Virtual Reality in Neonatal Ductal Stenting: A New Era in Preprocedural Planning for Ductus Arteriosus Stenting in Complex Congenital Heart Defects with Duct-Dependent Pulmonary Circulation

Abstract

Ductus arteriosus (DA) stenting is a well-established interventional alternative to surgical systemic-to-pulmonary shunts in neonates with congenital heart defects and ductal-dependent pulmonary circulation (DDPC). While it avoids sternotomy and cardiopulmonary bypass, the procedure remains technically demanding, especially in cases involving tortuous or misaligned ductal anatomy. Such anatomical complexity is associated with increased procedural difficulty and a higher rate of reinterventions. Although echocardiography, computed tomography (CT), and angiography remain key imaging modalities, they may be limited in providing comprehensive spatial context. Virtual reality (VR) has recently emerged as a promising tool that enables interactive 3D visualisation of cardiovascular structures and may enhance preprocedural assessment. This study aimed to evaluate the feasibility and potential benefits of CT-derived VR-based 3D modelling for preprocedural planning in neonates undergoing DA stenting for DDPC. The study compared VR-derived anatomical and procedural assessments with conventional angiographic imaging. Between March 2021 and August 2023, six neonates with DDPC underwent preprocedural contrast-enhanced CT and were included in this single-centre retrospective study. Three additional patients referred for DA stenting without CT were excluded. Dedicated VR software (VMersive, Poland) was used to automatically generate interactive 3D models from CT datasets without manual segmentation. The models were evaluated using immersive VR headsets and controller-based tools for precise anatomical assessment. DA morphology, tortuosity classification, total ductal length [L1], straight-line distance [L2], curvature index (CI), optimal fluoroscopic projection, and vascular access were analysed in VR and compared with angiographic data. Paired t-tests were used for statistical comparison, with significance set at p < 0.05. All six VR models were successfully created in under 5 min. Morphological and tortuosity classification: VR-based assessments were consistent with angiographic classification (Type I–III), confirming reliability. Length and curvature: VR-derived [L1] values were significantly longer than angiographic measurements (mean: 28.93 mm vs. 24.14 mm; p = 0.038). Curvature index was significantly higher in VR (mean: 0.46 vs. 0.29; p < 0.001), especially in Type III ducts, suggesting that angiography underestimates vascular tortuosity. Projection planning: Median angular deviation between VR-recommended and final procedural projections was 12° (range: 6–30°, p = 0.259). In cases requiring multiple contrast injections, final projections closely matched VR suggestions, indicating VR’s potential to reduce radiation exposure and procedural time. Vascular access: VR-predicted access matched initial operator choice in 3/6 cases (50%), rising to 5/6 (83%) when reinterventions were included. Adjacent structures: VR allowed simultaneous visualisation of DA and adjacent thoracic structures. In patients with anomalous pulmonary venous return, VR clarified spatial relationships and potential compression risk. Additionally, in one case, a persistent vertical vein was identified, guiding a rare, combined intervention strategy. VR-based 3D modelling from CT data are a feasible and effective tool for preprocedural planning in DA stenting. It provides accurate spatial insight into ductal anatomy, curvature, projection alignment, and access routes. VR enhances anatomical understanding, supports multidisciplinary planning, and may reduce procedural uncertainty, particularly in complex neonatal interventions.

Supplementary Information

The online version contains supplementary material available at 10.1007/s00246-025-03908-3.

Introduction

Interventional ductus arteriosus (DA) stenting is a valuable alternative to surgical systemic-to-pulmonary shunts in children with congenital heart defects (CHD) and ductal-dependent pulmonary circulation (DDPC) [1]. Increasing reports, demonstrate the advantages of this approach including avoidance of sternotomy and cardiopulmonary bypass, shorter intensive care unit stays, and reduced overall mortality and morbidity [2–4].

However, DA stenting for DDPC remains technically challenging, particularly in cases involving long, tortuous ducts or misaligned pulmonary artery (PA) branches. These anatomical complexities contribute to a failure rate of 16–19%, with a significant proportion of patients requiring reinterventions [5, 6].

