Quantitative and morphological assessment of early gestational sacs using three-dimensional ultrasonography

Authors

  • W. Lee,

    Corresponding author
    1. Division of Fetal Imaging, Department of Obstetrics and Gynecology, William Beaumont Hospital, Royal Oak, MI, USA
    2. Perinatology Research Branch, NICHD/NIH/DHHS, Bethesda, MD and Detroit, MI, USA
    3. Department of Obstetrics and Gynecology, Wayne State University/Hutzel Women's Hospital, Detroit, MI, USA
    • Division of Fetal Imaging, Department of Obstetrics and Gynecology, William Beaumont Hospital, 3601 West Thirteen Mile Road, Royal Oak, MI 48073-6769, USA
    Search for more papers by this author
  • R. L. Deter,

    1. Department of Obstetrics and Gynecology, Baylor College of Medicine, Houston, TX, USA
    Search for more papers by this author
  • B. McNie,

    1. Division of Fetal Imaging, Department of Obstetrics and Gynecology, William Beaumont Hospital, Royal Oak, MI, USA
    Search for more papers by this author
  • M. Powell,

    1. Division of Fetal Imaging, Department of Obstetrics and Gynecology, William Beaumont Hospital, Royal Oak, MI, USA
    Search for more papers by this author
  • M. Balasubramaniam,

    1. Division of Biostatistics, William Beaumont Hospital Research Institute, Royal Oak, MI, USA
    Search for more papers by this author
  • L. F. Gonçalves,

    1. Perinatology Research Branch, NICHD/NIH/DHHS, Bethesda, MD and Detroit, MI, USA
    2. Department of Obstetrics and Gynecology, Wayne State University/Hutzel Women's Hospital, Detroit, MI, USA
    Search for more papers by this author
  • J. Espinoza,

    1. Perinatology Research Branch, NICHD/NIH/DHHS, Bethesda, MD and Detroit, MI, USA
    2. Department of Obstetrics and Gynecology, Wayne State University/Hutzel Women's Hospital, Detroit, MI, USA
    Search for more papers by this author
  • R. Romero

    1. Perinatology Research Branch, NICHD/NIH/DHHS, Bethesda, MD and Detroit, MI, USA
    2. Department of Obstetrics and Gynecology, Wayne State University/Hutzel Women's Hospital, Detroit, MI, USA
    Search for more papers by this author

Abstract

Objective

Our main objective was to determine the value of three-dimensional ultrasonography (3DUS) and Virtual Organ Computer-aided AnaLysis (VOCAL) in the evaluation of gestational sac volume and morphology during early pregnancy.

Methods

Twenty-eight normal early pregnancies were scanned approximately every 2 weeks using transabdominal (TAS) and transvaginal (TVS) sonography. The VOCAL technique was used to create computerized surface models to classify gestational sac shapes as discoid or ellipsoid. Serial sac volume changes were analyzed using repeated measures ANOVA. Bland–Altman plots determined examiner bias and limits of agreement (LOA) for sac volume measurements. Gestational sac volumes were compared between the two-dimensional (2D) ellipsoid and VOCAL techniques. Differences between volume measurements were tested using the two-tailed paired t-test with statistical significance at the P < 0.05 level.

Results

Each subject was examined at a mean ± SD menstrual age of 7.9 ± 0.6 weeks (Scan 1), 9.9 ± 0.6 weeks (Scan 2), and 11.9 ± 0.6 weeks (Scan 3). Sac volumes significantly increased over time from 22 ± 11 mL at Scan 1, to 57 ± 21 mL at Scan 2 and 116 ± 35 mL at Scan 3 (P < 0.001). Predominant sac shapes were classified as ellipsoid (76.2%) or discoid (23.8%). Additional descriptors included: concave (60.7%), irregular (53.6%), or smooth (7.1%), with 19% of the overall group having more than one additional shape attribute. Clinically acceptable volume measurement bias and agreement were found for the following comparisons: (1) TAS versus TVS; (2) interobserver volume measurements; and (3) intraobserver volume measurements. The VOCAL technique yielded slightly greater sac volumes (64 ± 45.4 mL) when compared to the 2D ellipsoid model (48.6 ± 36.8 mL) (28.9 ± 24.3% (95% limit of agreement range, − 18.7 to 76.5%), P < 0.001).

