SEARCH

SEARCH BY CITATION

Keywords:

  • 3D ultrasound;
  • gestational sac;
  • SonoAVC;
  • VOCAL

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. REFERENCES

Objectives

To assess the novel three-dimensional (3D) tool, Sonography-based Automated Volume Count (SonoAVC) in the calculation of gestational sac volume at 11 + 0 to 13 + 6 weeks of gestation, to correlate the measurements with those obtained using Virtual Organ Computed-aided AnaLysis (VOCAL) and to study the reproducibility of SonoAVC volume calculation of this irregularly shaped structure.

Methods

We acquired 3D volumes of the uterus in 65 pregnancies at 11 + 0 to 13 + 6 weeks of gestation. We performed volume calculation of the gestational sac, excluding the fetus and the placenta, using VOCAL with 15° 12-step rotation. We then repeated the calculation with three different SonoAVC settings and compared both techniques. In 30 cases we assessed the reproducibility of the SonoAVC volume calculations.

Results

In 95% of cases it was possible to calculate the gestational sac volume with SonoAVC. This volume increased with advancing gestation and the volumes were expressed as delta values to compare the measurements made with VOCAL and the three different SonoAVC settings. There was no difference between delta values of gestational sac volume calculated using VOCAL and SonoAVC with high and medium growth settings. Reproducibility analysis showed good results.

Conclusion

Gestational sac volume calculation is feasible with SonoAVC in most cases and does not differ from that performed using VOCAL. High and medium growth SonoAVC settings seem to be more accurate for gestational sac volume calculation, although larger studies are required for standardization of the technique. The reproducibility analysis showed similar results to those previously published. Copyright © 2009 ISUOG. Published by John Wiley & Sons, Ltd.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. REFERENCES

As technology advances more tools become available for prenatal diagnosis, providing objective methods for volume calculation. The novel Sonography-based Automated Volume Count (SonoAVC, GE Healthcare Ultrasound, Zipf, Austria) mode of three-dimensional (3D) ultrasound was developed to simplify follicular volume assessment following ovarian stimulation1. Raine-Fenning et al.2 recently reported that in 51 cases follicular volume was measured more accurately by SonoAVC than by Virtual Organ Computed-aided AnaLysis (VOCAL, GE Healthcare Ultrasound) or using conventional two-dimensional (2D) diameters. Rizzo et al.3 reported that the fetal stomach was successfully measured by SonoAVC in 52 cases at 19–34 weeks of gestation and showed a significant reduction in the time needed to perform the calculations compared with using VOCAL. Several prenatal and in-vitro studies have reported high accuracy of VOCAL measurements and have validated the technique for performing volume calculation of different fetal and maternal structures with both regular and irregular shapes4–10.

The aims of this study were (1) to assess the new tool, SonoAVC, in the calculation of gestational sac volume at 11 + 0 to 13 + 6 weeks of gestation, (2) to evaluate the correlation between the automated volume calculation using SonoAVC and the manual volume calculation using VOCAL and (3) to assess the reproducibility of SonoAVC volume calculation of this irregularly shaped structure.

Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. REFERENCES

In this prospective study we examined 65 consecutive women who attended our unit for routine first-trimester screening with singleton pregnancies at 11 + 0 to 13 + 6 weeks of gestation in which no obvious fetal defect or ultrasound marker for chromosomal abnormality was detected. All patients gave signed informed consent to participate. All pregnancies were dated by known last menstrual period or early dating scan. We acquired 3D volumes of the uterus with a Voluson E8 (GE Healthcare Ultrasound) ultrasound machine equipped with a RAB 4-8-D transabdominal probe, using a highly contrasted 2D image.

Every patient transmits ultrasound in a different way, making it extremely difficult to standardize the gain settings. We therefore established that an adequate setting would be when the gain was sufficiently low so as to visualize fluid-filled structures as black with well-defined borders, in order to distinguish them from surrounding echoic structures. The sweep angle was set at 85° to ensure that the gestational sac was included completely in the volume, and maximum quality was selected. Volumes were stored for subsequent offline analysis. This was performed using 4D View software (version 7.0, GE Healthcare Ultrasound) by two sonographers with extensive experience in 3D ultrasound. The gestational sac volumes were measured with both VOCAL and SonoAVC, including the celomic cavity when visible, and excluding the fetus and the placenta (Figures 1 and 2).

thumbnail image

Figure 1. Sonographic measurement of gestational sac volume with SonoAVC, excluding fetus and placenta; multiplanar display and rendered images are shown.

Download figure to PowerPoint

thumbnail image

Figure 2. Sonographic measurement of gestational sac volume with VOCAL, analyzed with a manual 15° 12-step rotation and excluding the fetus.

