Three-Dimensional Virtual and Printed Prototypes in Complex Congenital and Pediatric Cardiac Surgery—A Multidisciplinary Team-Learning Experience

Three-dimensional (3D) virtual modeling and printing advances individualized medicine and surgery. In congenital cardiac surgery, 3D virtual models and printed prototypes offer advantages of better understanding of complex anatomy, hands-on preoperative surgical planning and emulation, and improved communication within the multidisciplinary team and to patients. We report our single center team-learning experience about the realization and validation of possible clinical benefits of 3D-printed models in surgical planning of complex congenital cardiac surgery. CT-angiography raw data were segmented into 3D-virtual models of the heart-great vessels. Prototypes were 3D-printed as rigid “blood-volume” and flexible “hollow”. The accuracy of the models was evaluated intraoperatively. Production steps were realized in the framework of a clinical/research partnership. We produced 3D prototypes of the heart-great vessels for 15 case scenarios (nine males, median age: 11 months) undergoing complex intracardiac repairs. Parity between 3D models and intraoperative structures was within 1 mm range. Models refined diagnostics in 13/15, provided new anatomic information in 9/15. As a team-learning experience, all complex staged redo-operations (13/15; Aristotle-score mean: 10.64 ± 1.95) were rehearsed on the 3D models preoperatively. 3D-printed prototypes significantly contributed to an improved/alternative operative plan on the surgical approach, modification of intracardiac repair in 13/15. No operative morbidity/mortality occurred. Our clinical/research partnership provided coverage for the extra time/labor and material/machinery not financed by insurance. 3D-printed models provided a team-learning experience and contributed to the safety of complex congenital cardiac surgeries. A clinical/research partnership may open avenues for bioprinting of patient-specific implants.


Introduction
"The essence of the virtual world is the freedom it allows for experimentation" [1]. Anatomical modeling of the patients' individual three-dimensional (3D) structures and 3D printing of the prototypes has won its place in personalized medicine and reconstructive surgery [2]. There are two types of 3D-printed objects in healthcare, as shown in Table 1. Table 1. Two types of 3D-printed objects in healthcare [3].

Types and Description Examples
3D-printed anatomical prototypes of an individual patient: replicate exact patient morphology; do not come into direct contact with the patient Anatomic models for demonstration, surgical planning, and emulations 3D-printed patient-specific medical hardware: newly-designed objects created by computer-aided design (CAD) based on and added to individual patient characteristics; direct patient contact Customized/personalized implants Prostheses External fixators Splints Surgical instrumentation and surgical cutting aides At present, pediatric and congenital cardiac surgery only utilizes 'type 1 anatomic models that promote a better understanding of complex anatomy by combining visual and tactile stimuli [4]. They are employed to plan and emulate the operation preoperatively. Models also improve communication within the multidisciplinary clinical team and towards patients and their relatives [5]. 3D models often bring new anatomical information and open the possibility of elaborating [6,7] and testing alternative surgical scenarios [8]. All these options are available before the actual intervention, so increased awareness contributes to greater procedural safety [9][10][11].
In this study, we report our single-center experience with 3D modeling and printed models in surgical planning of complex congenital cardiac surgery as an interdisciplinary team-learning process to realize and validate possible clinical benefits. We also present a sustainable co-operation model between applied/basic science and the clinical practice that broadens the possibilities of a learning organization and can overcome constraints in the financing of the prototypes and may initiate further developments towards the construction of 'type 2 3D-printed patient-specific implants: patches, valves, and conduits.

Materials and Methods
This study was carried out at Sheikh Khalifa Medical City (SKMC), a governmentowned, teaching hospital in the United Arab Emirates. There are approximately 150 advanced cardiac imaging procedures, around 150-200 mostly interventional cardiac catheters, and 400 congenital heart surgeries performed annually at this institution. Pediatric cardiac services offer tertiary care, comprehensive and uninterrupted treatment coverage for the majority of the population (approximately 9 million) [12]. The surgical patient population is skewed towards the complex, younger patients requiring more urgent interventions. Case scenarios requiring 3D virtual modeling and 3D printing were elected from the subgroup of patients requiring reoperations with the most complex anomalies (with Aristotle Basic Complexity Score over 10) [13] (approximately 10% of surgical patient material) spanning over a period of 18 months of the introduction of 3D-printed models. Indication for 3D-printed models was raised at the multidisciplinary meeting for a better understanding of the complex anatomical situation, consideration of alternative surgical solutions, intervention planning, and rehearsing. Institutional Research Ethics Committee review and approval were waived for this study due to the fact that anatomical models classified as research models were not coming into direct contact with the patients. Informed consent was obtained from the patients/guardians. Non-clinical participants of the study strictly adhered to patient data confidentiality.