Since the first detailed report of successful DA stenting, several attempts have been made to classify ductus arteriosus anatomy and morphology, which are crucial for preprocedural planning and optimizing outcomes. These classifications consider the number of bends observed in angiography, define ductal tortuosity using the tortuosity index (Ti) and incorporate curvature calculations such as the curvature index (CI). Additional anatomical factors, including DA origin, site of connection to the PA, and the presence of PA branch misalignment, are also taken into account [7]. Several studies have proposed practical procedural strategies to facilitate stent implantation, further underscoring the complexity of this intervention and the importance of comprehensive preprocedural imaging [1, 2, 8].

Detailed echocardiography remains the primary imaging modality, supplemented by computed tomography (CT) particularly in complex anatomies [9, 10]. Angiography remains essential for procedural guidance, while three-dimensional (3D) rotational angiography adds diagnostic value and may reduce radiation exposure and contrast doses [11]. Additionally, computer-based 3D modelling improves spatial comprehension, although it requires manual segmentation. While 3D printing supports preprocedural planning, it is associated with high costs and time-consumption [12].

Virtual reality (VR) tools have emerged as powerful alternatives for enhancing imaging, assessing spatial relationships, and planning procedures through virtual simulations [13, 14]. Dedicated VR software enables automatic 3D model generation from volumetric data, allowing for rapid analysis of cardiac defects with simultaneous evaluation of depth and surrounding structures, which is particularly valuable in complex cardiovascular anatomies.

At our institution, VR technology has been implemented as an adjunct imaging tool for the evaluation of complex CHD. This approach has notably improved imaging quality and supported clinical decision-making not only in larger patients [15–18], but also in neonates and infants with extremely complex anatomies [19–21]. Encouraged by these results, we conducted a retrospective analysis to evaluate the feasibility and clinical utility of CT-derived VR models in neonates undergoing DA stenting for DDPC.

Aim

This retrospective study evaluates the feasibility and potential benefits of CT-derived 3D VR- models in assessing DA morphology in patients with DDPC. It also compares 3D-VR modelling with standard angiographic imaging to highlight insights into the practical applications and advantages of VR in procedural planning DA stenting in DDPC cases.

Material and Methods

Patient Selection

This retrospective, single-centre study reviewed all cardiac catheterisation procedures performed at the University Children’s Hospital of Cracow between March 2021 and August 2023. Included were patients with CHD and DDPC referred for DA stenting who underwent preprocedural contrast-enhanced CT for DA morphology assessment. Patients without CT imaging were excluded.

Imaging Data

Computed Tomography and Image Processing

Contrast-enhanced, non-ECG-gated CT scans were performed based on clinical indications, following standard institutional protocol. Imaging data were stored in Digital Imaging and Communications in Medicine (DICOM) format. Informed parental consent was obtained for all imaging studies.

Virtual Reality Model Creation

Computed tomography datasets in DICOM format were imported into a dedicated VR software platform (VMersive; Poland), designed for immersive cardiovascular visualisation. The software enables automatic 3D-VR model generation from volumetric data without manual segmentation and offers a comprehensive set of tools for anatomical analysis and procedural planning. Core functionalities include selective rendering for isolating specific structures, layer filtering for stepwise visualisation, and tools for real-time model manipulation. The software supports precise linear and angular measurements, fluoroscopic angle simulation and virtual procedural simulation. Interactions occur via VR hand controllers, allowing intuitive spatial navigation within an immersive environment. Models are viewed through a head-mounted display to preserve depth perception and simultaneously projected onto an external computer screen for collaborative review and video documentation.

Data Analysis and Measurements: Angiography and VR

A comprehensive analysis of cardiac angiograms and essential angiographic measurements was conducted by an experienced interventional paediatric cardiologist. VR models were independently evaluated by a paediatric cardiologist and an interventional cardiology fellow. Emphasis was put on comprehensive evaluation of the DA and surrounding anatomical structures, including thoracic organs, major vessels, airways, and lung tissue. All findings were subsequently reviewed by a senior expert in paediatric and interventional cardiology. Importantly, the angiographic assessment was completed prior to exposure the VR models, to ensure an unbiased comparison between the two modalities.

