Quantitative assessment and clinical value of right-to-left shunt in patent foramen ovale using three-dimensional transesophageal echocardiography combined with right-heart contrast: a prospective study

Introduction

Patent foramen ovale (PFO) has a prevalence of about 25% in the general population, and its detection rate in patients with cryptogenic ischemic stroke without an identifiable arterial source and in those with migraine with aura can be as high as 40–50% (1). Right-to-left shunt (RLS) across the interatrial septum provides a pathway for venous thrombi to enter the systemic circulation directly, and is regarded as a potential mechanism of cerebral embolism as well as a key element of the “gas-neuron” hypothesis in migraine pathophysiology (2). Two-dimensional (2D) transthoracic bubble study and 2D transesophageal contrast imaging are commonly used clinical screening methods for PFO, but they provide only semi-quantitative “grade” information; the results are easily affected by operator experience, respiratory motion, and imaging-plane selection, leading to limited consistency (3,4). Although randomized controlled trials have confirmed that PFO closure can reduce stroke recurrence and improve symptoms in some patients with migraine, there is still no unified quantitative standard for precisely determining “significant shunt” in clinical pathways. Real-time three-dimensional transesophageal echocardiography (3D-TEE), integrating matrix probes and full-volume acquisition technology, can present the three-dimensional (3D) structure of the interatrial septum within a single cardiac cycle and record the instantaneous peak flow after the Valsalva maneuver (5,6). When synchronized with right-heart contrast echocardiography, voxel integration of microbubble distribution in 3D space directly yields the true volume of the transforamen shunt rather than indirect bubble counting (6). Theoretically, this method can overcome the drawback of 2D planes missing high-velocity shunts and provide a unified coordination for inter-center comparison (7). However, the lack of prospective studies verifying the accuracy and reproducibility of imaging-volume thresholds and their correlation with clinical outcomes has become a key bottleneck limiting the promotion of 3D technology (8). This study was designed as a prospective single-center cohort; 100 consecutive patients with suspected PFO-related stroke or migraine with aura were enrolled along with healthy controls, and training and validation cohorts were established. The study quantified interatrial shunt volume by 3D-TEE combined with right-heart contrast echocardiography and constructed a “hard reference standard” using two traditional imaging modalities to systematically evaluate diagnostic accuracy; a 3-year follow-up was conducted to observe the predictive value of the obtained volume threshold for stroke/TIA recurrence and migraine improvement, and its independence was verified using the Firth-Cox model. The primary objective of the study is to quantitatively determine the diagnostic accuracy, reproducibility and prognostic value of PFO shunt volume quantified by 3D-TEE combined with right-heart contrast echocardiography. This study is the first to propose a significant shunt-volume threshold of 22.93 µL, which was successfully extrapolated to the validation cohort, demonstrating that this threshold is operable in diagnosis, risk stratification, and efficacy monitoring, and providing a new quantitative basis for future updates to PFO imaging guidelines (Figure 1). We present this article in accordance with the TRIPOD reporting checklist (available at https://cdt.amegroups.com/article/view/10.21037/cdt-2025-aw-602/rc).

Figure 1 Schematic illustration of the shunt volume measurement across the PFO. ECG, electrocardiogram; IV, intravenous; PFO, patent foramen ovale.

Methods

Study design and ethical framework

This study was conducted as a prospective diagnostic investigation with a predefined training and validation cohort design. The study population will be randomly divided into two independent cohorts: training cohort: used to develop and optimize the quantitative model for RLS volume assessment using 3D-TEE combined with right-heart contrast echocardiography. Validation cohort: used to independently evaluate the diagnostic accuracy, reproducibility, and prognostic performance of the model established in the training phase. The study process comprised the following consecutive stages: first, baseline clinical evaluation and three imaging examinations were completed; subsequently, offline quantitation of 3D contrast data was performed on the hospital imaging server; third, the 3D quantitative results were compared with the multimodality “hard reference standard” to evaluate diagnostic performance; finally, a 36-month follow-up was conducted to test the prognostic value of the quantitative indices for ischemic stroke and transient ischemic attack. The protocol was approved by the ethics committee of Changxing County People’s Hospital (No. 2019-031), and all participants signed written informed consent. The study was conducted in accordance with the Declaration of Helsinki and its later amendments.

