyer A, Okuda S, Gering D, Zou K, et al. MR-based three-dimensional modelling of the normal pelvic floor in women: quantification of muscle mass. AJR Am J Roentgenol 2000;174:657–60.
16. Majida M, Hoff Braekken I, Bo K, Umek W, Dietz H, Ellstrom Engh M. 3D and 4D ultrasound of the pelvic floor: an interobserver reliability study. Int Urogynecol J 2006;17:S136–7.
17. Dietz H, Steensma A, Vancaillie T. Levator function in nulliparous women. Int Urogynecol J Pelvic Floor Dysfunct 2003;14:24–6.
18. Barry C, Dietz H, Bhuta S, Lim Y, Greenland H, Muller R. MRI versus 3-dimensional ultrasound: a comparative study of levator hiatal dimensions in women with pelvic organ prolapse. Aust N Z Continence J 2006;12:61–5.
19. Kraemer H, Thiemann S. How many subjects? Statistical power analysis in research. London (UK): Sage Publications; 1987.
20. Kruger J, Dietz H, Murphy B. Pelvic floor function in elite nulliparous athletes. Ultrasound Obstet Gynecol 2007;30:81–5.
21. Bland J, Altman D. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986;1:307–10.
22. Barbaric ZL, Marumoto AK, Raz SM. Magnetic resonance imaging of the perineum and pelvic floor. Top Magn Reson Imaging 2001;12:83–92.
23. Goh V, Halligan S, Kaplan G, Healy J, Bartram C. Dynamic MR Imaging of the pelvic floor in asymptomatic subjects. AJR Am J Roentgenol 2000;174:661–6.
24. Dietz HP, Lanzarone V. Levator trauma after vaginal delivery. Obstet Gynecol 2005;106:707–12.
25. Dietz HP. Quantification of major morphological abnormalities of the levator ani. Ultrasound Obstet Gynecol 2007;29:329–34.
26. Dietz HP, Steensma AB. The prevalence of major abnormalities of the levator ani in urogynaecological patients. BJOG 2006;113:225–30.
27. Delancey JO, Hurd WW. Size of the urogenital hiatus in the levator ani muscles in normal women and women with pelvic organ prolapse. Obstet Gynecol 1998;91:364–8.
28. Lanzarone V, Dietz HP. Three-dimensional ultrasound imaging of the levator hiatus in late pregnancy and associations with delivery outcomes. Aust N Z J Obstet Gynaecol 2007; 47:176–80.
© 2008 The American College of Obstetricians and Gynecologists
This article has been cited