1. Morphological assessment and tortuosity index

   Virtual models were analysed using dedicated software tools to optimize DA evaluation, including the anatomical origin and insertion site into the PA. Ductal tortuosity was classified according to established criteria: Type I (TI)—straight or minimally tortuous; Type II (TII)—moderately tortuous, with one turn; and type III (TIII)—highly tortuous with multiple turns [7]. VR-derived morphological assessment was compared with angiographic findings using the same classification parameters: origin, insertion site and degree of tortuosity.

2. Length measurements and curvature index

   Virtual reality models were used to measure total DA length [L1], defined as the center-line path within the vessel lumen. Subsequently, the straight-line length [L2], measured as the direct distance between the aortic and pulmonary ends of the DA. Both measurements obtained from VR models were compared to corresponding angiographic values. The Curvature index (CI) was calculated for both modalities using the formula: CI = (L1–L2)/L1.

3. Optimal projection for stent implantation

   VR-enabled spatial analysis allowed for manipulation of models in various planes to determine the most optimal fluoroscopic projection for guiding DA stenting. These VR-recommended projections were retrospectively compared with the final angiographic projections used during cardiac catheterization for ductal stenting.

   Additionally, the number of diagnostic angiographies performed before selecting the most optimal, final projection for stent implantation was also recorded to assess the potential procedural benefit of VR-based planning.

4. Vascular access

   The most optimal vascular access for cardiac intervention was determined using the virtual goniometer tool integrated into the VR software. A centerline path was traced from each evaluated access site through the following vessel to the DA lumen, allowing for an accurate assessment of the optimal catheter insertion angle. VR-based vascular access recommendations were retrospectively compared with those used during initial cardiac catheterisation procedure, and reintervention.

5. Virtual evaluation of adjacent structures

   The VR system was assessed for its ability to simultaneously visualise the DA and adjacent thoracic structures, including the trachea, bronchi, vascular structures, and their spatial relationships. The aim was to determine whether enhanced spatial context provide additional insight for intervention planning.

Statistical Methodology

Continuous variables were summarised using descriptive statistics (mean with standard deviation and median with range). Percentage differences were calculated to assess variability between VR and angiographic measurements. Paired t-tests (Student’s t-test for dependent samples) were used to compare values obtained from the same patients using both imaging modalities. Statistical significance was defined as p < 0.05.

Results

Patient Selection and Procedural Outcomes

During the study period, nine neonates were referred for ductal stenting. Preprocedural contrast-enhanced CT scans were performed in six of these cases, forming the basis of this retrospective analysis. Three patients, qualified solely on echocardiography, were excluded due to absence of CT imaging.

At the time of the procedure, patient ages ranged from 9 to 19 days (median: 13.5 days), and weight from 2.6 to 3.7 kg, (median: 3.35 kg). Underlying diagnoses included pulmonary atresia with a ventricular septal defect (n = 2), heterotaxy syndrome (n = 2), pulmonary atresia with intact ventricular septum (n = 1), and double outlet right ventricle with critical pulmonary valve stenosis (n = 1). Ductal stenting was successfully performed in all six cases. However, three required re-intervention, resulting in a 50% re-intervention rate (Table 1).

Table 1.

Patient characteristics and imaging-derived parameters assessed for procedural optimization in ductal stenting