Conclusions

Reproducible sac volume measurements can be obtained using VOCAL with either TAS or TVS. Early gestational sacs variably appear as discoid or ellipsoid structures with a concave indentation from the placenta. Sac volumes can be underestimated if an ellipsoid shape is assumed. Morphological and quantitative analysis of the gestational sac may provide baseline parameters for studying patients at risk for early pregnancy failure. Copyright © 2006 ISUOG. Published by John Wiley & Sons, Ltd.

Introduction

Over three decades have elapsed since gestational sac diameter was first introduced for the sonographic measurement of amniotic sac size1. This parameter was later correlated to pregnancy age and fetal growth1–4. Mean sac diameter was later found to correlate strongly with serum human chorionic gonadotropin (hCG) levels and this association raised the possibility that their combination could be used to evaluate threatened miscarriage. Other investigators subsequently applied three-dimensional ultrasonography (3DUS) to measure embryonic and gestational sac volumes5–9. The practical application of sac volume measurements, however, was limited by the extended amount of time required to manually trace surface contours.

In 2002 a novel semi-automated rotational technique (Virtual Organ Computer-aided AnaLysis, or VOCAL) was introduced for endometrial volume measurements10, 11. Our research group has examined its utility for evaluating fetal lung shape and volume12. Falcon et al.13 have more recently performed a cross-sectional study in which VOCAL was used to assess gestational sac volumes in early pregnancy at the time of chorionic villus sampling. Minimal attention was given to the morphologic appearance of gestational sacs during early pregnancy.

This investigation uses VOCAL to classify morphological characteristics of normal early pregnancies, quantifies longitudinal changes in gestational sac volume, and examines the reproducibility of this volume measurement technique.

Materials and Methods

The study population consisted of 28 pregnant women who were seen for clinical indications. Subjects were invited to participate under informed consent from the institutional review boards at William Beaumont Hospital and the National Institutes of Child Health and Human Development. Early pregnancies were scanned three times, every 2 weeks, beginning at 8–9 weeks' menstrual age. All subjects were subsequently confirmed to have normal pregnancy outcome. During the three visits, each subject underwent 3D ultrasound volume acquisitions using both transbdominal (TAS) and transvaginal (TVS) techniques (Voluson 730 Expert, GE Healthcare, Milwaukee, WI, USA). Data were stored on CD-ROM media. Care was taken to magnify the gestational sac until it filled at least two-thirds of the display screen as part of the standard volume acquisition procedure.

Multiplanar views of each gestational sac were obtained (4D View, Version 4.0, GE Healthcare). During each examination maximal sac diameters were measured for length (l), width (w), and height (h) to estimate volume using the ellipsoid formula (Figure 1):

equation image

Sac volume was also estimated by manually tracing the surface geometry with the VOCAL procedure. The examiner placed electronic markers on the anterior and posterior sac walls. The volume of interest was based on a semi-automated algorithm that defined geometric surfaces using six rotational steps of 30° each. Sac surfaces were manually outlined using a graphics pen and tablet (Intuos, WACOM Technology, Vancouver, WA, USA). Rotational sac contours were manually traced at 30° intervals until completion of a 180° sweep. In this manner, VOCAL provided a method for creating the computerized surface models that were used to classify predominant gestational sac shapes as discoid or ellipsoid. If present, other predominant shape attributes (irregular, smooth, concave) were also noted.

Figure 1.

Gestational sac volume measurement using multiplanar ultrasonography. Dotted lines between calipers show the maximal gestational sac diameters.

