Comparative evaluation of manual and VOCAL methods for amniotic sac volume measurement in early pregnancy

Introduction

The amniotic sac is a vital structure in early human development, serving both protective and regulatory functions for the embryo during the critical first trimester of pregnancy [1], [2], [3]. Accurate assessment of its volume provides significant diagnostic value, offering insight into gestational development and helping identify deviations that may indicate complications such as miscarriage, chromosomal abnormalities, or structural anomalies [4], [5], [6], [7]. The gestational period between 7 and 12 weeks represents a particularly dynamic window of fetal growth, during which subtle alterations in fluid volume or sac morphology may carry important clinical implications.

Traditionally, early pregnancy assessments have relied on two-dimensional (2D) ultrasound measurements such as amniotic sac diameter (ASD) and gestational sac diameter (GSD). Although widely used, these linear parameters are limited by geometric assumptions and an inability to fully capture the irregular shape of the developing amniotic sac [8], [9], [10]. These limitations contribute to inaccurate volume estimations and increased interobserver variability, particularly in routine clinical settings where imaging quality and operator skill can vary [1], 2], 8].

Advancements in three-dimensional (3D) ultrasonography have markedly improved the precision of fetal and embryonic measurements. Among these technologies, Virtual Organ Computer-aided Analysis (VOCAL) has emerged as a validated and practical method for volumetric assessment in obstetrics. VOCAL uses rotational axis tracing to reconstruct anatomical structures, reducing operator dependence and enhancing reproducibility – especially useful when measuring the complex contours of the amniotic sac in early pregnancy [11], 12]. Numerous studies have confirmed VOCAL’s reliability, showing it to outperform manual methods in both precision and consistency [3], 10], 13].

Prior work by Solangon et al. established reference values for ASD in early gestation using linear approaches [1]. While clinically useful, such measurements do not capture the full diagnostic potential of volumetric assessment. Despite increasing awareness of amniotic sac volume (ASV) as a sensitive indicator of fetal well-being, few studies have mapped its normative development using VOCAL, and even fewer have directly compared VOCAL with manual methods over a broad gestational range.

Manual volume estimation remains common in many clinical environments, yet it suffers from limited accuracy and considerable interobserver variability, due to its reliance on simplified geometric formulas and subjective image interpretation [6], 14], 15]. In resource-limited settings where access to 3D imaging is constrained, understanding the comparative performance of manual methods vs. advanced techniques like VOCAL is essential. Establishing such comparative data can help clinicians choose the most appropriate method for early gestational screening in varying clinical contexts.

This study aims to establish reference intervals for ASV between 7 and 12 weeks of gestation, evaluate the accuracy and reproducibility of VOCAL vs. manual measurement, and analyze interobserver and intraobserver variability for both techniques. We hypothesize that VOCAL will demonstrate superior accuracy and reliability in volumetric assessment, although manual methods may retain clinical utility in settings with technological limitations. By addressing this gap, the research seeks to strengthen early pregnancy diagnostic frameworks and promote standardization in ultrasound-based volumetric assessment.

Although previous studies have contributed foundational data using diameter-based assessments, volumetric measurement – especially via VOCAL – remains underexplored. This study adds to the literature by presenting normative ASV reference values for early pregnancy using both VOCAL and manual methods, while directly comparing their diagnostic performance. These findings may guide clinical practice, improve consistency in prenatal care, and provide a platform for future research on abnormal pregnancies.

Materials and methods

Ultrasound image acquisition and volume measurement

Transvaginal 3D ultrasound imaging was performed during routine clinical evaluations [17]. For each participant, 3D datasets of the amniotic sac were acquired and stored for offline analysis. Figure 1 depicts a representative 3D ultrasound image demonstrating fetal and extraembryonic structures at 8–9 weeks’ gestation, providing anatomical context for subsequent measurements. All measurements were conducted by two certified sonographers experienced in obstetric ultrasonography.

Figure 1:

3D ultrasound image of fetal and extraembryonic structures at 8–9 weeks gestation. This 3D ultrasound image demonstrates the structural relationship between the fetus and extraembryonic components during early gestation. Key features such as the head, body, upper and lower limbs, amniotic sac, amniotic membrane, amniotic cavity, and gestational sac are clearly labeled. The fetus is shown encapsulated within the amniotic sac, with the amniotic membrane visibly separating it from the surrounding gestational space. Image rendered using 3D surface rendering mode (Silhouette), which enhances structural contrast and minimizes acoustic artifacts for improved anatomical clarity.