Financing
In an earlier phase of the project, segmentation and model printing was outsourced to Materialise, Leuven, Belgium, and expenses were covered by a grant from Hamdan bin Rashid Al Maktoum Foundation for Distinguished Academic Performance, Dubai, UAE. In the second phase, a cost-efficient sustainable manufacturing and financing model was

Imaging and Manufacturing
ECG-gated computed tomography (CT) was performed combined with contrastenhanced angiography recorded during expiration at a resolution of 0.3 to 0.7 mm. CT was done within 6 weeks prior to the planned surgical procedure. A virtual model (segmentation) was created from the digital imaging and communication in medicine (DICOM) data set using commercial software (Mimics ® , Materialise, Leuven, Belgium [14]) and an open-source 3D Slicer [15]. In all cases, a research engineer carried out semi-automatic segmentation with the clinician's participation. The segmentation steps involved loading the DICOM data into Mimics or 3D Slicer, using a combination of intensity-based thresholding to segment the blood pool signal, and creating a hollow model of 2-3 mm wall thickness to outline the surface of the vessels. Manual segmentation of the heart tissue was performed when needed for anatomical clarity. Manual model refinement (preprocessing) was done at the end and approved by clinicians. A fully-rotatable and sliceable virtual 3D model of the cardiac structures was presented in a pdf file on the computer's screen (2D). From the stereolithography (STL) file, a life-size 'blood volume' model made of VeroMagenta (Stratasys, Eden Prairie, MN, USA), hard opaque material and another 1.5× scaled 'hollow' made of TangoPlus (Stratasys, Eden Prairie, MN, USA) and/or HeartPrint Flex (Materialise, Leuven, Belgium), a flexible, translucent material, models were printed (see below). Magnification of the hollow models was used to facilitate surgical emulation. Figure 1 illustrates the steps of 3D modeling and creating 3D-printed prototypes and holograms.
In an earlier phase of the project, segmentation and model printing was outsourced to Materialise, Leuven, Belgium, and expenses were covered by a grant from Hamdan bin Rashid Al Maktoum Foundation for Distinguished Academic Performance, Dubai, UAE. In the second phase, a cost-efficient sustainable manufacturing and financing model was established in a clinical/research partnership with the Core Technology Platforms at New York University Abu Dhabi (NYUAD).

Imaging and Manufacturing
ECG-gated computed tomography (CT) was performed combined with contrast-enhanced angiography recorded during expiration at a resolution of 0.3 to 0.7 mm. CT was done within 6 weeks prior to the planned surgical procedure. A virtual model (segmentation) was created from the digital imaging and communication in medicine (DICOM) data set using commercial software (Mimics ® , Materialise, Leuven, Belgium [14]) and an opensource 3D Slicer [15]. In all cases, a research engineer carried out semi-automatic segmentation with the clinician's participation. The segmentation steps involved loading the DI-COM data into Mimics or 3D Slicer, using a combination of intensity-based thresholding to segment the blood pool signal, and creating a hollow model of 2-3 mm wall thickness to outline the surface of the vessels. Manual segmentation of the heart tissue was performed when needed for anatomical clarity. Manual model refinement (preprocessing) was done at the end and approved by clinicians. A fully-rotatable and sliceable virtual 3D model of the cardiac structures was presented in a pdf file on the computer's screen (2D). From the stereolithography (STL) file, a life-size 'blood volume' model made of Vero-Magenta (Stratasys, Eden Prairie, MN, USA), hard opaque material and another 1.5× scaled 'hollow' made of TangoPlus (Stratasys, Eden Prairie, MN, USA) and/or HeartPrint Flex (Materialise, Leuven, Belgium), a flexible, translucent material, models were printed (see below). Magnification of the hollow models was used to facilitate surgical emulation. Figure 1 illustrates the steps of 3D modeling and creating 3D-printed prototypes and holograms.  The printed prototypes were classified as 'research models' and did not come into direct contact with the patient's tissues [16]. The blood volume model was utilized for exact measurements of anatomical structures, such as vessel diameters. The hollow model featured endo-and epicardial surfaces, i.e., the wall thickness of the given heart cavity and/or great vessel between them. In order to simulate natural tissues, material characteristics of the hollow model were chosen to be similar to that of the human arteries (Young's modulus Biomolecules 2021, 11, 1703 4 of 20 between 0.2 and 9 MPa; distensibility between 1.2 × 10 −3 and 6.6 × 10 −3 mmHg −1 ) and, to practice surgical emulation enlarged models (scale 1.5×) were created. The accuracy of the 3D-printed models was cross-checked with the intraoperative surgical conditions.

