**4. Developing patient-specific models for research purposes**

The possibility of printing 3D patient-specific CHD models is appealing not only for important clinical applications (e.g. testing devices and procedures) but also for research purposes. Models can in fact be incorporated into experimental setups for the detailed analysis of local fluid dynamics and for overall hydrodynamic measurements. This is of particular relevance in CHD patients, as often in vivo measurements are not feasible or, if feasible, are technically challenging, e.g. measuring pressure-volume loops in single ventricle patients [17]. The use of appropriately tuned experimental systems has the advantage that parametric studies can be performed in a controlled and reproducible setting, i.e. changing one variable at a time incrementally. This can generate hydrodynamic data which can increase the understanding of complex physiology.

#### **4.1. Multi-scale hydrodynamic setups: studying single ventricle patients**

Patients with single-ventricle physiology (e.g. hypoplastic left heart syndrome, HLHS) present with a fragile physiology, as their cardiovascular system relies on a single pumping chamber supporting both the systemic and pulmonary circulations [ref ]. Hydrodynamic measurements are relevant to assess the effect of comorbities, e.g. aortic coarctation, on the fluid dynamics, as well as accurate measurements of quantities that are difficult to acquire in vivo, e.g. pressure drop in the surgically enlarged aorta in patients with palliated HLHS. To this end, a mock circulatory loop (MCL) can be constructed, whereby the cardiovascular system is experimen‐ tally summarised by appropriately tuned resistive and compliant elements. A system was built [18] using a Berlin Heart® ventricular assist device to simulate the single ventricle, and 3Dprinted patient-specific models were manufactured and 'plugged' into the MCL to perform parametric experiments. This allowed us to test different aortic morphologies in the same controllable setup, e.g. tubular aortic arch, aortic arch with a severe coarctation, or an aortic arch in which the patching performed during the Norwood procedure had resulted in a very dilated transverse arch therefore creating a size mismatch with the descending aorta. 3D models are shown in **Figure 5**. Data on pressure drop along the aorta or flow distribution (upper body, lower body, pulmonary circulation) can be obtained using catheters and flow probes. MCLs can also be modified to simulate different surgical procedures. In the case of HLHS, for example, different surgical options exist to perform the first palliative procedure, i.e. different shunts can be used to source pulmonary blood flow to the pulmonary circulation. An MCL was built simulating first-stage palliation of HLHS with a modified Blalock-Taussig shunt [18] from the innominate artery to the pulmonary bed of the circuit and then modified to simulate first-stage palliation of HLHS with a Sano shunt [19] from the ventricle itself (in the experimental setup, from the de-airing valve of the Berlin Heart®) to the pulmonary bed of the circuit. The coupling of a detailed 3D model with a circuit summarising the remainder of the circulation can be considered a multi-scale [20] in vitro modelling approach for CHD. With regard to HLHS palliative surgery and the valuable insight that can be gathered exper‐ imentally, MCLs can be modified to incorporate the effect of respiration [21] which becomes relevant when these patients receive their second and third surgeries, i.e. the Glenn procedure and the total cavopulmonary connect (or Fontan completion) [22]. A system incorporating a patient-specific 3D-printed Fontan model has indeed been realised [23].

**4. Developing patient-specific models for research purposes**

**4.1. Multi-scale hydrodynamic setups: studying single ventricle patients**

complex physiology.

128 New Trends in 3D Printing

The possibility of printing 3D patient-specific CHD models is appealing not only for important clinical applications (e.g. testing devices and procedures) but also for research purposes. Models can in fact be incorporated into experimental setups for the detailed analysis of local fluid dynamics and for overall hydrodynamic measurements. This is of particular relevance in CHD patients, as often in vivo measurements are not feasible or, if feasible, are technically challenging, e.g. measuring pressure-volume loops in single ventricle patients [17]. The use of appropriately tuned experimental systems has the advantage that parametric studies can be performed in a controlled and reproducible setting, i.e. changing one variable at a time incrementally. This can generate hydrodynamic data which can increase the understanding of