Manual method

Manual volume estimation was based on two-dimensional measurements of the amniotic sac in three perpendicular planes – longitudinal, transverse, and anteroposterior. The ellipsoid formula was applied to calculate volume: V=4/3π × L/2 × T/2 × AP/2 where L is longitudinal, T is transverse, and AP is anteroposterior diameter. Figure 2 illustrates manual ASV measurements at 7 and 11 weeks’ gestation, highlighting application of this technique.

Figure 2:

Manual amniotic sac volume estimation at 7 and 11 Weeks gestation using 3D ultrasound. This image displays manual ASV measurements at two gestational ages: Panel A shows 7w6d and panel B shows 11w1d. Measurements were obtained by manually tracing the amniotic sac borders across three orthogonal planes, with visible calipers marking CRL and sac dimensions. The fetus and amniotic cavity are identifiable in both stages of development. Images rendered using 3D surface rendering mode (Silhouette) to enhance contour accuracy and reduce visualization artifacts.

VOCAL method

Volume acquisition was performed using a 30° rotational angle, which balances detail and acquisition efficiency in first-trimester volumetric ultrasound. The VOCAL technique involved manual tracing of the amniotic sac boundary across multiple rotational planes, with subsequent automated 3D volume reconstruction. Figure 3 illustrates the VOCAL-based measurement process at 8 weeks of gestation. This method compensates for irregular sac shapes and reduces operator dependency, improving precision and reproducibility.

Figure 3:

VOCAL-based amniotic sac volume measurement at 8 weeks gestation. The left panel displays three orthogonal planes from transvaginal ultrasound showing the manually traced amniotic sac in sagittal (A), coronal (B), and transverse (C) views. The right panel shows the 3D reconstructed amniotic sac using the VOCAL method, visualized as a green volume. Traced borders correspond to sac contours on each plane, and volume is calculated from rotational analysis. Rendered using the VOCAL surface reconstruction mode, this method improves accuracy by modeling irregular sac geometry in 3D and reducing dependency on geometric assumptions. All key anatomical boundaries are labeled.

Results

This study analyzed amniotic sac dimensions and volumes across 68 participants with gestational ages ranging from 7 weeks 0 days to 12 weeks 6 days. Mean crown-rump length (CRL), amniotic cavity (AC) diameter, and amniotic cavity volume measurements demonstrated progressive increases with advancing gestational age, confirming expected early gestational growth trajectories. Figure 4 presents representative 3D ultrasound images from 7 to 12 weeks, illustrating progressive fetal and amniotic cavity development. These images reinforce the strong correlation between gestational age and amniotic/fetal expansion.

Figure 4:

Serial 3D ultrasound images depicting amniotic Sac and fetal Development from 7 to 12 weeks gestation. This progression series illustrates key stages of first-trimester fetal development using surface-rendered 3D ultrasound imaging. Each panel represents a distinct gestational week (7–12 weeks) and highlights anatomical changes in the fetus, amniotic sac, yolk sac, and surrounding structures. Visible features include the evolving head contour, limb buds, and overall fetal morphology, along with expansion of the amniotic cavity. Images were acquired using 3D surface rendering mode to enhance spatial contrast and minimize artifacts. All key structures are labeled to aid anatomical interpretation.

Amniotic cavity diameter exhibited a strong positive linear correlation with gestational age, modeled by the equation: AC (cm)=0.980 × Gestational Age − 6.008 with r=0.999, R²=0.999, p<0.001 (Figure 5). Similarly, AC diameter was strongly correlated with CRL: AC (cm)=1.052 × CRL + 0.234, r=0.999, R²=0.999, p<0.001 (Figure 6). Regression models based on CRL demonstrated predictive accuracy comparable to those based on gestational age, further validating CRL as a robust parameter for early volumetric assessment.

Figure 5:

The linear regression shows a strong positive correlation between gestational age (weeks) and amniotic cavity diameter (cm), with the equation: Diameter=0.980 × gestational age − 6.008. The analysis reveals a correlation coefficient of 0.999, an R-squared value of 0.999, and a statistically significant p-value of 0.001, indicating that 99.9 % of the variation in amniotic cavity diameter is explained by gestational age.