Each volume dataset was further analyzed by two different examiners using VOCAL in a blinded manner. Three-dimensional multiplanar sac volume measurements and the first set of VOCAL calculations were both performed separately by Examiner 1 (W.L.) using offline software. Blinded volume measurements of the same volume data sets were made a second time by Examiner 1 to determine intraobserver bias and agreement. Examiner 2 (B.M.) also made blinded sac volume measurements using VOCAL to examine interobserver bias and agreement by comparing her results to Examiner 1's first round of sac volume measurements. Finally, Examiner 1 used VOCAL to measure sac volumes for comparison between TAS and TVS techniques. Bland–Altman plots (% difference versus average) were used to determine the 95% limits of agreement and bias for a single examiner and between examiners (GraphPad Prism, version 4.0b for Macintosh, GraphPad Software, San Diego, CA, USA)14. Repeated measures analysis of the variance-covariance data structure was also used to evaluate how correlated measurements and measurement differences changed over time (PROC MIXED, SAS System for Windows, version 9.1.3, Service Pak 2). Differences between volume measurements were compared using the two-tailed paired t-test. Mean values were reported with their standard deviations, and P < 0.05 was considered statistically significant.

Results

Subjects were predominantly Caucasian (n = 25), but also included African–American (n = 1), Asian (n = 1), and Hispanic (n = 1) ethnicities; 10 women were primigravidae. The mean birth weight of 14 female and 14 male infants was 3572 ± 523 g at a mean menstrual age of 39.3 ± 1.2 weeks.

Three ultrasound examinations were performed, at menstrual ages 7.9 ± 0.6 weeks (Scan 1), 9.9 ± 0.6 weeks (Scan 2) and 11.9 ± 0.6 weeks (Scan 3). Mean sac volumes significantly increased over time from 22 ± 11 mL at Scan 1 to 57 ± 21 mL at Scan 2 and 116 ± 35 mL at Scan 3 (P < 0.001; Figure 2).

Figure 2.

Serial gestational sac volumes estimated using Virtual Organ Computer-aided AnaLysis and three-dimensional ultrasonography in 28 pregnancies with normal outcomes. Mean sac volumes increased from 22 ± 11 mL to 116 ± 35 mL from approximately 8 to 12 weeks' menstrual age. P < 0.001, repeated measures ANOVA.

Predominant gestational sac shapes were classified as ellipsoid (76.2%) or discoid (23.8%). Additional descriptors included sacs that were concave (60.7%), irregular (53.6%), or smooth (7.1%), with 19% of the overall group having more than one additional attribute (Figure 3). Eight of 28 study subjects (28.6%) demonstrated variable shape changes with advancing pregnancy (Figure 4, Tables 1 and 2).

Figure 3.

Variable surface-rendered gestational sac shapes using Virtual Organ Computer-aided AnaLysis. (a) Ellipsoid; (b) disk–concave; (c) ellipsoid–concave; (d) ellipsoid–irregular.

Figure 4.

Serial gestational sac shape changes using Virtual Organ Computer-aided AnaLysis at three different time points in one pregnancy (Case 8, Tables 1 and 2): (a) 9.4 weeks; (b) 10.1 weeks; (c) 12.0 weeks.

Table 1. Qualitative comparison of main sac characteristics among eight early pregnancies with morphological changes over time
CaseScan 1 (8.7 ± 0.7 weeks)Scan 2 (10.6 ± 0.7 weeks)Scan 3 (12.9 ± 0.6 weeks)
DiskEllipsoidDiskEllipsoidDiskEllipsoid
1  
2  
3  
4  
5  
6 
7 
8 
Table 2. Qualitative comparison of secondary shape attributes of the sac among eight early pregnancies with morphological changes over time
 Scan 1 (8.7 ± 0.7 weeks)Scan 2 (10.6 ± 0.7 weeks)Scan 3 (12.9 ± 0.6 weeks)
CaseIrregularConcaveIrregularConcaveIrregularConcave
1   
2  
3  
4  
5  
6  
7  
8 