Parameter	Patient 1	Patient 2	Patient 3	Patient 4	Patient 5	Patient 6
Diagnosis	TS, PA, RV-hypopl VSD	PA, DORV	Htx, DORV, PA, RV-hypopl, PAPVR	Htx, DORV, PA, RV-hypopl, TAPVR	PA, IVS, RV-hypopl	TS, PA, RV-hypopl VSD
Age-CT [day]	11	13	5	6	6	14
Age-cath [day]	17	19	9	10	11	16
Weight [kg]*	2.6	3.7	3.4	3.3	3.7	2.8
Lenght [cm]*	50	53	55	54	55	44
Aortic Arch	RAA	RAA	LAA	LAA	LAA	LAA
DA origin**	L-InnA	L-InnA	BCT	Ao Arch	Ao Desc	Ao Desc
DA insertion**	LPA	PA-Bif	PA-Bif	RPA	RPA	LPA
Tortuosity index (Ti): based on the number of DA bends
Ti-VR	TIII	TII	TII	TII	TIII	TII
Ti-cath	TIII	TII	TII	TII	TIII	TII
Measurements: overall ductal length [L1] and straight line [L2] between DA origin and insertion
L1-VR [mm]	38.3	22.8	35.7	25.6	35.5	15.7
L1-cath [mm]	27.1	20.0	31.4	25.0	27.0	14.3
L2-VR [mm]	24.9	12.2	11.5	10.1	23.4	10.1
L2-cath [mm]	22.5	13.0	16.3	16.0	22.9	11.0
Curvature index (CI): calculated based on the formula: CI = (L1–L2)/L1
CI-VR	0.35	0.46	0.68	0.60	0.34	0.35
CI-cath	0.17	0.35	0.48	0.36	0.15	0.23
Angiographic projections
VR-proposed	LAO 75	LAO 30	RAO 15	RAO 35	RAO 65/CRA 30	LAO 60/CRA 10
Used-cath	P-A, LL	LAO 60	RAO 27	RAO 29	LAO 90	LAO 54/CRA 21
Projection deviation [°]	15°	30°	12°	6°	155°	6° (LAO), 12° (CRA)
Total angio***	2	3	2	4	2	4
Vascular access: based on measurements of the catheter insertion angle
VR-proposed	CaA	FA or FV	FA	AxilA	FA	AxilA
Used-cath	Hybrid	FV	FA	FA	FA	FA
Used-re-cath	FA	N/A	N/A	AxilA	FA	N/A

Ao Desc descending aorta, AxilA axillary artery, BCT brachiocephalic trunk, CaA carotid artery, Cath cardiac catheterization, CRA cranial, CT computed tomography, DA ductus arteriosus, DORV double outlet right ventricle, FA femoral artery, FV femoral vein, Htx heterotaxy syndrome, Hybrid hybrid procedure, IVS intact ventricular septum, L-InnA left innominate artery, LAA left aortic arch, LAO left anterior oblique, LL left lateral, LPA left pulmonary artery, N/A not applicable (due to absence of reintervention), PA pulmonary atresia, P-A postero-anterior, PA-Bif pulmonary artery bifurcation, PAPVR partial anomalous pulmonary venous return, RAA right aortic arch, RAO right anterior oblique, Re-cath cardiac re-intervention, RPA right pulmonary artery, RV-hypopl right ventricular hypoplasia, TAPVR total anomalous pulmonary venous return, TII moderately tortuous ductus arteriosus, TIII highly tortuous ductus arteriosus, VSD ventricular septal defect, VR virtual reality

*Weight and length refer to the time of cardiac catheterization

**The origin and insertion of the ductus arteriosus were consistent between angiographic and VR-based imaging

***Total Angio—total number of angiographies to define optimal working projection

Imaging Data

Angiographic Analysis and VR Model Assessment

Angiographic images were first analysed to evaluate detailed DA morphology. Separately, all CT DICOM datasets were successfully processed in a dedicated VR application, and fully immersive 3D VR models were generated within approximately 5 min. This enabled accurate and interactive anatomical analysis without the need for manual segmentation. However, additional software tools were used to apply tissue-specific filters, enhancing visualization of soft tissues, vessel walls, blood-filled structures, bone, and fat tissue (Fig. 1).

1. Morphological assessment and tortuosity index

   In all cases, VR models accurately depicted the DA origin, insertion site, the presence of PA branches and the tortuosity. Four patients had a left aortic arch with a right-sided DA, originating from the proximal descending aorta in two, from the aortic arch in one, and from the brachiocephalic trunk in another. Two patients had a right aortic arch with a left-sided DA arising from the left innominate artery.

   The pulmonary insertion site varied, with two cases connecting at the PA bifurcation, while in the remaining four cases, PA offset was observed, with insertion to the left pulmonary artery (LPA) in two cases and to the right pulmonary artery (RPA) in the other two.

   Anatomical assessment derived from VR models were fully consistent with angiographic findings, confirming the accuracy of VR in identifying anatomical variations (Table 1).

   Morphological evaluation of DA tortuosity also confirmed complete agreement between VR and angiography based on current classification standards. Four cases were classified TII, including two with an L-shape and two with a U-shape. Two cases were classified as TIII: one with an S-shape and one with a twisted morphology (Fig. 1, Table 1).