Subject recruitment, inclusion and exclusion criteria, and sample size

The recruitment period was from March 2019 to March 2022. Inclusion criteria were: age 17–86 years; at least one episode of cryptogenic ischemic stroke or transient ischemic attack previously confirmed by the neurology department, or moderate-to-severe migraine with aura meeting the International Headache Society ICHD-3 criteria (9); ability, with the aid of a mouthpiece pressure monitor, to sustain a 40 mmHg Valsalva strain for 15 s; RLS indicated by routine transthoracic bubble study. Exclusion criteria were: atrial fibrillation, patent ductus arteriosus (PDA), ventricular septal defect (VSD), systolic pulmonary arterial pressure >70 mmHg, estimated glomerular filtration rate <30 mL·min⁻¹·1.73 m⁻², pregnancy or lactation, allergy to right-heart contrast agents, inability to tolerate transesophageal intubation, or refusal of follow-up. A pilot diagnostic-efficacy test showed that the area under the curve (AUC) of 3D shunt volume for predicting a significant shunt was 0.82. Setting the null hypothesis AUC =0.60, significance level α=0.05, and power 1 − β =0.80, MedCalc 24.0 calculated a minimum requirement of 90 cases; allowing for a 10 % loss to follow-up, the final planned enrollment was 100 cases. Patients with PDA or VSD were excluded to avoid potential confounding related to the presence of additional intracardiac or extracardiac shunts that may alter hemodynamics and shunt detection on contrast echocardiography. Although PDA and VSD are typically associated with left-to-right shunting in adulthood, shunt direction and magnitude may vary under certain physiological conditions, and their presence could interfere with the accurate assessment of RLS specifically attributable to PFO.

A total of 100 patients were consecutively enrolled, and 25 age- and sex-matched healthy volunteers were recruited to establish the normal reference interval. After enrollment was complete, the study database was divided by computer-generated random numbers into a training cohort of 70 cases and a validation cohort of 30 cases at a 7:3 ratio.

Imaging examination procedures

All examinations were completed on a single hospitalization day in the order of two-dimensional transthoracic echocardiography (2D-TTE), two-dimensional transesophageal echocardiography (2D-TEE), and real-time 3D-TEE-right-heart contrast, with each examination team blinded to the results of the others. 2D-TTE was performed with a Philips EPIQ 7C console and an S5-1 probe in the left lateral decubitus position to obtain a parasternal four-chamber view; a 10 s baseline video was recorded before injection of contrast agent. The contrast agent consisted of 8 mL normal saline, 1 mL air, and 1 mL autologous venous blood mixed back and forth 10 times in a three-way stopcock, and 5 mL was rapidly injected through an 18-G indwelling catheter in the right elbow. Injection was synchronized with ultrasound recording via a coaxial pulse trigger. Shunt was quantified by Spencer grade 0–4. 2D-TEE employed a Philips EPIQ 7C and an X7-2t probe; after topical oropharyngeal anesthesia with 2 % lidocaine spray 10 mL and intravenous midazolam 0.03 mg/kg (maximum 3 mg) for sedation, the probe was inserted. Resting and Valsalva sequences were recorded at 0° four-chamber and 60° short-axis views, the same dose of contrast was injected, and results were evaluated by Nyboe grade 0–3. Real-time 3D-TEE used the same probe in Full-Volume mode, depth 12 cm, volume angle 60°×60°, electrocardiography (ECG)-gated acquisition of 4 cardiac cycles, with 10 mL fresh microbubble suspension injected during both rest and Valsalva. Echo Navigator 3.0 software completed real-time fusion at a frame rate of 20–25 volume/s. All data were stored in Digital Imaging and Communications in Medicine (DICOM) 3.0 format.

Image post-processing and quantification of 3D shunt volume

Using 3DQ Advanced 12.1, load the volume data and adjust the three orthogonal planes to ensure full-length display of the interatrial septum. In mid-systole measure the PFO tunnel length (straight-line distance from the right-atrial rim of the valve to the left-atrial rim), the minimum and maximum slit width (diameters of the cross-section at the narrowest closing and widest fully open moments, respectively), the leaflet free-edge excursion (maximum minus minimum displacement value), and determine the peak-to-peak difference of the leaflet-mitral annulus plane angle. The volume data are then exported in RVL format and imported into a MATLAB R2022a script. The script first normalizes voxel intensity to 0–255, calculates in real time the mean grayscale (G_m) of the adjacent interatrial muscular wall, and uses the threshold (T) =1.45×G_m to screen microbubble voxels, marks them with value 1, and integrates within the 3D coordinate grid to obtain the instantaneous shunt volume (V_t). The continuous V_t-t curve is trapezoidally integrated to yield the total shunt volume (V_total; µL), while the peak value (V_peak) and shunt duration (t_dur; s) are recorded. Each case outputs a 3D visualization video and a comma-separated values (CSV) data file. To assess reproducibility, 30 cases were randomly selected for a second analyst to repeat all quantifications in a fully blinded manner; the first analyst re-analyzed the same data 14 days after the initial measurement.