Manufacturing of the 3D-Printed Models
3D printing was performed using the Stratasys J750 resin-based 3D printer (Stratasys, Eden Prairie, MN, USA). This multijet 3D printing technology allows concurrent jetting and UV-curing of several materials. The materials have different physical properties and can be mixed to obtain a range of intermediate physical properties and colors that would match the properties and colors of the different tissues. The digital 3D models obtained from the image segmentation were sliced and 3D-printed layer by layer. The empty domains within the model were filled with water-soluble support material. After 3D printing, the models were submerged in a water bath with 2% NaOH and 1% Na 2 SiO 3 to remove the support.

Preoperative Team Emulations
Repeated rehearsals and emulation sessions were held with the intraoperative team as well as with the multidisciplinary team. The added value and acceptance of models in clinical decision making were assessed as a team-learning experience. Members in the multidisciplinary team and the relatives of patients gave informal feedback and filled questionnaires in throughout the project. Questionnaires polled the opinion separately for the medical and non-medical participants on the added value of 3D-printed models over virtual ones, model accuracy and cost/benefit adequacy, and medical and non-medical participants' willingness to assume extra work and expenses related to model production.

Patient Characteristics and Material
3D virtual and printed models of the heart and the great arteries were created for fifteen complex congenital cardiac patients (nine boys, six girls, median age: 11 months; range: 6.5-96 months) in preparation for cardiac surgery. Thirteen out of fifteen patient (86.66%) were planned for redo-operations, in 10 cases as part of uni-or biventricular staging. No operative morbidity/mortality occurred. Table 2 shows the study characteristics of the patients.
Despite the complex anatomy, restoration of biventricular circulation was possible for patients with two ventricles (10/15), but in one scenario (Case 15, Figure 3), both the 3D virtual and printed models were extremely helpful in disproving the feasibility of reconnection of the left ventricle to the aorta, and thus, biventricular circulation. Biventricular repairs (9/15 = 60%)-mostly (re)operations-associated with an Aristotle Basic Complexity Score [13] of the mean of 10.64 ± 1.95. Owing to detailed and strategic surgical rehearsing on the 3D models, successful complete biventricular repair-consisting of repair of pulmonary venous stenosis, atrial separation, AV-valve repair, intraventricular rerouting, take-down of previous superior bidirectional cavopulmonary anastomosis, and implantation of RV-PA conduit-could be performed for the most complex case scenario (Case 10) demonstrated on Figures 4 and 5.    Despite the complex anatomy, restoration of biventricular circulation was possible for patients with two ventricles (10/15), but in one scenario (Case 15, Figure 3), both the 3D virtual and printed models were extremely helpful in disproving the feasibility of reconnection of the left ventricle to the aorta, and thus, biventricular circulation.  Biventricular repairs (9/15 = 60%)-mostly (re)operations-associated with an Aristotle Basic Complexity Score [13] of the mean of 10.64 ± 1.95. Owing to detailed and strategic surgical rehearsing on the 3D models, successful complete biventricular repair-consisting of repair of pulmonary venous stenosis, atrial separation, AV-valve repair, intraventricular rerouting, take-down of previous superior bidirectional cavopulmonary anastomosis, and implantation of RV-PA conduit-could be performed for the most complex case scenario (Case 10) demonstrated on Figures 4 and 5.    Patients with univentricular physiology (6/15 = 40%) could have progressed on their staging, except for one patient with intractable pulmonary hypertension (Case 4). Three patients (20%) had significant issues with coronary artery origin and/or course anticipated from clinical imaging but only fully revealed on the virtual and/or printed 3D model. Figure 6 shows the anomalous origin of the right coronary artery from the left anterior descending branch of the single left coronary artery. Operative procedures were performed along with preoperative plans; no deviation from the rehearsed steps occurred. No reoperation during follow-up was necessary.