Transabdominal sac volumes (62.1 ± 44.8 mL) were minimally smaller than those obtained using TVS (62.6 ± 43.8 mL) (−3.7 ± 7.2% (95% limits of agreement (LOA) − 17.8 to + 10.5%), P < 0.001) (Figure 5). Intraobserver sac volumes were not significantly different (1.9 ± 6.5% (95% LOA − 10.8 to 14.6%), P = 0.30) (Figure 6). Interobserver sac volumes (64.0 ± 45.4 mL vs. 66.5 ± 47.2 mL) were slightly different (–3.3 ± 7.3% (95% LOA − 17.6 to 10.9%), P = 0.003) (Figure 7). VOCAL yielded slightly greater sac volumes when compared to the ellipsoid model (64 ± 45.4 vs. 48.6 ± 36.8 mL) (28.9 ± 24.3% (95% LOA − 18.7 to 76.5%), P < 0.001) (Figure 8).

Figure 5.

Gestational sac volumes using Virtual Organ Computer-aided AnaLysis for a single examiner. A blinded comparison of measurement bias (dashed line) and 95% limits of agreement (solid lines) is demonstrated between abdominal and transvaginal scans. Sac volumes from transabdominal scans were slightly smaller than were those obtained using transvaginal sonography (P < 0.001). Percent difference was calculated as [(transabdominal sac volume) − (transvaginal sac volume) ÷ average of sac volumes] × 100.

Figure 6.

Gestational sac volumes using Virtual Organ Computer-aided AnaLysis for a single examiner: intraobserver variability. Measurement bias (dashed line) and 95% limits of agreement (solid lines) are demonstrated between two different sessions. Percent difference was calculated as [(sac volume from Trial 1) − (sac volume from Trial 2) ÷ average of sac volumes] × 100.

Figure 7.

Gestational sac volumes using Virtual Organ Computer-aided AnaLysis for two different examiners: interobserver variability. Measurement bias (dashed line) and 95% limits of agreement (solid lines) are demonstrated between separate examiners. Percent difference was calculated as [(sac volume from Observer 1) − (sac volume from Observer 2) ÷ average of sac volumes] × 100.

Figure 8.

Volume measurement compared using two-dimensional (2D) sonography and three-dimensional ultrasonography (3DUS) for a single examiner. Measurement bias (dashed line) and 95% limits of agreement (solid lines) are demonstrated between both techniques. Percent difference was calculated as [(sac volume using 3DUS) − (sac volume using 2D sonography) ÷ average of sac volumes] × 100. The Virtual Organ Computer-aided AnaLysis technique calculated slightly greater sac volumes as compared with results from a prediction model based on mean sac diameters and an assumed ellipsoid shape.

Sac volume measurements increased with advancing menstrual age (Type 3 test of fixed effects, P = 0.02) for a given examiner. By comparison, sac volume measurement variability was not significantly different between examiners over time (Type 3 test of fixed effects, P = 0.31). Sac volume measurement variability, resulting from comparing the two-dimensional (2D) ellipsoid model and VOCAL, did not significantly change over time (Type 3 test of fixed effects, P = 0.49).

Discussion

Only 50–60% of all conceptions advance beyond menstrual age 20 weeks15. Although the precise mechanisms responsible for early pregnancy loss are usually unknown, they may be associated with implantation problems16–18, genetic abnormalities19, unfavorable hormonal alterations20, 21, and hostile immunologic factors22. In this context sonographic markers such as the mean sac diameter and heart rate have been proposed as prognostic indicators of early pregnancy loss23, 24. The current investigation evaluated the use of VOCAL for rapid and accurate quantification of gestational sac volume. Surface-rendered computer models were also used to examine the morphological appearances of these gestational sacs over time.