Download figure to PowerPoint

The VOCAL technique used a 15° 12-step rotation along the y-axis in manual mode as previously described6, 7. Each volume was then displayed for the automated SonoAVC calculation as follows: first, the SonoAVC mode of display was activated; second, the region of interest was adjusted to the minimum size that contained the entire gestational sac; third, the threshold was modified to improve visualization of the gestational sac separately from surrounding echoic and anechoic structures; fourth, medium growth and medium separation settings were selected. The growth setting has five different display options (from minimum to maximum) and the effect of this tool is to make the tracing for the volume calculation closer to or further from the border of the echoic and anechoic structures. The separation setting also has five display options and affects the discrimination between two or more structures which are next to each other; in this study we did not modify the separation setting because only one anechoic structure was being measured. Once the volume calculation had been obtained, manual corrections were made to the outline when necessary, using the editing tools to add, remove, cut and merge. The fourth step was then repeated with high growth and low growth settings while maintaining the medium separation setting, in order to evaluate the difference in gestational sac volume according to the setting employed.

In all cases comparisons between gestational sac volume calculated using VOCAL and SonoAVC were made by one operator. Intraobserver reproducibility was analyzed in this operator, who measured 30 cases twice. Interobserver reproducibility was analyzed in 30 cases, each calculated by this and a second operator, using medium, high and low growth settings combined with medium separation. The operators were blinded to each others' and their own results.

Statistical analysis

Linear regression was used to assess the association between gestational sac volume, measured with VOCAL as the gold standard, and gestational age. All measurements performed with SonoAVC were then expressed as difference from the expected mean for gestation (delta values) to correct for gestational age. The Kolmogorov–Smirnov test was used to confirm normal distribution. The t-test for independent samples was used to determine the significance of the difference between the mean delta gestational sac volumes calculated using VOCAL and those calculated using the medium, high and low growth SonoAVC settings individually, and to compare the mean difference amongst the three SonoAVC settings. The Bland–Altman test11 was used to assess the intra- and interobserver agreement and bias between paired measurements with SonoAVC volume calculation. The intraclass correlation coefficient (ICC) was calculated between VOCAL and each of the three SonoAVC settings in all cases. The data were analyzed using the statistical software package SPSS 12.0 (SPSS Inc., Chicago, IL, USA). P < 0.05 was considered statistically significant.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. REFERENCES

The median gestational age was 85 (range, 77–97) days. In all 65 cases a 3D volume was obtained, and all the images were found to be satisfactory for the calculations with VOCAL. Overall there were 10 (15%) cases in which the measurements were excluded from the analysis because at least one of the settings was not satisfactory to provide a SonoAVC volume calculation. In three of these 10 cases the calculations failed with all three of the SonoAVC settings due to poor quality of the stored image, presence of shadows or inaccuracies of the software.

In the 55 cases measured with VOCAL and SonoAVC, the gestational sac volume calculated using VOCAL increased with advancing gestational age, from a mean of 50.4 cm3 at 77 days' gestation to 78.9 cm3 at 97 days' gestation (gestational sac volume = −59.324 + 1.425 × gestational age in days; r = 0.400, P < 0.002, SD = 16.568). The mean delta value of gestational sac volume calculated using VOCAL was −0.02 (95% CI, −4.457 to 4.500). The mean differences of the delta values between volume calculations using VOCAL and SonoAVC with each of three different settings are shown in Table 1 and Figure 3.

thumbnail image

Figure 3. Error plot showing the mean and 95% CI of the delta gestational sac volume (GSV) calculated with VOCAL and with SonoAVC with medium, high and low growth settings.

Download figure to PowerPoint

Table 1. Mean difference in delta values of gestational sac volume between VOCAL and SonoAVC calculations and between SonoAVC calculations with three different growth settings
MethodMean difference in delta values (cm3 (95% CI))P*
  • *

    t-test. High, high growth; low, low growth; mid, medium growth.

VOCAL vs. SonoAVC mid1.86 (−4.468 to 8.101)0.568
VOCAL vs. SonoAVC high−1.29 (−7.611 to 5.014)0.684
VOCAL vs. SonoAVC low4.97 (−1.409 to 11.368)0.125
SonoAVC mid vs. high−3.11 (−9.449 to 3.220)0.332
SonoAVC mid vs. low3.16 (−3.247 to 9.574)0.300
SonoAVC high vs. low6.27 (−0.160 to 12.716)0.056

The mean delta values of gestational sac volume calculations using SonoAVC were not different when using medium and high growth settings (−1.79 vs. 1.32; P = 0.332) or medium and low growth settings (−1.79 vs. −4.96; P = 0.300); however, the level of significance was marginal when comparing high and low growth settings (1.32 vs. −4.96; P = 0.056). The mean differences in delta values of the three SonoAVC settings are shown in Table 1.