Indication for 3D Modeling and Printing
Indication of 3D modeling was (1) to further illuminate the spatial relationship in the segmental anatomy and for 3D printing (2) to facilitate surgical planning. 3D virtual models were always available, from which blood volume (N = 10) and hollow (N = 9) prototypes were printed. In three cases, analysis of the virtual 3D model was sufficient to refine the anatomy and formulate a surgical plan. These scenarios revealed (1)

Indication for 3D Modeling and Printing
Indication of 3D modeling was (1) to further illuminate the spatial relationship in the segmental anatomy and for 3D printing (2) to facilitate surgical planning. 3D virtual models were always available, from which blood volume (N = 10) and hollow (N = 9) prototypes were printed. In three cases, analysis of the virtual 3D model was sufficient to refine the anatomy and formulate a surgical plan. These scenarios revealed (1) abnormal right coronary artery from the left anterior descending branch (Case 5), (2) an obstructed left main coronary orifice compressed by a dilated right pulmonary artery (Case 6), and (3) exact location of left-sided intra-atrial obstruction (Case 8).

Added-Value and Accuracy of 3D Modeling/Printing
Exact size measurements were taken from blood volume models. The shapes of implants in relation to the intracardiac defects were assessed, and steps of the repair were designed and rehearsed on the hollow models. Registration of the AV valves (on the CT and/or MRI dataset) is currently suboptimal to include them in the 3D-printed models.
Virtual and printed 3D models revealed new anatomic findings relevant for the surgery in 9/15 (60%). For three patients (Cases 3: site of subaortic resection; Case 6: external coronary compression as the cause of LV failure; and Case 15: the presence of a VSD obstructed by LV thrombus) previously classified inoperable due to high surgical risk, 3D models gave new insights in the anatomy and opened the possibility of setting up an alternative surgical plan. In other words, these patients became 'operable'. Models also had a high yield in clarifying the surgical anatomy (13/15 = 86.66%) that translated into specific recommendations on the surgical approach (e.g., which cardiac chamber should be opened), perfusion technique (e.g., where to cannulate), surgical steps of complex repairs. The added value of the 3D-printed models is that they enabled surgical emulation and preprocedural planning for the shape and size patches, conduits needed [17].

Added-Value and Accuracy of 3D Modeling/Printing
Exact size measurements were taken from blood volume models. The shapes of implants in relation to the intracardiac defects were assessed, and steps of the repair were designed and rehearsed on the hollow models. Registration of the AV valves (on the CT and/or MRI dataset) is currently suboptimal to include them in the 3D-printed models.
Virtual and printed 3D models revealed new anatomic findings relevant for the surgery in 9/15 (60%). For three patients (Cases 3: site of subaortic resection; Case 6: external coronary compression as the cause of LV failure; and Case 15: the presence of a VSD obstructed by LV thrombus) previously classified inoperable due to high surgical risk, 3D models gave new insights in the anatomy and opened the possibility of setting up an alternative surgical plan. In other words, these patients became 'operable'. Models also had a high yield in clarifying the surgical anatomy (13/15 = 86.66%) that translated into specific recommendations on the surgical approach (e.g., which cardiac chamber should be opened), perfusion technique (e.g., where to cannulate), surgical steps of complex repairs. The added value of the 3D-printed models is that they enabled surgical emulation and preprocedural planning for the shape and size patches, conduits needed [17]. Figure 7 demonstrates precise anatomic details of complex pulmonary atresia with MAPCAs (Case 7). Hands-on use of the 3D-printed blood-volume model was essential to assess the spatial relationship of the MAPCAs and the tracheobronchial tree and to rehearse the plan of unifocalization. For the lack of an adequate control group, we could not draw comparisons on the 3D models' impact on saving surgical time. Intraoperative identification of the structures seen on the models, however, reportedly created a 'déjà vu' effect, and it improved the surgical flow and confidence in executing the complex operations.
demonstrates precise anatomic details of complex pulmonary atresia with MAPCAs (Cas 7). Hands-on use of the 3D-printed blood-volume model was essential to assess the spatia relationship of the MAPCAs and the tracheobronchial tree and to rehearse the plan o unifocalization. For the lack of an adequate control group, we could not draw compar sons on the 3D models' impact on saving surgical time. Intraoperative identification of th structures seen on the models, however, reportedly created a 'déjà vu' effect, and it im proved the surgical flow and confidence in executing the complex operations. Accuracy of the 3D virtual models showed exact matching with the CT-dataset the were based on. Accuracy depended on two factors, manual segmentation for the virtua model and 3D printed resolution for the printed model, which was always better tha imaging resolution. 3D-printed models were then validated by comparing the diameter of the inferior vena cava and ascending aorta from the CT scan and comparing with thos from the blood volume 3D-printed model (data were extrapolated for the scaled proto types). No significant difference was found between CT scans, the virtual, and 3D-printe models. The 3D-printed prototypes were also evaluated with the intraoperative finding Accuracy of the 3D virtual models showed exact matching with the CT-dataset they were based on. Accuracy depended on two factors, manual segmentation for the virtual model and 3D printed resolution for the printed model, which was always better than imaging resolution. 3D-printed models were then validated by comparing the diameters of the inferior vena cava and ascending aorta from the CT scan and comparing with those from the blood volume 3D-printed model (data were extrapolated for the scaled prototypes). No significant difference was found between CT scans, the virtual, and 3Dprinted models. The 3D-printed prototypes were also evaluated with the intraoperative findings, and the models were precise to the 1 mm range (quantitative identifier). No morphological mismatch (qualitative identifier) was found between the 3D-printed models and the intraoperative anatomy in the series. Figure 8 (Case 13) demonstrates the accuracy of the model compared to the intraoperative situation. and the models were precise to the 1 mm range (quantitative identifier). No morphological mismatch (qualitative identifier) was found between the 3D-printed models and the intraoperative anatomy in the series. Figure 8 (Case 13) demonstrates the accuracy of the model compared to the intraoperative situation.

