Endometrial Receptivity Beyond Thickness: The Predictive Role of 3D Ultrasound and Doppler Parameters in IVF

Background: Two-dimensional endometrial thickness (EMT) is a weak and inconsistent predictor of live birth in IVF, prompting interest in whether three-dimensional (3D) ultrasound, particularly endometrial and subendometrial volume and power Doppler vascularity indices, and uterine artery Doppler velocimetry can provide additional, clinically useful information about endometrial receptivity beyond thickness alone. 

Objective: This narrative review synthesizes the available evidence on 3D power Doppler vascularity parameters (vascularization index [VI], flow index [FI], and vascularization flow index [VFI]) and uterine artery Doppler indices as predictors of pregnancy and live birth in IVF. 

Methods: A structured search of academic databases and search engines was conducted for primary studies and systematic reviews/meta-analyses examining 3D ultrasound or Doppler parameters in relation to pregnancy, implantation, or live birth outcomes in IVF and frozen embryo transfer. Six primary studies and one systematic review/meta-analysis met inclusion criteria. 

Results: Findings across 3D power Doppler studies are heterogeneous and partly time- and context-dependent. Several early cohorts found that subendometrial or endometrial vascularity indices, particularly VFI, discriminated between pregnant and non-pregnant cycles and outperformed endometrial volume, but this predictive value in one prospective cohort held only when few or no top-grade embryos were available for transfer, and in a separate frozen embryo transfer cohort no 3D power Doppler parameter discriminated pregnancy outcome when measured at a single time point. A study directly comparing live birth with miscarriage found endometrial vascularization index to be an independent predictor of live birth even after adjustment for confounders, and more recent work has shifted toward composite, multiparametric receptivity scoring systems and non-linear, optimal-range modelling of vascularity indices rather than single cutoffs. A recent meta-analysis of uterine artery Doppler found that women who achieved clinical pregnancy had significantly lower pulsatility index and higher peak systolic velocity than those who did not.

Discussion: The evidence suggests that 3D and Doppler parameters carry real, biologically plausible prognostic information about endometrial receptivity, but that this information is context-dependent, sensitive to measurement timing, and most useful as part of a composite assessment or in specific clinical subgroups (e.g., limited embryo quality, distinguishing live birth from miscarriage) rather than as a universal, single-index replacement for endometrial thickness.

Conclusion: 3D ultrasound vascularity indices and uterine artery Doppler parameters add biologically meaningful, but inconsistent and context-dependent, predictive information beyond endometrial thickness; multiparametric and composite modelling approaches appear more promising than any single 3D or Doppler index used in isolation, but current evidence remains insufficiently standardized to support routine clinical adoption of a specific cutoff or algorithm.

Endometrial thickness (EMT), measured by two-dimensional transvaginal ultrasound, remains the most widely used clinical marker of endometrial receptivity in IVF, but accumulating evidence indicates that EMT performs poorly as an individual-patient predictor of live birth, with several well-designed cohorts finding no meaningful discriminatory ability once other factors are considered. This has motivated interest in whether more sophisticated ultrasonographic techniques, particularly three-dimensional (3D) power Doppler angiography and uterine artery Doppler velocimetry, can capture additional, functionally relevant information about endometrial receptivity that a simple linear thickness measurement cannot.

3D power Doppler ultrasound allows volumetric, semi-quantitative assessment of tissue vascularity within a defined region of interest, typically summarized through three indices: the vascularization index (VI, the proportion of color-coded voxels representing blood vessels within the tissue volume, reflecting vessel density), the flow index (FI, the average power Doppler signal intensity within those vessels, reflecting the intensity of blood flow), and the vascularization flow index (VFI, a composite of the two, reflecting both vessel density and flow intensity). These indices can be measured separately in the endometrial and subendometrial (junctional zone) regions, which are proposed to reflect different aspects of the vascular supply relevant to implantation. Separately, uterine artery Doppler assesses upstream vascular resistance through indices such as the pulsatility index (PI) and resistance index (RI), and flow velocity through the peak systolic velocity (PSV), providing a complementary, more proximal measure of uterine perfusion.