2. Length measurements and curvature index (CI)

   In all cases, across both TII and TIII tortuosity, VR-derived center-line lengths [L1] were consistently longer than those obtained from angiographic measurements. The median [L1] value in VR was 30.6 mm (range: 15.7–38.3 mm; mean: 28.9 mm; SD: 8.2), compared to 26.0 mm in angiography (range: 14.3–31.4 mm; mean: 24.1 mm; SD: 5.5). The paired t-test demonstrated a statistically significant difference (t = 2.80, p = 0.038), with percentage differences ranging from 2.4 to 41.3% and the largest discrepancies observed in two patients classified as TIII.

   For straight-line measurements [L2], the median difference was − 0.85 mm (range: − 5.90 to 2.40 mm), with percentage differences ranging from − 36.88 to 10.67%, with was not statistically significant (t = − 1.22, p = 0.276). However, VR-derived [L2] measurements were longer in TIII cases, whereas angiographic [L2] values tend to be greater in patients with TII tortuosity (Fig. 2, Table 1, Supplementary Video 1).

   The curvature index (CI) was significantly higher in VR-based measurements, with median of 0.4 (range: 0.34–0.68; mean 0.46; SD: 0.13) versus median 0.29 (range: 0.15–0.48; mean: 0.29; SD: 0.12) from angiography. The paired t-test confirmed a statistically significant difference (t = 8.55, p < 0.001) with VR-derived CI exceeding angiographic measurements by 31.4–126.7%. The largest discrepancies were observed in patients with type III tortuosity, where VR-based CI values were 2.08 and 2.28 times higher than angiographic values achieved from angiography (patients No.1 and 5, respectively; Table 1).

3. Projection angle alignment: VR-based vs. angiographic projections

   The retrospective comparison between VR-recommended and procedurally implemented angiographic projections, excluding one outlier case (patient No.5, where a completely different approach was used), revealed an angular deviation ranging from 6 to 30°, with a median of 12° Statistical analysis showed no significant difference (t = − 1.32, p = 0.259), indicating good agreement between VR-based planning and the final procedural projections. Additionally, the number of diagnostic angiographies performed before selecting the final projection was recorded for each case, ranging from 2 to 4, to evaluate the procedural value of VR planning. The smallest angular deviation (6°) was observed in two cases: patient 4 and the left anterior oblique (LAO) component of patient 6. Notably, these both patients also required the highest number of contrast injections (n = 4) to determine the final optimal angiographic projection. This indicated that increased effort to optimise projection selection intraoperatively was associated with closer alignment with VR recommendation. In one case, VR-recommended a right anterior oblique (RAO 65°) projection with cranial (CRA 30°) angulation, while the operator ultimately used a LAO 90° view, achieving procedurally adequate imaging after two contrast injections. (Fig. 3, Table 1).

4. Vascular access assessment (vascular access prediction)

   In all cases, VR-based assessment successfully identified the most favourable vascular access route, defined as the approach that ultimately provided the most gentle catheter entry angle into the ductus arteriosus and enabled smooth advancement. Initial concordance between VR predictions and procedural access routes was 50% (3 of 6 cases). Including reinterventions performed by the same operator, already familiar with the DA anatomy, concordance increased to 83% in 5 out of 6 cases (Fig. 4, Table 1, Supplementary Video 2).

   The only discordant case involved a 2.6 kg neonate in whom a hybrid approach was initially selected due to low body weight and PA narrowing with offset. During reintervention, femoral artery access was used instead, as both the carotid and subclavian arteries presented with insufficient vascular dimensions to allow safe vascular access.

5. Additional structures

   Virtual reality imaging consistently enabled visualization of additional thoracic structures, including lung tissue and main bronchi, in all cases, using dedicated filters from the software’s tool library. In patients with anomalous pulmonary venous return—one with partial (PAPVR) and one with total (TAPVR)—VR accurately depicted the trajectory of the anomalous vessel and its spatial relationship to the DA, allowing assessment of the potential compression risks during stent implantation.

   In another case, VR reconstruction identified a persistent vertical vein connecting the left superior vena cava and the left atrium, providing essential anatomical insight for planning both DA stenting and static atrial septostomy as the first stage of palliation for a complex congenital defect [19] (Fig. 5, Supplementary Video 3).