Technical, Organizational and Financial Aspects
Creation and printing 3D patient-specific prototypes in this series involved data acquisition (from CT angiography) and segmentation by 2-4 h of computer work with special software and expertise. The printing process required additional 4-6 h depending on the model's size and complexity. In the starting phase of the project, for the lack of established hardware and infrastructure, outsourcing the production phases to an internationally renowned company (Materialise, Leuven, Belgium) seemed a viable solution. Foundation of clinical/research co-operation with specialized technical expertise for image segmentation and additive manufacturing at NYUAD opened up an avenue for organic development in regular 3D printing and set the basis for related future projects, e.g., in 3D bioprinting. The shift from outsourcing to local co-operation resulted in significant improvements in the speed of manufacturing the prototypes by taking away geographical barriers and provided a sustainable financing structure.

Team-Learning Experience
3D modeling and printing added a new modality for preoperative planning and communication. Sporadic need for 3D models (N = 3) in the first half of the study period of 18 months swapped to a surge of case scenarios (N = 12) in the second half. Rapid learning of the complex cardiac anatomy was observed among clinical engineers supported by clinicians' participation. Team rehearsals significantly contributed to establishing a shared

Technical, Organizational and Financial Aspects
Creation and printing 3D patient-specific prototypes in this series involved data acquisition (from CT angiography) and segmentation by 2-4 h of computer work with special software and expertise. The printing process required additional 4-6 h depending on the model's size and complexity. In the starting phase of the project, for the lack of established hardware and infrastructure, outsourcing the production phases to an internationally renowned company (Materialise, Leuven, Belgium) seemed a viable solution. Foundation of clinical/research co-operation with specialized technical expertise for image segmentation and additive manufacturing at NYUAD opened up an avenue for organic development in regular 3D printing and set the basis for related future projects, e.g., in 3D bioprinting. The shift from outsourcing to local co-operation resulted in significant improvements in the speed of manufacturing the prototypes by taking away geographical barriers and provided a sustainable financing structure.

Team-Learning Experience
3D modeling and printing added a new modality for preoperative planning and communication. Sporadic need for 3D models (N = 3) in the first half of the study period of 18 months swapped to a surge of case scenarios (N = 12) in the second half. Rapid learning of the complex cardiac anatomy was observed among clinical engineers supported by clinicians' participation. Team rehearsals significantly contributed to establishing a shared mental model and stepwise operative plan with alternative scenarios and respective responsibilities among operative team members. Questionnaires on different aspects (Table  3) revealed that virtual 3D models had already provided adequate information for most members of the multidisciplinary team; however, surgeons and patient relatives preferred the hand-held, 3D-printed prototypes. Accuracy was considered as good, and most responders reported improved communication as a primary advantage. Patient relatives were dedicated to subscribing additional costs related to 3D modeling and printing. Table 3. Average values of opinions of the multidisciplinary team and patient relatives on 3D modeling based on a questionnaire survey. Range of values: 1 = strongly disagree, 2 = disagree, 3 = indifferent, 4 = agree, 5 = strongly agree; n/a: non-applicable.