Hard reference standard and evaluation of diagnostic accuracy

2D transoesophageal grading ≥ Nyboe 2 (10) and 2D transthoracic grading ≥ Spencer 3 (11) were defined as indicators of a “significant shunt”. A subject was regarded as “hard reference standard” positive only when both examinations reached their respective significant thresholds; if either examination failed to meet the criterion, the subject was judged negative (12). Using this hard reference standard as the control, we calculated the sensitivity and specificity of the 3D shunt volume and further derived the positive likelihood ratio, negative likelihood ratio, and diagnostic odds ratio (DOR).

Threshold determination, internal validation and extrapolation

In the training set of 70 cases, an ROC curve was plotted with the hard reference standard as the dependent variable and the 3D shunt volume as the independent variable, and the optimal cut-off (V_cut) was obtained according to the Youden index. V_cut was then directly applied to the independent validation set of 30 cases, and the AUC was recalculated; the DeLong test was used to compare the AUCs between the training and validation sets. If the validation-set AUC did not differ statistically from that of the training set, V-cut was confirmed to have a stable diagnostic value.

Follow-up, intervention blinding and clinical outcomes

The 3D-TEE results were managed in a double-blind manner: the imaging workstation automatically hid all 3D quantitative fields, and only the study statistician held the decryption key. The decision on whether to perform percutaneous PFO closure was made by a joint consultation among cardiology, neurology and interventional imaging departments, based on 2D-TTE, 2D-TEE and clinical data; the decision-makers could not view the 3D results, thereby avoiding verification bias. The closure procedure was recorded as a time-varying covariate. All subjects were followed at scheduled outpatient visits at 3, 6, 12, 24 and 36 months, and were contacted by telephone every 3 months. The primary endpoint was new or recurrent ischemic stroke or transient ischemic attack, requiring 1.5 T magnetic resonance imaging (MRI) diffusion-weighted imaging within 24 h of symptom onset, adjudicated independently by two blinded neuroradiologists; if their opinions differed, a third neuroradiologist made the decision. Secondary endpoints included the absolute change from baseline in monthly frequency of migraine with aura, residual shunt volume 6 months after closure, and all-cause death. All follow-up data were entered into the REDCap electronic database and double-checked before database lock.

Statistical analysis

Sample size was calculated a priori based on an expected AUC with 85% power and a two-sided α of 0.05. Normality of continuous variables was assessed with the Kolmogorov-Smirnov test; normally distributed data are expressed as mean ± standard deviation and were compared with the independent-sample t-test or paired t-test; non-normally distributed data are expressed as median (interquartile range) and were compared with the Mann-Whitney U or Wilcoxon signed-rank test. Categorical variables are presented as counts and percentages and were compared with the χ² test or Fisher exact test. Reproducibility was assessed with the two-way mixed absolute-agreement intraclass correlation coefficient (ICC) and Bland-Altman 95% limits of agreement. ROC curves and AUC comparisons used the pROC package DeLong test. Prognostic analysis adopted the Firth-corrected Cox proportional hazards model, treating the closure procedure as a time-varying covariate; candidate covariates were screened by least absolute shrinkage and selection operator (LASSO) regression while maintaining an event-to-variable ratio ≥10. All tests were two-sided, and a significance level of P<0.05 was considered statistically significant. Statistical analyses were performed with SPSS 30.0, R 4.4.0 and STATA 18.

Results

Study population and follow-up

A total of 100 patients were enrolled and randomly assigned to the training and validation groups. There were no significant differences between the two groups in age, sex ratio, body mass index, hypertension, diabetes, smoking proportion, number of previous ischemic stroke/TIA events, MIDAS score for migraine with aura, or antiplatelet therapy proportion (t-test, χ² test or Mann-Whitney U test, all P>0.05). Data from the healthy control group (25 cases) are for reference only and were not included in statistical comparisons (Table 1).