Fig. 1.

Virtual reality-based Three-Dimensional (3D) models of the ductus arteriosus (DA) generated from computed tomography (CT) imaging in all analysed patients. Models are shown in various spatial orientations and with different imaging filters to optimize anatomical assessment. A: Left lateral view: Patient with a left-sided DA originating from the left innominate artery classified as Type III (twisted) based on tortuosity index. The DA initially courses vertically downward and exhibits increased tortuosity near its insertion into the left pulmonary artery, with marked offset from the right pulmonary artery, posing a potential risk of branch pulmonary artery jailing during DA stenting. B: Anterior view: Patient with right aortic arch and a left innominate artery. The DA follows an L-shaped course, classified as Type II. C: Anterior view: Patient with heterotaxy syndrome and left aortic arch. The DA arises from the brachiocephalic trunk and follows a U-shaped trajectory, classified as Type II. D: Anterior view: Patient with a DA arising from the undersurface of the aortic arch, showing a Type II, U-shaped morphology. E: Right cranio-lateral view: Patient with a DA originating from the proximal descending aorta. The DA has two significant bends, forming an S-shaped trajectory, classified as Type III. F: Right lateral view: Patient with a DA originating from the proximal descending aorta. Standard CT underestimated anatomical detail, but VR modelling enabled enhanced spatial interpretation. A specific software filter was applied to improve tissue contrast, which moderately enhanced overall model clarity. The DA follows an L-shaped trajectory, classified as Type II. AoArch aortic arch, AoAsc ascending aorta, AoDesc descending aorta, BCT brachiocephalic trunk, DA ductus arteriosus, L-InnA left innominate artery, LPA left pulmonary artery, RPA right pulmonary artery, TAPVR total anomalous pulmonary venous return

Fig. 2.

Comparison of ductus arteriosus (DA) length measurements using conventional angiography and virtual reality (VR) models derived from CT imaging. Two parameters were assessed: total DA length [L1], measured along the ductal center-line traced through the ductal lumen, and straight-line length [L2], defined as the direct distance between the aortic origin and pulmonary insertion. A: Patient No. 1—Angiography (posteroanterior view): Angiographic assessment of DA morphology showing L1 (green segmented lines, X1–X5) and L2 (green straight-line, X6). B: Patient No. 1—VR Model (posteroanterior view): 3D VR reconstruction allowing interactive spatial assessment. L1 (blue continuous line) and L2 (blue dashed line) were measured within the VR environment, preserving vessel curvature and anatomical context. C: Patient No. 5—Angiography (left lateral view): Conventional angiographic measurement of L1 (green segmented lines, X1–X3) and L2 (green straight-line, X4). D: Patient No. 5—VR Model (right lateral view): VR-based analysis of L1 (blue continuous line) and L2 (blue dashed line), demonstrating more precise spatial representation of DA morphology, and notable difference in length compared to angiography

Fig. 3.

Selection process of the optimal angiographic projection for ductus arteriosus intervention: comparison between (4) angiographic views and the projection recommended by virtual reality. A–C: Initial angiographic views. Three consecutive projections were performed to identify the optimal working projection for DA stenting; however, none provided sufficient visualisation of ductal morphology to be selected as the final working plane. D: Final angiographic projection. The fluoroscopic view ultimately selected for stent implantation, obtained with the fourth contrast injection. E: VR-recommended projection: optimal projection generated within the VR environment with precise angulation adjustment. Notably, the VR-suggested projection closely corresponded to the final angiographic view, which was only achieved after multiple angiographies. Ductus arteriosus is marked with black arrows in all panels. CRA cranial, DA ductus arteriosus, LAO left anterior oblique, PA posteroanterior view

Fig. 4.