Figure 6:

The linear regression analysis between crown-rump length (CRL, cm) and amniotic cavity diameter (cm) shows a strong positive correlation, with the equation: Diameter=1.052 × CRL + 0.234. The analysis reveals a correlation coefficient of 0.999, an R-squared value of 0.999, and a statistically significant p-value of 0.001, indicating that 99.9 % of the variation in amniotic cavity diameter is explained by CRL.

In contrast, the gestational sac-to-amniotic cavity (GS/AC) diameter ratio followed a non-linear (curvilinear) declining trend with both gestational age and CRL. The best-fit second-degree polynomial equations were: GS/AC Ratio=0.2165 × (GA)² – 5.1874 × (GA)+32.6026 for gestational age (R²=0.666; Figure 7), and GS/AC Ratio=0.4114 × (CRL)² – 5.6461 × (CRL)+19.3965 for CRL (R²=0.613; Figure 8). These models highlight the non-linear expansion dynamics of the amniotic cavity relative to the gestational sac during early pregnancy.

Figure 7:

Demonstrates a second-degree polynomial regression model describing the relationship between gestational age (GA) and the gestational sac-to-amniotic cavity (GS/AC) diameter ratio. The fitted curvilinear equation is: GS/AC ratio=0.2165 × (GA)2 – 5.1874 × (GA) + 32.6026. The model shows a non-linear decreasing trend with a coefficient of determination (R²) of 0.666, indicating that 66.6 % of the variation in the GS/AC diameter ratio can be explained by gestational age. This curvilinear trend reflects the accelerated expansion of the amniotic cavity compared to the gestational sac in early pregnancy.

Figure 8:

Presents a second-degree polynomial regression analysis illustrating the relationship between crown-rump length (CRL) and the gestational sac-to-amniotic cavity (GS/AC) diameter ratio. The fitted curvilinear equation is: GS/AC ratio=0.4114 × (CRL)2 – 5.6461 × (CRL) + 19.3965. The model demonstrates a non-linear decline in GS/AC diameter ratio as CRL increases, with a coefficient of determination (R²) of 0.613. This trend emphasizes the rapid expansion of the amniotic cavity relative to the gestational sac as embryonic development progresses during the first trimester.

Amniotic sac volumes measured using both VOCAL and manual techniques also showed strong positive correlations with gestational age and CRL. Manually measured sac volumes followed the linear model: Volume (cc)=24.935 × Gestational Age − 180.334 with r=0.995, R²=0.990, p<0.001 (Figure 9). VOCAL-based volume measurements were modeled by: Volume (cc)=2.244 × Gestational Age − 17.561, r=0.997, R²=0.994, p<0.001, while manual measurements (fetus included) followed: Volume (cc)=21.364 × Gestational Age − 169.509, r=0.993, R²=0.986, p<0.001.

Figure 9:

The linear regression analysis between gestational age (weeks) and manually measured gestational sac (GS) volume (cc) shows a steady increase in GS volume, with the equation: Volume=24.935 × gestational age − 180.334. The analysis reveals a correlation coefficient of 0.995, an R-squared value of 0.990, and a statistically significant p-value of 0.001, indicating that 99.0 % of the variation in GS volume is explained by gestational age.

Curvilinear regression analyses between gestational age or CRL and the GS/AC volume ratio (manual) are shown in Figures 10 and 11, demonstrating decreasing ratios as gestation progresses (R²=0.813 and R²=0.761, respectively). Figure 12(A–D) further illustrates curvilinear growth models for amniotic cavity volume, with polynomial models providing a better fit than linear approximations, particularly beyond 9 weeks of gestation.

Figure 10:

Non-linear regression analysis showing the relationship between gestational age (in weeks) and the ratio of gestational sac volume (GS) to amniotic cavity volume (AC), measured manually. A second-degree polynomial (curvilinear) model was applied, providing a better fit than a linear model to capture the biological progression. The analysis revealed a statistically significant association, with the GS/AC volume ratio decreasing as gestational age increases (R²=0.813). This trend reflects the relatively rapid expansion of the amniotic cavity during early pregnancy compared to the gestational sac.