Questions
Multidisciplinary

Discussion
This study represents our multidisciplinary team's learning experience with 3D virtual and printed models in preparing for complex, mostly redo pediatric cardiac procedures. High anatomical and procedural complexity in our series warranted a 3D understanding of the scenarios. 3D-printed models naturally contributed to an interactive team experience; at rehearsals, they allowed that the entire clinical team would appropriate a shared mental image and detailed plan. Parents not familiar with reading images of traditional medical imaging themselves preferred touchable physical objects to virtual ones. Furthermore, interaction with clinical engineers, experts in additive manufacturing, and bioengineers promoted knowledge of each other's fields that could inspire continuing crosstalk and co-operation in biofabrication.
The risk of open-heart surgery for complex congenital heart defects and in reoperations is still significantly higher compared to other surgical activities prompting for safe surgery measures [24]. Operative efficiency and learning curve are nowadays not expected to impact outcomes [25]. 3D-printed models improve understanding of 3D anatomy and allow anticipation and communication of technical challenges [19,21,24]. Anatomical specimens have a long history, and they significantly contributed to abstracting individual features into general rules, connecting function to structure [26][27][28]. Generations of students of congenital heart disease familiarized themselves with the complex anatomical relationships in the pathological museum [29,30]. The unique advantage of 3D models is that they convey haptic information and binocular vision to complement and strengthen multisensory convergence in creating a mental model of an object [31][32][33]. The strength of palpation is illustrated by that tactile information can even suppress image perception transmitted from the dominant eye under experimental conditions, when the two eyes look at different sights [34,35]. The combination of vision and haptics leads to better analysis, faster decisions, and reduced number of touches [36]. In contrast, advanced clinical imaging, e.g., 3D/4D echocardiography, 3D rotational angiography, CT angiography, and magnetic resonance imaging/MRI, creates 3D models that still exist in the two-dimensional space of a computer screen [36,37]. In order to overcome this possible problem, new technologies have emerged, where the virtual 3D model is projected the virtual and/or mixed reality space [38,39] and the spectator wearing special glasses can interact with them with/out haptic feedback [40][41][42] (Figure 1B). These developing techniques are in the process of maturation and in finding their niche in the clinical armamentarium [43][44][45][46]; however, the need for 3D-printed models remains, as they readily exist in the physical reality and they accurately demonstrate rather complex morphologies [47], which can be printed with the physical properties of native tissues [48].

3D-Printed Models Promote Team Learning
Planning for complex congenital cardiac operations is both a cognitive process for the operating surgeon [49][50][51] and for the interdisciplinary team [52]. The objective is to build shared mental models of the anatomy, steps of the operation with respective responsibilities, and anticipating avenues to overcome possible complications. Team members, however, are bound by their own perspectives, knowledge gaps, and team dynamics. "Marrying teamwork to one's own ego is quite difficult at times-and now we're still learning that" [53]. In the present study, pediatric cardiologists were already content with the 3D virtual models, whereas surgeons and the operative team facing the context of increasing complexity still required the added demonstrating value of 3D-printed models. Non-medical personnel (i.e., patient relatives), being unfamiliar with 2D medical imagery, disfavored virtual models. The learning process, i.e., building a mental model, started with the clinicians' participation in the segmentation process. Overall, we observed an increasing interest in 3D modeling and printed prototypes demonstrated by the escalating incidence of case scenarios during the study period. Preoperative emulation sessions with the 3D-printed models and with multidisciplinary participation team facilitated shared team-learning experience by taking away time constraints typically present in the operating theatre, allowing reversible actions and repetitions and simplifying complexity. Our experience replicates the reports by others [54,55] that building shared mental models and language promotes shared responsibilities, improved teamwork, and communication. The utilization of 3D-printed models remains a central element of the team experience [56].