Virtual reality-based evaluation of optimal catheter insertion angles for ductus arteriosus (DA) stenting. This figure demonstrates the benefits of using immersive VR models for preprocedural vascular access planning. By providing full 3D spatial perception, via head-mounted VR goggles and hand controllers, the optimal catheter trajectory can be assessed, allowing for precise evaluation of insertion angles and vascular curvature. The VR-based assessment facilitates the selection of the most favourable vascular access route. A: Posterior view—femoral artery access: The catheter path is traced from the femoral artery through the descending aorta (AoDesc) to the DA. The required catheter angulation for successful transition from the descending aorta to the DA is approximately 110°, the most favourable of all tested access options in this patient. B: Anterior View—femoral vein access: The catheter trajectory extends from the femoral vein, through the ventricular septal defect (VSD) into the ascending aorta (AoAsc), and then to the DA. The required catheter angulation for the transition from AoAsc to DA is 95°. When considering this approach, additional challenges include navigation through intracardiac structures. C: Anterior view—right axillary artery (R-AxA) access: The angle required to reach the DA from R-AxA is 52°, the sharpest angle among all tested approaches. It necessitates considerable catheter curvature, making this access technically challenging. D: Anterior view – right carotid artery (R-CaA) access: The angle for transition from R-CaA to the DA is 67°, also steep and requiring significant catheter bending. E: Anterior view—left carotid artery (L-CaA) access: The required catheter angle from L-CaA to DA is 90°, which, while still steep, remains more favourable than the R-CaA access in this case. F: Posterior and cranial view—left axillary artery (L-AxA) access: Due to vascular offset, the catheter trajectory from L-AxA to the DA requires at least two significant bends, increasing the technical difficulty of this approach

Fig. 5.

Virtual reality-based spatial assessment of thoracic structures surrounding the ductus arteriosus (DA) in complex congenital heart defect. This figure illustrates the added value of VR-derived 3D models in visualising adjacent thoracic structures, including vascular anomalies and airway anatomy, which are essential for effective procedural planning. A: Identification of additional vascular anomaly. Patient with total anomalous pulmonary venous return (TAPVR): VR reconstruction (posterior view) reveals close spatial proximity between the anomalous pulmonary vein and the DA, helping assess potential risks of vessel compression during interventional. B: The same patient as in panel A, with a similar posterior view. Combined vascular and airway assessment in TAPVR: VR-enabled simultaneous 3D visualization of vascular structures and airway, supporting detailed spatial evaluation not easily achievable with conventional imaging. C: Identification of additional vascular anomaly. Patient with partial anomalous pulmonary venous return (PAPVR). VR reconstruction depicted left upper pulmonary vein connected with left superior vena cava (L-SVC) [19]. D: Same patient as in panel A, with PAPVR. The VR model revealed a persistent vertical vein connecting the L-SVC to the left-sided pulmonary veins (upper and middle), which drained into the left atrium. This anatomical configuration was important for planning a staged interventional approach [19]. AoDesc descending aorta, DA ductus arteriosus, L-SVC left superior vena cava, LA left atrium, LPA left pulmonary artery, LUPV left upper pulmonary vein, RPA right pulmonary artery, TAPVR total anomalous pulmonary venous return

Discussion

This study demonstrates that VR-based 3D modelling provides added value in the procedural planning of DA stenting, particularly in cases with complex anatomy. CT-derived VR models enabled fast and highly detailed visualisation of DA characteristics, while immersive interaction supported more accurate interpretation of spatial relationships compared to the standard imaging. Moreover, VR offers a deeper spatial understanding of vascular structures, allowing for precise measurements, improved angulation assessment, and interactive procedural simulation.

Morphological classification of ductal tortuosity (Types I–III) was consistent across VR and angiography, validating the reliability of VR for anatomical assessment. Moreover, VR-derived total ductal length [L1] was significantly greater than angiographic measurements, particularly in highly tortuous ducts (TIII), which is consistent with previous findings using 3D computational or printed models [12]. Importantly, underestimation of total ductal length may lead to inappropriate stent sizing or suboptimal positioning, increasing the risk of migration or incomplete ductal coverage [8].

Conversely, straight-line length [L2] was often overestimated in angiography, especially in TII ducts. This discrepancy likely stems from the inherent limitations of two-dimensional (2D) angiographic projections. The process of selecting the optimal angiographic projection typically prioritizes clear visualization of both the DA origin and insertion. However, such projections do not always reflect the true 3D spatial configuration. As a result, the flattening effect of 2D imaging may artificially overestimate [L2] values, as demonstrated in our study. These findings suggest that VR models preserve the authentic 3D spatial relationships of vascular structures, avoiding the geometric distortion inherent in conventional 2D angiography.