The physiological rationale for these techniques is strong: adequate endometrial and subendometrial blood flow is considered essential for the nutrient and oxygen supply that supports decidualization and implantation, and reduced uterine perfusion has been mechanistically linked to implantation failure. However, translating this physiological plausibility into a clinically useful predictive tool has proven difficult, with published studies reporting inconsistent, sometimes directly contradictory, findings regarding which parameters, if any, reliably predict pregnancy or live birth. This review aims to (a) synthesize the available evidence on 3D power Doppler vascularity indices and uterine artery Doppler parameters as predictors of pregnancy and live birth in IVF; (b) identify the conditions (measurement timing, embryo quality, outcome definition) under which these parameters appear to add predictive value beyond endometrial thickness; and (c) evaluate the field's progression toward composite, multiparametric assessment approaches.

This is a narrative review rather than a formally registered PRISMA systematic review. A structured search was conducted across academic search engines and publisher platforms combining terms related to the exposure ("three-dimensional ultrasound," "3D power Doppler," "vascularization index," "flow index," "VFI," "uterine artery Doppler," "pulsatility index") with terms related to the population and outcome ("endometrial receptivity," "IVF," "frozen embryo transfer," "pregnancy," "live birth," "implantation"). Reference lists of retrieved reviews were also hand-searched.

Inclusion criteria comprised studies that (a) reported primary empirical data using 3D power Doppler ultrasound or uterine artery Doppler velocimetry in relation to pregnancy, implantation, or live birth outcomes in IVF or frozen embryo transfer, or were themselves systematic reviews/meta-analyses; and (b) were available in English. Studies examining only two-dimensional endometrial thickness or pattern without a 3D or Doppler vascularity component, and studies focused exclusively on non-IVF populations (e.g., natural-cycle infertility without ART), were excluded, with the exception of foundational endometrial-thickness literature retained for comparative framing.

Six primary studies and one systematic review/meta-analysis met inclusion criteria and form the basis of the synthesis below (Table 1). For each study, the following were extracted: authorship and year, design, sample, the specific parameters assessed, and the key finding. Given substantial heterogeneity in measurement timing (hCG day, LH+1, embryo transfer day), cycle type (fresh vs. frozen), outcome definition (biochemical/clinical pregnancy vs. live birth vs. miscarriage), and analytic approach, a narrative rather than a meta-analytic synthesis was undertaken, organized around three evidence strands: (a) early foundational 3D power Doppler cohorts with mixed findings, (b) more recent composite and non-linear modelling approaches, and (c) uterine artery Doppler meta-analytic evidence.

Early 3D power Doppler cohorts: mixed and context-dependent findings

The foundational literature on 3D power Doppler vascularity indices in IVF presents a genuinely mixed picture. In one of the earliest prospective cohorts, subendometrial VFI outperformed vascularization index, flow index, and endometrial volume in receiver operating characteristic (ROC) analysis for predicting pregnancy, with the best discrimination achieved at a VFI cutoff greater than 0.24 (Wu, Chiang, Huang, Chao, Wang & Soong, 2003). A separate prospective cohort of 80 IVF/ICSI patients found that endometrial volume and power Doppler indices (VI, FI, VFI) were significantly higher in the pregnant group, with statistically significant ROC discrimination (area under the curve 0.72–0.83), but critically, this predictive value held only in the subgroup receiving zero or one top-grade embryo; when two or three top-grade embryos were transferred, embryo quality appeared to dominate and the Doppler indices lost their discriminatory power (Mercé, Barco, Bau & Troyano, 2008).

In direct contrast, a prospective cohort specifically examining frozen embryo transfer cycles found that endometrial thickness, pattern, volume, uterine artery PI/RI, and endometrial and subendometrial 3D power Doppler indices were all statistically similar between pregnant and non-pregnant groups, with ROC areas under the curve of approximately 0.5, no better than chance, for every ultrasound parameter tested; only patient age predicted pregnancy in this cohort (Ng, Chan, Tang, Yeung & Ho, 2006). The authors explicitly attributed this null finding to the limitation of measuring vascularity at only a single time point, raising the possibility that dynamic, repeated-measurement approaches might perform better than single-occasion assessment.