Cost-Reimbursement Constraints and Technical Limitations
At present, 3D modeling and 3D printing does not have an internationally accepted current procedural terminology (CPT) code [57]. Since remuneration of treatment activities is based on these codes, their absence represents an obstacle with insurance companies and prompts for alternative funding. Costs of 3D medical modeling and printing are funded by insurance in Japan, and negotiations at the American Medical Association are reportedly underway to include them in the reimbursed activities [58]. A questionnaire survey in the present study showed high willingness among patient relatives to subscribe to extra expenses associated with model production. 3D printing of 'rigid' blood-volume models is relatively inexpensive due to the available techniques and single inexpensive material (typically polylactic acid, PLA). However, printing a 'hollow' model with flexible, vessel/chamber-mimicking material is an expensive and difficult task because it requires advanced printing capabilities only offered by a handful of 3D printers (such as Stratasys J750 multijet printer, Stratasys, Eden Prairie, MN, USA), using special materials made of multiple resins. A clinical/research institutional co-operation may relieve the financial burden for individual case scenarios, and it can also promote co-operation, e.g., in translational research in the treatment of congenital heart disease.

Future Directions
'Type 1' 3D-printed cardiac models do not come to direct contact with the patient; however, computer-aided design and 3D bioprinting can now produce individually implantable ('type 2') prototypes in many domains of reconstructive surgery [59][60][61]. Congenital and pediatric cardiac surgery mostly deals with reconstructive patient scenarios when defects are closed, various segments of the heart are connected, valves are implanted. Congenital cardiac surgery applies prosthetic material in the majority (>75%) of its procedures [62]. However, biomaterials currently available in our profession lack the potential of growth, the implanted conduits and valves derange over time surrendering recipients to repeated reoperations [63], and currently, numerous projects address biofabrication of the patches, valves or conduits [64][65][66]. 3D-printed prototypes can play a significant role in biofabrication as templates for computer-aided design, bioscaffolds [67]. Availability of patient-specific, autologous (i.e., non-immunogenic), structurally sound, viable and growing implants could cancel reoperations and could entail improved quality-of-life of the individual patient and relief of a significant public health burden [68]. We believe that clinical/research co-operation in biofabrication and 3D bioprinting-as in our study-holds the promise of realizing these future goals [69,70].
As another direction, rapid development in holographic technology carries potentials for 3D modeling. Holograms offer quicker production time and low budget solution compared to 3D printing [45,46]. Applications allow free maneuverability, free magnification, the possibility of virtual tours inside the cardiac chambers and segments as a shared 3D team experience. Superimposed holograms on intraoperative scenarios are currently introduced as surgical decision making and adjuvant tool in several surgical disciplines [71]. Provided that computational strength increases in future versions, higherresolution cardiac-cycle models could be imported into the virtual/mixed reality from dynamic imaging sources, e.g., 3D echocardiography, structure and function will conjoin [45]. The application will also solve the problem of the currently suboptimal representation of the cardiac valves [45].

Study Limitations
This report is a single-center, initial experience of 3D modeling and printing focusing on complex congenital cardiac surgical scenarios without a control group. Technical, financial and logistic limitations are represented in focusing on the most complex, mostly redo case scenarios; thus, potential selection bias may exist. The advantages of 3D modeling and printing mostly rely on qualitative opinion. No quantitative data are currently available on reduction of complications, savings in operating time. A multicentric study is also warranted to formulate a strong case for financial reimbursement. By the essence of the data acquisition methodology, registration of the cardiac valves is suboptimal, and current 3D models are static images addressing the structure rather than cardiac function.

Conclusions
3D modeling and 3D printing is a relatively new modality in congenital cardiac surgery based on multi-and interdisciplinary teamwork. It offers multiple advantages in team learning for safe surgery, education, and communication. At present, only anatomic ('type 1 ) prototypes are available in our discipline. 3D-printed models gained higher acceptance among the surgical team as they provided additional haptic information and allowed surgical emulation; thus, they significantly contributed to team learning. 3D virtual modeling advances into 4D functional models in virtual/mixed reality. In combination with bioprinting and biofabrication, 3D-printed models could represent an avenue for the creation of 'type 2 individual cardiac implants. Funding: This research received no external funding, but the production of some 3D models was supported by the Hamdan bin Rashid Al Maktoum Foundation for Distinguished Academic Performance, Dubai, UAE.
Institutional Review Board Statement: The study was conducted according to the guidelines of the Declaration of Helsinki. Ethical review and approval were waived for this study due to the fact that anatomical models classified as research models did not come into direct contact with the patients.
Informed Consent Statement: Informed consent was obtained from all subjects involved in the study.
Data Availability Statement: Data supporting reported results can be found in the hospital database and are kept with patient confidentiality, i.e., the data are not publicly available due to patient confidentiality reasons.