Similarly, the CI, a recognized predictor of catheter navigation difficulty, was significantly higher in VR, again most pronounced in TIII ducts. This suggests that 2D angiography underrepresents true curvature, which is crucial for optimizing procedural strategy, as it directly impacts wire positioning, catheter stability, and the accuracy of stent deployment.

In our study, re-intervention was required in 3 of 6 cases, including both with type III tortuosity. One hybrid case required additional stent implantation due to deep surgical sheath placement at the ductal margin (TIII), another required proximal re-stenting (TIII), and a third needed balloon re-dilation without additional implants (TII). These findings support the predictive value of VR-based length assessment in procedural planning.

Virtual models also contributed to optimal projection planning. Angular deviations between VR-recommended and implemented fluoroscopic projections were generally small (median 12°). One outlier, in which the procedural projection deviated from the VR recommendation, suggest the value of non-standard angles not typically used in routine practice.

In cases with the highest number of contrast injections, the final projection closely aligned with VR suggestions, highlighting its potential to reduce trial-and-error during angiographic acquisition. This is particularly important consideration in neonatal catheterisation, where minimising contrast volume, radiation exposure, and overall procedure time is crucial.

Optimal vascular access is essential for successful DA stenting, with femoral vessels commonly used, and axillary or carotid assess preferred particularly for vertical ducts, despite potential neurological risks [2, 8, 22, 23]. Vascular access planning showed initial concordance between VR predictions and operator choice in 3 of 6 cases, increasing to 5 of 6 (83%) when reinterventions were included. This reflects VR’s value not only in procedural foresight but also in aligning with real-world strategies as anatomical familiarity improves.

When procedural success depends on careful planning, VR technology may offer added value beyond anatomical analysis. By facilitating early strategy formulation and supporting less experienced clinicians, VR has the potential to enhance procedural efficiency and reduce unexpected intraoperative adjustments. Furthermore, the immersive format facilitates multidisciplinary communication by providing a shared understanding of complex anatomy [8, 21].

A notable advantage of VR is its ability to visualise adjacent thoracic structures. In patient with TAPVR, it enabled precise evaluation of surrounding tissues and compression risk. In another case, VR revealed a persistent vertical vein, allowing a rare, combined intervention that might not have been considered without this insight [19].

Overall, VR-based modelling offers comprehensive, 3D evaluation of cardiovascular and thoracic anatomy, supporting more precise, informed, and collaborative interventional planning.

Limitations and Future Directions

This study has several limitations. The small sample size restricts the generalizability of the findings, underscoring the need for larger cohorts to confirm the reproducibility and broader applicability of VR-based planning. Although VR showed clear benefits in anatomical assessment and procedural planning, its direct impact on procedural success and complication rates was not evaluated. Further prospective studies are needed to assess its clinical utility and long-term outcomes in real-world interventional practice.

Conclusion

This study confirms the feasibility and accuracy of VR-based preprocedural planning in patients with DDPC. Virtual reality modelling provides an immersive, detailed understanding of DA anatomy, with clear advantages in length assessment, curvature evaluation, fluoroscopic projection planning, and vascular access selection. By enabling preprocedural evaluation of spatial relationships, anatomical angles, and vascular complexity, VR supports more predictable and precise interventions. In complex cases, its use may reduce complications and contribute to improved long-term outcomes in interventional cardiology.

Supplementary Information

Below is the link to the electronic supplementary material.

Author Contributions

Conceptualization and methodology: J.S.; Original draft and figures preparation: J.S. Resources: K.C.-W.; Data curation: J.S. and K.C-W.; Software: J.S. and K.C.-W.; Conceptual supervision, guidance on study design and final aproval of the manuscript: S.G. All authors have read and agreed to the published version of the manuscript.

Funding

Virtual Reality project was supported by the Jagiellonian University Medical College internal grant No. N41/DBS/001219 for 2023-2024.

Data Availability

No datasets were generated or analysed during the current study.

Declarations

Conflict of interest

The authors declare no competing interests.

Ethical Approval

This retrospective study was approved by the Institutional Medical Research Ethics Committee of Jagiellonian University Medical College.