A related study by the same research group examined a different, arguably more clinically important, outcome distinction: live birth versus miscarriage among patients who had already achieved pregnancy. Among 161 pregnant patients following stimulated IVF and FET, of whom 28.0% subsequently miscarried, endometrial and subendometrial vascularity were significantly higher in those who went on to live birth than in those who miscarried, and endometrial vascularization index remained an independent predictor of live birth in multivariate analysis after adjustment for confounders (odds ratio 1.384, 95% CI 1.025–1.869) (Ng, Chan, Tang, Yeung & Ho, 2007). This suggests that 3D power Doppler vascularity may carry more consistent prognostic information for the maintenance of an established pregnancy than for the initial achievement of pregnancy or implantation itself.

Recent composite and non-linear modelling approaches

More recent work has moved away from testing single vascularity indices in isolation toward composite, multiparametric assessment. A prospective cohort study developed and evaluated a composite ultrasonographic endometrial receptivity scoring system incorporating endometrial thickness, pattern, and blood flow parameters together, rather than relying on any single measure, and found this composite score to be associated with pregnancy outcome following frozen-thawed embryo transfer (Ouyang et al., 2024), consistent with the broader pattern that no single 3D or Doppler parameter reliably predicts outcome across all contexts. Separately, a large retrospective cohort of 338 frozen embryo transfer patients used restricted cubic spline modelling to examine the relationship between endometrial VFI and pregnancy outcome across the full VFI distribution, finding a non-linear relationship and proposing an optimal VFI range rather than a single cutoff value, directly addressing the inconsistency in previously proposed VFI thresholds across the literature (Zhao, Che, Li, Song, Zhou, Long & Zhang, 2026).

Uterine artery Doppler: meta-analytic evidence

At the systematic review level, a recent meta-analysis pooling 12 studies and 3,317 women undergoing assisted reproductive technology found that women who achieved clinical pregnancy had a significantly lower pre-implantation uterine artery pulsatility index (mean difference −0.26, 95% CI −0.46 to −0.06) and a significantly higher peak systolic velocity (mean difference 8.59, 95% CI 2.31 to 14.87) than those who did not conceive, with the PI difference most pronounced when measured during the menstrual cycle preceding stimulation (Siargkas et al., 2025). This meta-analysis represents the most robust quantitative evidence currently available linking a specific Doppler parameter, upstream uterine artery resistance, to IVF outcome, though it should be noted this evidence concerns clinical pregnancy rather than live birth specifically.

Table 1 summarizes the design, sample, parameters assessed, and key findings of the seven studies included in this review.

Table 1: Characteristics of Included Studies on 3D Ultrasound and Doppler Parameters in Endometrial Receptivity Assessment

Note: ART = assisted reproductive technology; FET = frozen embryo transfer; FI = flow index; hCG = human chorionic gonadotropin; PI = pulsatility index; PSV = peak systolic velocity; RI = resistance index; ROC = receiver operating characteristic; UtA = uterine artery; VFI = vascularization flow index; VI = vascularization index.

The evidence reviewed here does not support a simple conclusion that 3D power Doppler or uterine artery Doppler parameters are either reliable predictors of IVF outcome or uninformative; rather, the literature indicates that their predictive value is highly context-dependent. Three specific sources of context-dependence emerge clearly from the included studies. First, measurement timing and dynamics matter: the null finding in frozen embryo transfer cycles (Ng et al., 2006) was explicitly attributed by its authors to single-time-point measurement, while the positive early finding using a single hCG-day measurement in a different fresh-cycle cohort (Wu et al., 2003) suggests that the relevant timing may differ between fresh and frozen cycle types, an important source of heterogeneity that complicates cross-study comparison. Second, embryo quality moderates the apparent predictive value of endometrial vascularity: in the Mercé et al. (2008) cohort, 3D power Doppler indices discriminated outcome only when few or no top-grade embryos were transferred, suggesting that a sufficiently high-quality embryo may implant successfully across a wider range of endometrial receptivity than a lower-quality embryo, effectively masking any endometrial vascularity signal when embryo quality is uniformly excellent. Third, outcome definition matters: vascularity indices showed clearer, independently significant predictive value when distinguishing live birth from miscarriage among already-pregnant patients (Ng et al., 2007) than when predicting whether pregnancy would occur at all, suggesting these markers may better capture processes related to the maintenance of early pregnancy than the initial implantation event.