Figure 11:

Non-linear regression analysis showing the association between crown-rump length (CRL, cm) and the ratio of gestational sac volume (GS) to amniotic cavity volume (AC), based on manual measurements. A second-degree polynomial (curvilinear) regression was applied to better reflect the natural trend, which is not adequately captured by a linear model. The analysis demonstrates a significant inverse relationship, where the GS/AC volume ratio decreases with increasing CRL. This pattern is consistent with the accelerated development of the amniotic cavity relative to the gestational sac during early pregnancy. The model achieved a strong fit with R²=0.761.

Figure 12:

Curvilinear modeling of amniotic cavity volume in early pregnancy. To better reflect the non-linear growth pattern of the amniotic cavity in early pregnancy, second-degree polynomial regression models were applied to key clinical variables, replacing the initial linear analyses. (A) Shows a strong curvilinear relationship (R²=0.997) between gestational age and manually measured amniotic cavity (AC) volume. The model captures the accelerated growth trend that a linear fit failed to represent. (B) Presents the same relationship using VOCAL measurements, with a similarly strong fit (R²=0.996). This confirms that even with more precise techniques, amniotic volume growth follows a non-linear pattern. (C) Demonstrates the correlation between crown-rump length (CRL) and manual AC volume. The curvilinear model (R²=0.997) accurately reflects volumetric changes as the fetus develops. (D) Applies the same model to CRL and VOCAL-based AC volume, again achieving excellent fit (R²=0.997). This supports CRL as a reliable predictor of amniotic volume, particularly when measured with high-precision methods. Together, these models highlight the non-linear nature of early amniotic volume growth and reinforce the importance of using appropriate regression techniques in both manual and VOCAL-based assessments.

In addition to regression modeling, reference intervals were established for key amniotic measurements. Table 4 presents the 5th percentile, median, and 95th percentile values for AC diameter, GS/AC diameter ratio, amniotic sac volume (manual), and GS/AC volume ratio across gestational weeks 7–11. These intervals reflect expected biological trends during early pregnancy, providing clinically useful benchmarks for the evaluation of amniotic development.

Error analysis revealed significantly greater variability in manual measurements compared to VOCAL-based assessments. Mean Absolute Error (MAE) and Root Mean Square Error (RMSE) were lower for VOCAL (MAE: 0.35 cc; RMSE: 0.6 cc) than for manual measurement (MAE: 16.94 cc; RMSE: 26.19 cc), highlighting VOCAL’s superior precision.

Initial Bland–Altman analysis revealed increasing differences between manual and VOCAL methods with larger volumes, violating assumptions of constant bias. Therefore, corrected percentage Bland–Altman plots were constructed (Figures 13 and 14). These confirm a systematic overestimation bias in manual measurements and demonstrate significant proportional bias, particularly at higher volume ranges. The regression lines included in the plots indicate that the difference between methods is not constant, and caution is warranted when interpreting manual data across the full measurement spectrum. Observer 1 demonstrated a mean percentage bias of 52.11 %, while Observer 2 showed a mean percentage bias of 54.77 %.

Figure 13:

Percentage Bland–Altman plot comparing manual (observer 1) and VOCAL (observer 2) measurements of amniotic sac volume. The mean percentage difference (red line) indicates systematic overestimation by the manual method. Limits of agreement (blue dashed lines) reflect the 95 % confidence interval. A regression line (green) reveals a modest but persistent proportional bias, underscoring the need for caution when interpreting manual measurements, especially at higher volumes.

Figure 14:

Percentage Bland–Altman plot comparing manual (observer 2) and VOCAL (observer 2) measurements of amniotic sac volume. The red line represents the mean percentage difference, suggesting systematic overestimation by the manual method. Blue dashed lines represent the 95 % limits of agreement. A mild upward trend (green line) reflects residual proportional bias, particularly at higher volume measurements, warranting cautious interpretation.

Intraclass correlation coefficients (ICC) further supported these findings:

1. Intraobserver reliability: VOCAL=0.846 (95 % CI: 0.75–0.92); Manual=0.451 (95 % CI: 0.30–0.60)

2. Interobserver reliability: VOCAL=0.910 (95 % CI: 0.85–0.96); Manual=0.488 (95 % CI: 0.32–0.65) (Table 1) Note: The presence of proportional bias in manual measurements limits the interpretability of ICC values, especially across varying volume ranges.