Morphological quantification by 3D-TEE and measurement consistency

Anatomical parameters of PFO and quantitative shunt indices measured by 3D-TEE showed no significant differences between the training and validation cohorts in tunnel length, slit width (minimum, maximum), leaflet free-edge excursion, leaflet-mitral annulus peak-to-peak angle difference, or shunt volume (including total volume, peak volume and duration) (Mann-Whitney U test, all P>0.05) (Table 2). Repeat-measurement consistency analysis of 3D-TEE shunt-volume indices demonstrated good intra-observer (ICC: 0.91–0.94) and inter-observer (ICC: 0.90–0.92) agreement for V_total, peak shunt volume and shunt duration (Table 3).

Diagnostic accuracy and threshold validation

Analysis of the diagnostic performance of the imaging methods for identifying significant shunt showed that the quantitative index from 3D-TEE had high sensitivity (92.31%), specificity (89%) and DOR (89.78), clearly superior to 2D-TTE and 2D-TEE (Table 4). Using the optimal cut-off (22.93 µL) obtained from the training set for external validation, ROC analysis of 3D shunt volume yielded an AUC of 0.92 in the training set and 0.89 in the validation set; the difference between the two groups was not statistically significant (DeLong test, P=0.27, Figure 2).

Figure 2 Combined ROC curves for training and validation cohorts. AUC, area under the curve; CI, confidence interval; ROC, receiver operating characteristic; V_cut, the optimal cut-off.

Prognostic value of 3D shunt volume for 3-year stroke/TIA

Kaplan-Meier analysis showed a significant difference in ischemic stroke/TIA survival curves between the high-shunt group (≥ V_cut, n=48) and the low-shunt group (< V_cut, n=52) after stratification by the 3D shunt-volume threshold (total events =12, of which 10 occurred in the high-shunt group and 2 in the low-shunt group; log-rank test P=0.01). At 36 months of follow-up, event-free survival rates were 79% and 96%, respectively, indicating clear prognostic stratification by this threshold (Figure 3). Because the number of events was limited (total =12), a Firth-penalized Cox model (EPV ≈3) including only four key covariates was used to ensure model robustness. The results showed that 3D shunt volume ≥ threshold [hazard ratio (HR) =5.27, P=0.04] and age (HR =1.04, P=0.049) independently predicted 3-year stroke/TIA risk, whereas hypertension and PFO closure were not statistically significant (both P>0.05, Table 5).

Figure 3 Kaplan-Meier curves for ischemic stroke/TIA-free survival in groups stratified by 3D shunt volume (≥ V_cut vs. < V_cut). 3D, three-dimensional; TIA, transient ischemic attack; V_cut, the optimal cut-off.

Secondary clinical outcomes

The monthly attack frequency of migraine with aura decreased significantly during follow-up (P=0.001, Figure 4A). Residual RLS volume 6 months after closure was significantly lower than in the non-closure group (P<0.001, Figure 4B), indicating concordance between imaging improvement and migraine symptom relief.

Figure 4 Changes in monthly migraine with aura frequency and residual right-to-left shunt volume 6 months after closure. (A) Monthly migraine attacks, analysed by repeated-measures one-way ANOVA; (B) residual right-to-left shunt volume, analysed by Mann-Whitney test. ANOVA, analysis of variance.

Discussion

3D-TEE combined with right-heart contrast echocardiography shows overall performance superior to traditional 2D imaging in the identification of significant shunts; this advantage stems from two levels of technological innovation. The 3D full-volume mode captures the complete stereoscopic structure of the interatrial septum within a single cardiac cycle and, through real-time inter-frame coupling with microbubble perfusion, converts trans-foramen bubble trajectories into quantifiable voxels that are integrated in real time, thereby overcoming plane deviation and repeat-counting errors caused by probe rotation and respiratory motion in 2D planes (13,14). The volumetric shunt threshold of 22.93 µL maintains stable discriminative power across different populations, with sensitivity, specificity and DOR all reaching high levels, indicating that the threshold is advantageous both in excluding weak positives and in capturing truly high-load shunts. The ECG-gated strategy fully presents the transient peak flow after the Valsalva maneuver, avoiding 2D misses caused by “rapid in-and-out” microbubbles (15). Such peak-flow compensation is particularly important clinically because stroke-related PFOs mostly present as instantaneous high-volume shunts rather than continuous small shunts. The grayscale threshold is automatically normalized to the adjacent myocardial wall, eliminating the influence of different gain settings on voxel integration and further improving inter-equipment and inter-operator consistency (16). In comparison with existing literature, where 2D-TEE shows only about 75% sensitivity for significant shunts (17), the 3D threshold validated in this study still maintained an AUC of 0.89 in the independent validation cohort, demonstrating good generalizability. The clinical decision chain is therefore simplified: when the volumetric shunt exceeds this threshold, diagnostic confidence is already sufficient to proceed to therapeutic discussion without repeated multimodal confirmation. The new threshold also fills the gap of lacking quantitative reference for 3D techniques and is expected to become a candidate indicator in the quantitative section of future guidelines for shunt assessment.