The uterine artery Doppler meta-analysis (Siargkas et al., 2025) provides the field's strongest current quantitative evidence, benefiting from pooling across multiple studies and a substantial combined sample size, and its finding of both lower resistance (PI) and higher flow velocity (PSV) in women who achieve clinical pregnancy is physiologically coherent, both indices reflecting more favourable uterine perfusion. However, even this meta-analysis addressed clinical pregnancy rather than live birth, and substantial heterogeneity in measurement timing (menstrual cycle vs. stimulation cycle vs. day of embryo transfer) was noted across the pooled studies, mirroring the timing-related heterogeneity seen in the 3D power Doppler literature.

The field's more recent shift toward composite scoring systems (Ouyang et al., 2024) and non-linear, optimal-range modelling (Zhao et al., 2026) represents a methodologically important development, implicitly acknowledging that no single 3D or Doppler index, measured at a single time point and interpreted via a single cutoff, is likely to capture the full biological complexity of endometrial receptivity. This mirrors a broader pattern also seen in the endometrial-thickness literature, where population-level associations coexist with poor individual-patient discriminatory performance for any single measure; combining multiple structural and vascular parameters, and modelling them as continuous or non-linear rather than simple threshold variables, appears a more promising direction than continued searches for a single superior index.

Several methodological limitations affect the strength of these conclusions. Sample sizes across the primary studies are modest (54 to 338 participants), limiting statistical power, particularly for subgroup analyses such as the embryo-quality stratification in Mercé et al. (2008). 3D power Doppler measurement is operator- and equipment-dependent, and measurement reproducibility across centres and machines, a concern explicitly raised in published correspondence regarding some of the foundational studies, remains incompletely characterised. Outcome definitions vary considerably (biochemical pregnancy, clinical pregnancy, live birth, live birth versus miscarriage), limiting direct comparability across studies and, in several cases, preventing a clear answer to the live-birth-specific question that is of greatest clinical relevance. Finally, most primary studies are single-center and were not designed or powered with live birth as a prespecified primary outcome, and the meta-analytic evidence currently available addresses clinical pregnancy rather than live birth specifically.

Priorities for future research include: (a) larger, multi-center studies with live birth as the prespecified primary outcome, adequately powered to test 3D power Doppler and uterine artery Doppler parameters specifically, rather than clinical pregnancy as a surrogate; (b) studies using repeated or dynamic measurement across the peri-implantation window, rather than a single time point, to directly test whether this explains the discrepant findings between fresh- and frozen-cycle cohorts; (c) further development and external validation of composite, multiparametric receptivity scoring systems and non-linear modelling approaches, building on the recent work of Ouyang et al. (2024) and Zhao et al. (2026); and (d) formal assessment of inter-operator and inter-equipment reproducibility for 3D power Doppler indices, to establish whether these measures can be standardized sufficiently for routine multi-center clinical use.

Three-dimensional power Doppler vascularity indices and uterine artery Doppler parameters carry real, biologically plausible information about endometrial receptivity beyond simple endometrial thickness, but this information is inconsistent across studies and strongly context-dependent, varying with measurement timing, cycle type, embryo quality, and the specific outcome examined. The clearest and most consistent signal to date concerns the distinction between live birth and miscarriage among already-pregnant patients, and the association between lower uterine artery resistance and clinical pregnancy at the meta-analytic level, while single-time-point 3D power Doppler measurement in frozen embryo transfer cycles has shown no discriminatory value in at least one well-designed cohort. The field's recent movement toward composite, multiparametric scoring systems and non-linear modelling of individual indices likely represents a more promising direction than continued reliance on any single 3D or Doppler cutoff, but current evidence remains insufficiently standardized and validated to support routine clinical adoption of a specific parameter or algorithm in place of, or in addition to, endometrial thickness.

This manuscript was prepared as a structured draft to support submission and is intended for the author's review before journal submission. Before submission: (1) independently verify every citation against its primary source, and complete the full author list for Ouyang et al. (2024), abbreviated here as "et al."; (2) search for any more recent studies or meta-analyses published after this draft was prepared, given the continued methodological evolution in this area (e.g., composite scoring systems, artificial-intelligence-based models); (3) consider whether a full PRISMA systematic review (with protocol registration, dual independent screening, and a PRISMA flow diagram) is required by the target journal, as this narrative review does not claim that level of methodological rigor; and (4) adapt formatting, length, and citation style to the target journal's author guidelines.