Estimations of gestational sac volume by 3DUS were originally carried out using a lengthy procedure that required multiple surface contour tracings. For example, Steiner et al.25 examined 31 normal and eight abnormal pregnancies between 5 and 11 weeks' gestation. Transabdominal 3DUS and manual tracing were used to estimate sac volume. These measurements were significantly correlated with gestational age, even more so than most of the measured hormonal parameters. Inter-observer reproducibility of the technique was acceptable, and normal sac volume had a high negative predictive value (97%) for their small study group. In another investigation, Muller et al.5 used transvaginal gestational sac volumetry to compare 2D sonography with 3DUS for 130 pregnancies between 5 and 12 weeks' menstrual age. Conventional sac measurements were based on the maximum transverse diameters in three planes, using an ellipsoid formula for volume. These results were compared to 3DUS, where the sac contours were manually traced in 5–19 parallel planes. Mean gestational sac volume increased from 1.5 ± 2 mL to 127 ± 27 mL over the study period. 3DUS led to a smaller degree of variability in sac volume measurements. Sac volumetry, however, did not appear to provide prognostic information for pregnancy outcome—a finding that was similar to that of Acharya and Morgan7. On the other hand, Babinszki et al.6 evaluated 94 early pregnancies using 3DUS. They found that gestational sac volumetry was a significant predictor of abnormal outcome, with an odds ratio of 4.3 (95% CI, 1.18–15.91).

The recent application of VOCAL during early pregnancy offers the practical advantage of more rapid volume analysis. 3DUS has already been reported to be superior to conventional sonography for volume measurements of small, irregular objects5, 26, 27. Falcon et al.13 used VOCAL in a cross-sectional study to measure gestational sac volume in 500 consecutive singleton pregnancies prior to chorionic villus sampling between 11 and 13.9 weeks' menstrual age. Gestational sac volume increased from 69 mL at 11 weeks to 144 mL at 13.9 weeks. Fetal chromosomes were abnormal in 83 cases; when based on menstrual age, sac volumes were smaller in fetuses with triploidy and trisomy 13, but normal for trisomy 21, trisomy 18, and Turner syndrome. When the sac volume was normalized to crown–rump length, this parameter was larger in trisomy 18, smaller in triploidy and trisomy 13, and normal in trisomy 21 and Turner syndrome. Acceptable inter- and intraobserver reproducibility for transabdominal sac volume measurements was documented. However, it was concluded that gestational sac volumetry was unlikely to provide useful prediction of major chromosomal defects. This conclusion was based on the significant overlap with results from karyotypically normal pregnancies and the current availability of effective maternal serum screening. Our longitudinal study yielded similar sac volume measurements from genetically normal pregnancies using the VOCAL technique and now extends the observation period to include earlier pregnancies (between 8 and 10 completed weeks' menstrual age). In addition to confirming satisfactory examiner agreement for transabdominal sac volume measurements, we documented satisfactory reproducibility for the transvaginal 3D scans as well.

Prior longitudinal studies of gestational sac changes during early pregnancy have involved sac diameter measurements using 2D sonography28–30. Although the irregular nature of early gestational sacs has been known for years4, their fundamental morphological characteristics have not been well characterized using 3DUS. In our investigation, three-quarters of the sacs demonstrated a predominantly ellipsoidal shape. They commonly appeared as irregular concave disks with variable shape changes over time. An important limitation to our study, however, was reflected in the difficulty of classifying complex sac shapes into discrete categories. Indeed, the more conventional use of linear measurements to estimate mean sac diameter or volume should be used with caution because a regular sac shape cannot be assumed.

To summarize, reproducible sac volume measurements can be obtained using VOCAL with either TAS or TVS. Early gestational sacs variably appear as discoid or ellipsoid structures with a concave indentation from the placenta. Sac volume measurements, however, can be underestimated if an ellipsoid shape is assumed. Future research should apply these observations to a broader range of normal and abnormal pregnancies. The relative significance of qualitative sac shape appearance versus sac volumetry is at present unclear. More sophisticated methods are required to optimally classify sac features that best predict pregnancy outcome. For example, spatial coordinates that describe surface contours of the gestational sac could be utilized for quantitative shape analysis31. Such an approach would offer a basis for the interpretation of differences in anatomic shapes using statistical methods32, 33. Combined morphologic and quantitative assessment of the gestational sac may provide important baseline parameters for identifying patients at risk for early pregnancy failure.

Acknowledgements

This research was supported [in part] by the Intramural Research Program of the National Institute of Child Health and Human Development, NIH, DHHS.

Ancillary