The 3D-TEE quantitative indices showed extremely high ICCs and low coefficients of variation in both same- and different-day repeat measurements, indicating that algorithmic and operator factors contributed little to measurement error. The matrix probe provides isotropic voxel resolution, so the automatic search for the narrowest and widest slit cross-sections is highly repeatable and no longer dependent on the operator’s fine-tuning experience with 2D planes. The grayscale threshold is set as a proportion of the average grayscale of the surrounding myocardium, thereby minimizing the influence of system gain, depth and contrast-agent mixing uniformity on volume integration and fundamentally ensuring consistency across different scanning batches (18). The angle difference between the interatrial septal leaflet and the mitral annulus was found to be positively correlated with volumetric shunt, meaning that greater leaflet compliance and larger opening-angle changes allow a larger volume of microbubbles to enter the left atrium during the same Valsalva maneuver. This structure-function coupling phenomenon is often overlooked in 2D studies because of static section measurements, whereas 3D dynamic tracking reveals the true clinical significance of leaflet compliance (19). Previous 2D slit-length measurements reported ICCs often below 0.8; the 3D ICCs >0.9 in this study indicate that the new method markedly improves research reproducibility and also suggest that adopting a unified volume threshold in future multicenter cohorts is more feasible. Structural indicators showed no differences between the training and validation populations, further confirming that random allocation balanced the distribution of interatrial-septum anatomy in this cohort and laid a foundation for subsequent efficacy comparisons and prognostic analyses.

The volumetric shunt threshold is effective not only at the diagnostic level but also in demonstrating marked stratification capability for long-term outcomes. Three-year follow-up showed that patients whose shunt volume reached or exceeded 22.93 µL had a clearly higher incidence of stroke or TIA, with survival curves separating early and remaining divergent until the end of follow-up, indicating that high volumetric shunt is not a transient driver of paradoxical embolism (20). The Firth-Cox model further excluded the confounding effects of age, hypertension and percutaneous closure, confirming that the volume threshold itself possesses independent prognostic value. Unlike previous qualitative gradings such as “more than 20 bubbles” or the “shower sign”, this finding provides an operable quantitative cutoff for interventional decision-making, enabling physicians to build a more direct pre-procedural dialogue between shunt magnitude and potential benefit (21). Meanwhile, the reduction in shunt volume paralleled the decrease in migraine-with-aura frequency, suggesting that the same pathological pathway may link thromboembolism and headache attacks: when interatrial bubbles decline, blood-brain barrier inflammation and trigeminovascular activation are simultaneously attenuated (22). Residual shunt volume 6 months after closure was markedly lower than in non-intervened patients, indicating that the 3D-guided closure strategy achieved success in both anatomical apposition and functional sealing, and also providing a monitorable surrogate endpoint for borderline cases with high shunt volume but relatively low stroke risk (23). Consistency of the threshold across both stroke and migraine outcomes helps redefine “functionally significant shunt” and is expected to allow future trials to incorporate both hard vascular events and soft symptomatic benefits, thereby comprehensively evaluating the gains from closure.