Cohen’s Kappa values indicated high categorical agreement for both methods (κ=0.961 for VOCAL and κ=1.0 for Manual; Table 2), although perfect agreement in the manual method likely reflects categorical oversimplification rather than true measurement consistency.

Paired t-tests confirmed significant differences between VOCAL and manual measurements (mean difference=22.88 cc; 95 % CI: 17.65–28.11; t=7.92; p<0.001). Repeated Measures ANOVA demonstrated a significant method effect (F=47.834, p<0.001; η²=0.64), affirming VOCAL’s superior consistency.

Overall, these findings support VOCAL as a more accurate, reproducible, and clinically valuable technique for ASV measurement in early pregnancy, especially where precision is critical. The corrected Bland–Altman approach further underscores the need for caution when applying manual methods in high-volume cases or for diagnostic thresholds.

Discussion

This study presents a comprehensive comparison between manual and VOCAL techniques for assessing amniotic sac volume (ASV) in early pregnancy, specifically from 7 to 12 weeks’ gestation. The findings clearly demonstrate the superior accuracy, consistency, and reproducibility of the VOCAL method, reinforcing its clinical value in maternal-fetal medicine. The exceptionally high correlation coefficients (R²=0.999) between ASV and key parameters such as gestational age (GA) and crown-rump length (CRL) reflect the internal validity of the dataset, likely facilitated by standardized imaging protocols, high-resolution technology, and homogeneity of the study population.

Importantly, regression models using CRL yielded results comparable to those based on GA, reaffirming CRL’s utility not only in gestational dating but also as a strong and reliable predictor of ASV in early pregnancy. These findings are in line with prior literature [5], [18], [19], [20], supporting CRL as a cornerstone in both linear and volumetric assessment strategies.

Accurate ASV measurement plays a pivotal role in identifying deviations from normal growth patterns that may signal adverse pregnancy outcomes. Conditions such as miscarriage, oligohydramnios, subchorionic hematoma, and chromosomal anomalies are known to affect early gestational fluid dynamics [6], 7], 21]. Precise volumetric assessment thus facilitates early detection, guiding risk stratification and timely intervention [22], [23], [24], [25].

Building upon the work of Solangon et al. [1], who established reference intervals for amniotic sac diameter using 2D ultrasound, this study advances the field by employing volumetric methods that account for irregular sac morphology. The VOCAL technique, in particular, offers more accurate modeling of amniotic geometry, consistent with findings from prior 3D ultrasound research [3], 10], 13]. By directly comparing VOCAL with manual methods, this study provides important normative data and further validates VOCAL’s superiority in terms of precision and interobserver agreement.

While manual methods remain commonly used due to their accessibility and simplicity, they demonstrated consistent overestimation of ASV and substantially greater variability. Table 3 highlights these limitations through intraclass correlation coefficient (ICC) comparisons. Much of this variability arises from reliance on geometric assumptions (e.g., the ellipsoid formula) and operator-dependent variability [14], 15], 26]. Despite these drawbacks, manual methods may retain clinical value in low-resource settings, particularly when supported by computational tools to enhance consistency.

A critical methodological refinement involved the evaluation of agreement between methods. Initial Bland-Altman analysis using raw differences revealed a marked proportional bias in manual measurements, with increasing overestimation at higher volumes. This violated the core assumptions of Bland–Altman analysis (constant bias and homoscedasticity), prompting the adoption of percentage Bland–Altman plots. These corrected visualizations showed that manual overestimation is not constant but grows significantly with volume, underscoring the importance of cautious interpretation in higher-volume cases. Consequently, ICC values alone may inadequately represent the reliability of manual measurements, especially across a wide volume range.

The presence of curvilinear growth patterns, particularly in the gestational sac-to-amniotic cavity (GS/AC) diameter and volume ratios, further supports the need for advanced modeling. Second-degree polynomial regression provided a more accurate representation of the amniotic cavity’s accelerated expansion, capturing biological dynamics that linear models failed to reflect. This reinforces the utility of non-linear modeling in early pregnancy assessments.