The 95% limits of agreement and bias for SonoAVC volume calculation in 30 paired measurements by the same operator with medium growth setting are shown in Table 2 and Figure 4 and and those by two operators with SonoAVC settings of medium, high and low growth are shown in Table 2 and Figure 5. The ICC analysis showed the correlation between gestational sac volume calculation by VOCAL and by SonoAVC to be good. For the correlation between VOCAL and SonoAVC with medium growth setting, the ICC was 0.858 (95% CI, 0.769–0.915), for that between VOCAL and SonoAVC with high growth setting it was 0.882 (95% CI, 0.806–0.929) and for that between VOCAL and SonoAVC with low growth setting it was 0.924 (95% CI, 0.873–0.955).

thumbnail image

Figure 4. Mean percentage difference and 95% limits of agreement between paired calculations of gestational sac volume (GSV) using SonoAVC with a medium growth setting performed by the same operator.

Download figure to PowerPoint

thumbnail image

Figure 5. Mean percentage difference and 95% limits of agreement between paired calculations of gestational sac volume (GSV) using SonoAVC with medium (a), high (b) and low (c) growth settings performed by two different operators.

Download figure to PowerPoint

Table 2. Mean percentage difference and 95% limits of agreement (LOA) between paired measurements of gestational sac volume by the same operator (intraobserver) made with SonoAVC with a medium setting and by two different operators (interobserver) made with SonoAVC with medium, high and low growth settings
SonoAVC reproducibilityMean % difference (95% LOA)
  1. High, high growth setting; low, low growth setting; mid, medium growth setting.

Intraobserver−1.50 (−9.315 to 6.315)
Interobserver mid−0.54 (−9.339 to 8.247)
Interobserver high0.13 (−8.879 to 9.153)
Interobserver low0.09 (−10.615 to 10.812)

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. REFERENCES

Our findings show that: (1) gestational sac volume calculation is feasible with the novel automated tool SonoAVC with at least one selected growth setting in about 95% of cases; (2) volume calculations obtained by SonoAVC and VOCAL are similar and (3) assessment of gestational sac volume with SonoAVC is reproducible. The intra- and interobserver reproducibility showed a 95% agreement of measurements within 20%.

Like all other 3D tools, SonoAVC relies on good quality of the 2D image. It is not always possible to standardize such quality, mainly because of the different abilities of the maternal tissues to conduct the ultrasound beam. According to our data, any of the three growth settings could be used for gestational sac volume calculation, there being no statistical difference between the volume calculations of each of the SonoAVC settings and VOCAL. However, the volumes calculated with the SonoAVC low growth setting were close to being significantly different from those calculated with VOCAL; it is possible that with a larger number of cases the difference would become significant. In addition, the difference between SonoAVC high and low growth settings was only marginal. With the low growth setting, discrimination between fluid-filled and surrounding structures is reduced, requiring more post-process steps, and therefore negatively affecting the applicability of the method. It is likely that future studies will find for each fetal structure that one particular setting provides the most accurate volume calculation.

SonoAVC was developed to provide automated follicle counting following ovarian stimulation1. Rizzo et al.3 found that it can be used to calculate the volume of the fetal stomach, which has a shape similar to that of the ovarian follicles. They reported encouraging results in favor of SonoAVC compared with VOCAL, with a high success rate and less time spent for the calculation. Our finding that in about 95% of cases SonoAVC could be used for gestational sac volume calculation agrees with previous studies2, 3; however, the success rate may differ depending on the SonoAVC setting used. Reasons for the failures of SonoAVC in our study were the presence of shadows and adjacent anechoic and echoic structures, such as the maternal bladder and the irregularly shaped fetus within the gestational sac. This is a disadvantage of SonoAVC compared with VOCAL, since in the latter the operator can use the other two planes of the multiplanar mode for orientation and to avoid mistakes during the tracing of the structure in the presence of shadows. It seems critically important for SonoAVC calculations to avoid shadows during volume acquisition.

Falcon et al.7 found that measurement of the gestational sac volume using VOCAL in 500 cases was not a useful predictor of fetal aneuploidy. It was not the purpose of our study to reassess the usefulness of this measurement, but to assess the accuracy of this new tool for volume calculation of fluid-filled structures in prenatal ultrasound. In contrast to their study, we performed the gestational sac volume calculation excluding fetal structures, which appears to be easier to do with SonoAVC than with VOCAL. Several prenatal ultrasound studies that focused on the utilization of VOCAL for assessment of fetal anechoic structures have validated the method4–10. Some of them reported a longer time needed to obtain the measurements compared with other methods of volume calculation3, 12, 13. SonoAVC may be useful in reducing the time for volume calculations and for the learning process of sonographers with experience in 3D ultrasound. We report good results for reproducibility of the gestational sac volume calculation using SonoAVC, comparable to those previously reported for VOCAL in vivo and in vitro and in previous publications about SonoAVC3–5, 10.

It is possible that in the future SonoAVC will become a useful tool in prenatal diagnosis; however, at present more studies are required to validate this technique. It is likely that validation in gynecological studies will aid the development of standards for prenatal diagnosis.

REFERENCES

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. REFERENCES