Limitations

The study was conducted at a single center; the case source was relatively homogeneous, making it difficult to cover differences in multiethnic populations, multiple devices and different contrast-agent formulations. Although random allocation and healthy volunteer controls were used in advance, the results still need validation by external cohorts. The total number of events was only twelve; Firth-Cox calculation can converge with an event per variable of about three, but the confidence intervals remain wide, so the effects of hypertension or closure intervention may have been underestimated in the small sample, and their true impact requires larger-scale long-term follow-up to clarify. The right-heart contrast agent consisted of autologous blood-air-saline mixture; different laboratories may use commercial microbubble agents whose echo intensity and decay rate differ, so the algorithm threshold may need adjustment. Shunt-volume calculation was based on grayscale threshold multiplied by a fixed coefficient, and cross-platform standardization for different machine brands has not yet been performed. Migraine assessment relied only on self-reported monthly attack frequency, lacking headache diaries and quality-of-life scales, and other preventive migraine drugs were not excluded during follow-up, which may introduce confounding into the symptom-shunt relationship. The final threshold was derived from a standard Valsalva of 40 mmHg monitored by mouthpiece and cannot be extrapolated to elderly or critically ill patients who cannot perform the manoeuvre. Future studies need to adopt unified contrast agents and threshold algorithms in a multicentre setting, accumulate hard-endpoint events, and introduce objective symptom scales and advanced imaging to verify the pathological mechanism.

Conclusions

3D-TEE combined with right-heart contrast echocardiography can objectively quantify the RLS volume of PFO; 22.93 µL is a significant shunt threshold confirmed by the training-validation dual cohorts. This threshold is diagnostically superior to 2D-TTE and 2D-TEE; the consistency of 3D measurements is extremely high. Patients whose volumetric shunt reaches or exceeds the threshold have a significantly increased 3-year stroke/TIA risk, and extrapolation of the threshold to migraine outcomes is equally effective. Closure markedly reduces residual shunt volume, accompanied by further relief of migraine. 3D volumetric shunt indices can serve as reliable imaging biomarkers for ischaemic-event risk stratification, closure decision-making and efficacy follow-up.

Acknowledgments

We would like to express our gratitude to all the consortium studies for making the summary association statistical data publicly available. We also thank the reviewers for their insightful comments and their contributions to this manuscript.

Footnote

Reporting Checklist: The authors have completed the TRIPOD reporting checklist. Available at https://cdt.amegroups.com/article/view/10.21037/cdt-2025-aw-602/rc

Data Sharing Statement: Available at https://cdt.amegroups.com/article/view/10.21037/cdt-2025-aw-602/dss

Peer Review File: Available at https://cdt.amegroups.com/article/view/10.21037/cdt-2025-aw-602/prf

Funding: None.