Despite its clear advantages, VOCAL’s implementation may be constrained by cost, equipment availability, and the need for operator training – barriers that are more pronounced in resource-limited settings. Nonetheless, emerging solutions such as portable 3D ultrasound systems, AI-assisted segmentation, and automated real-time volume measurement offer promising strategies to expand access [27], [28], [29], [30], [31], [32], [33]. Hybrid approaches that combine manual methods with artificial intelligence could further bridge this gap by enhancing measurement precision without requiring full reliance on high-end equipment.

This study acknowledges several limitations. While adequate for modeling, the sample was restricted to structurally normal pregnancies. Future research should extend to diverse populations, including pregnancies complicated by miscarriage, fluid anomalies, or embryonic demise, to test the robustness and clinical relevance of ASV measurements [7], 17], 25], 34], 35]. Longitudinal studies correlating early ASV trends with perinatal outcomes would further clarify the predictive value of volumetric analysis.

Ethical considerations must accompany the adoption of advanced imaging technologies. Ensuring equitable access to VOCAL and emerging tools is essential to prevent disparities in prenatal care. Additionally, the growing role of AI in ultrasound interpretation requires transparency, validation, and clinical oversight to ensure unbiased, safe, and trustworthy decision-making.

In conclusion, this study affirms VOCAL as a highly accurate and reproducible method for ASV measurement in early pregnancy. Its precision and strong correlation with key gestational markers make it ideal for both routine and high-risk prenatal assessments. The corrected Bland–Altman analysis reveals important limitations in manual methods, particularly related to proportional bias, which must be acknowledged when applying manual techniques clinically. By advancing volumetric measurement practices and embracing innovation, VOCAL holds promise for elevating the standards of maternal-fetal care globally.

Future directions

Future research should aim to expand the study population to encompass larger and more heterogeneous cohorts, including pregnancies with abnormal outcomes. Comparative analyses beyond the first trimester would further elucidate the longitudinal evolution of ASV and its clinical implications. Particular focus should be given to high-risk pregnancies, such as those complicated by fetal anomalies, early pregnancy loss, or amniotic fluid abnormalities, to validate the utility of ASV as a biomarker for adverse outcomes. Advances in portable 3D ultrasound technology, along with AI-assisted volume measurement tools, should be explored to facilitate broader access and improve diagnostic precision in resource-constrained environments. Finally, long-term, prospective studies linking early volumetric trends to perinatal outcomes are essential to fully establish the clinical relevance of ASV measurements within real-world clinical workflows and to refine early pregnancy risk stratification protocols.

Conclusions

This study affirms the clinical value and diagnostic precision of the Virtual Organ Computer-aided Analysis (VOCAL) method for measuring amniotic sac volume (ASV) during early pregnancy. VOCAL demonstrated superior accuracy, reproducibility, and consistency compared to traditional manual techniques, which remain limited by geometric assumptions and operator variability. By providing more reliable volumetric data, VOCAL enhances early gestational assessment and may support the early detection of pregnancy-related complications. Importantly, while VOCAL requires specialized equipment and operator expertise, it offers a relatively accessible and cost-effective alternative to more complex volumetric technologies. Its adaptability and potential integration into routine obstetric workflows suggest that it can help bridge diagnostic gaps in both high-resource and low-resource clinical settings. The high intraclass correlation coefficients and strong statistical agreement observed in this study reinforce VOCAL’s suitability for widespread clinical and research applications.

Furthermore, the alignment of ASV measurements with CRL-based predictions enhances VOCAL’s clinical relevance, particularly in early pregnancy assessments where CRL is already a fundamental parameter. Notably, the incorporation of curvilinear modeling for the gestational sac-to-amniotic cavity (GS/AC) ratio adds to the physiological accuracy of early pregnancy evaluations, capturing dynamic developmental changes that linear models may fail to represent. Looking forward, VOCAL’s role in maternal-fetal medicine could be further strengthened through integration with artificial intelligence and automated image analysis systems. These innovations could streamline implementation, democratize access, and standardize operator training across diverse healthcare environments. Future research should focus on validating VOCAL’s performance in high-risk or abnormal pregnancies and investigating its predictive value for long-term maternal and fetal outcomes.

In conclusion, this study positions VOCAL as a pivotal advancement in prenatal diagnostics. Its superior measurement performance not only redefines current imaging protocols but also sets a new benchmark for early pregnancy evaluation. With continued technological innovation and broader validation, VOCAL is poised to become an essential tool for advancing the precision, equity, and quality of maternal-fetal healthcare globally.