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://cdt.amegroups.com/article/view/10.21037/cdt-2025-aw-602/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The protocol was approved by the ethics committee of Changxing County People’s Hospital (No. 2019-031), and all participants signed written informed consent. The study was conducted in accordance with the Declaration of Helsinki and its later amendments.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Van der Giessen H, Wilson LC, Coffey S, et al. Review: Detection of patent foramen ovale using transcranial Doppler or standard echocardiography. Australasian Journal of Ultrasound in Medicine 2020;23:210-9. [Crossref] [PubMed]
2. Cao W, Shen Y, Zhong J, et al. The Patent Foramen Ovale and Migraine: Associated Mechanisms and Perspectives from MRI Evidence. Brain Sci 2022;12:941. [Crossref] [PubMed]
3. Rubin MN, Shah R, Devlin T, et al. Robot-Assisted Transcranial Doppler Versus Transthoracic Echocardiography for Right to Left Shunt Detection. Stroke 2023;54:2842-50. [Crossref] [PubMed]
4. Larsen BC, Harmsen MC, Vase H, et al. Comparison of Patent Foramen Ovale Sizing by Transesophageal Echocardiography and Balloon Sizing in Patients Undergoing Percutaneous Closure. Struct Heart 2025;9:100375. [Crossref] [PubMed]
5. El Doklah EH, El Shedoudy SA, El Saied AM, et al. Assessment of Value of Three Dimensional Transesophageal Echocardiography versus Conventional Two Dimensional Transesophageal Echocardiography in Guiding Transcatheter Closure of Atrial Septal Defects and Patent Foramen Ovale. J Saudi Heart Assoc 2022;34:85-99. [Crossref] [PubMed]
6. Gwak SY, Kim K, Lee HJ, et al. Three-dimensional agitated saline contrast transesophageal echocardiography for the diagnosis of patent foramen ovale. Sci Rep 2025;15:29136. [Crossref] [PubMed]
7. Wang L, Sun H, Shen H. Anatomical Significance of the Patent Foramen Ovale by Real-Time 3D TEE in Cryptogenic Stroke and Migraine. Echocardiography 2024;41:e70018. [Crossref] [PubMed]
8. He Y, Pu Z. Hotspots and frontiers in patent foramen ovale research: a bibliometric and visualization analysis from 2003 to 2023. Front Cardiovasc Med 2025;12:1483873. [Crossref] [PubMed]
9. Headache Classification Committee of the International Headache Society (IHS). The International Classification of Headache Disorders (3rd ed.). Cephalalgia 2018;38:1-211. Available online: https://ihs-headache.org/en/resources/ichd/
10. Zhang H, Huang W, Lan T, et al. The Value of Contrast-Enhanced Transesophageal Echocardiography in the Detection of Cardiac Right-to-Left Shunt Related with Cryptogenic Stroke and Migraine. Biomed Res Int 2020;2020:8845652. [Crossref] [PubMed]
11. Lim ICZY, Teo YH, Fang JT, et al. Association of Shunt Size and Long-Term Clinical Outcomes in Patients with Cryptogenic Ischemic Stroke and Patent Foramen Ovale on Medical Management. J Clin Med 2023;12:941. [Crossref] [PubMed]
12. Xu J, Cui F, Zhang H. Multimodal ultrasound imaging for patent foramen ovale and pulmonary arteriovenous malformation in patients with cryptogenic stroke or migraine: A prospective diagnostic study. Medicine (Baltimore) 2024;103:e39791. [Crossref] [PubMed]
13. Datta T, Ruggiero N, Peters A, et al. Three-dimensional Transesophageal Echocardiography for Transcatheter Patent Foramen Ovale Closure: Standardizing Anatomic Nomenclature and Novel Sizing Concepts. CASE (Phila) 2023;7:14-20. [Crossref] [PubMed]
14. Hahn RT, Saric M, Faletra FF, et al. Recommended Standards for the Performance of Transesophageal Echocardiographic Screening for Structural Heart Intervention: From the American Society of Echocardiography. J Am Soc Echocardiogr 2022;35:1-76. [Crossref] [PubMed]
15. Chen A, Zhu J, Zhu L, et al. Neglected intrapulmonary arteriovenous anastomoses: A comparative study of pulmonary right-to-left shunts in patients with patent foramen ovale. Front Cardiovasc Med 2023;10:1111818. [Crossref] [PubMed]
16. Pino PG, Madeo A, Lucà F, et al. Clinical Utility of Three-Dimensional Echocardiography in the Evaluation of Mitral Valve Disease: Tips and Tricks. J Clin Med 2023;12:2522. [Crossref] [PubMed]
17. Zhang D, Jiang L, Chen YN, et al. The diagnostic value of contrast-enhanced transcranial Doppler and contrast-enhanced transthoracic echocardiography for right to left shunt in patent foramen ovale: a systematic review and meta-analysis. Front Neurol 2024;15:1447964. [Crossref] [PubMed]
18. Steffel CN, Brown R, Korcarz CE, et al. Influence of Ultrasound System and Gain on Grayscale Median Values. J Ultrasound Med 2019;38:307-19. [Crossref] [PubMed]
19. Turc G, Lee JY, Brochet E, et al. Atrial Septal Aneurysm, Shunt Size, and Recurrent Stroke Risk in Patients With Patent Foramen Ovale. J Am Coll Cardiol 2020;75:2312-20. [Crossref] [PubMed]
20. Mas JL, Saver JL, Kasner SE, et al. Association of Atrial Septal Aneurysm and Shunt Size With Stroke Recurrence and Benefit From Patent Foramen Ovale Closure. JAMA Neurol 2022;79:1175-9. [Crossref] [PubMed]
21. Sun PY, Tobis JM, Daneshvar SA, et al. Quantification of patent foramen ovale shunt severity by transesophageal echocardiogram and transcranial doppler in routine clinical practice. J Stroke Cerebrovasc Dis 2025;34:108306. [Crossref] [PubMed]
22. Lang Y, Zhang S, Xie P, et al. Short-term efficacy of right-to-left shunt closure in patients with vestibular migraine. Front Neurol 2024;15:1500918. [Crossref] [PubMed]
23. Witte LS, El Bouziani A, Beijk MAM, et al. Persistence of residual shunt at 6 and 12 months after transoesophageal echocardiography-guided percutaneous closure of a patent foramen ovale for cryptogenic stroke. Heart 2024;110:1172-9. [Crossref] [PubMed]