Patients with single-ventricle physiology (e.g. hypoplastic left heart syndrome, HLHS) present with a fragile physiology, as their cardiovascular system relies on a single pumping chamber supporting both the systemic and pulmonary circulations [ref ]. Hydrodynamic measurements are relevant to assess the effect of comorbities, e.g. aortic coarctation, on the fluid dynamics, as well as accurate measurements of quantities that are difficult to acquire in vivo, e.g. pressure drop in the surgically enlarged aorta in patients with palliated HLHS. To this end, a mock circulatory loop (MCL) can be constructed, whereby the cardiovascular system is experimen‐ tally summarised by appropriately tuned resistive and compliant elements. A system was built [18] using a Berlin Heart® ventricular assist device to simulate the single ventricle, and 3Dprinted patient-specific models were manufactured and 'plugged' into the MCL to perform parametric experiments. This allowed us to test different aortic morphologies in the same controllable setup, e.g. tubular aortic arch, aortic arch with a severe coarctation, or an aortic arch in which the patching performed during the Norwood procedure had resulted in a very dilated transverse arch therefore creating a size mismatch with the descending aorta. 3D models are shown in **Figure 5**. Data on pressure drop along the aorta or flow distribution (upper body, lower body, pulmonary circulation) can be obtained using catheters and flow probes. MCLs can also be modified to simulate different surgical procedures. In the case of HLHS, for example, different surgical options exist to perform the first palliative procedure, i.e. different shunts can be used to source pulmonary blood flow to the pulmonary circulation. An MCL was built simulating first-stage palliation of HLHS with a modified Blalock-Taussig shunt [18] from the innominate artery to the pulmonary bed of the circuit and then modified to simulate first-stage palliation of HLHS with a Sano shunt [19] from the ventricle itself (in the experimental setup, from the de-airing valve of the Berlin Heart®) to the pulmonary bed of the circuit. The coupling of a detailed 3D model with a circuit summarising the remainder of the circulation can be considered a multi-scale [20] in vitro modelling approach for CHD. With regard to HLHS palliative surgery and the valuable insight that can be gathered exper‐ imentally, MCLs can be modified to incorporate the effect of respiration [21] which becomes relevant when these patients receive their second and third surgeries, i.e. the Glenn procedure

**Figure 5.** Model of the aortic arch of a baby with hypoplastic left heart syndrome following the Norwood operation and including a visible aortic coarctation. The 3D volume (left) is modified for the purpose of carrying out hydrody‐ namic experiments, hence including funnelled attachments to the MCL and ports for pressure measurements along the aorta. The volume is 3D printed with rigid resin (centre) and inserted in the MCL (right), which is here represented with its electrical equivalents, including a Berlin Heart® device to simulate the single ventricle.

#### **4.2. Visualisation studies and validation of computational models**

An MCL which incorporates 3D-printed anatomical models can be utilised to two additional ends: (a) serving the purpose of validation test beds for computational models and (b) generating detailed information on local fluid dynamics by means of visualisation experi‐ ments. In the first case, computational models have been introduced as a powerful resource for generating knowledge in CHD and possibly aiding in the decision-making process [24], but, in order for numerical models to be reliable, they require thorough validation against 'real world data' [25] which can be obtained in vitro. In the second case, the MCL can be modified to be compatible with different visualisation techniques, e.g. particle image velocimetry [23] or magnetic resonance imaging, thereby generating detailed visualisation data. A case study of CHD, namely transposition of the great arteries (TGA) repaired with the arterial switch procedure (ASO), was modelled for both purposes [26]. In this study, an MCL was built in order to be compatible with CMR image acquisition, and a 3D patient-specific model was inserted in the MCL for visualisation purposes. The model was reconstructed from CMR data of a patient (16-year-old male) with repaired TGA with ASO and Lecompte manoeuvre, showing typical features of TGA aortic arches (i.e. a dilated aortic root and a gothic aortic arch), thereby representing a representative test bed for studying haemodynamics in this clinical scenario. The model was 3D printed with transparent rigid resin. Visualisation data were acquired with 4D CMR flow, which generates exquisite images of flow streamlines within the volume of interest over the cardiac cycle. An age-matched control model (i.e. no CHD) was included in the study for comparison purposes. Furthermore, the model was replicated in silico in a multi-scale model including the same 3D domain (i.e. TGA or control aortic arch) and a lumped parameter network summarising the remainder of the circulation. Qualitative and quantitative comparison of flow and pressure data was carried out in order to validate the computational model, demonstrating good agreement between the two for both TGA and control scenarios. An example for this modelling paradigm is shown in **Figure 6**.

**Figure 6.** A model of the aortic arch 3D printed for visualization CMR experiments (top) and meshed for computation‐ al simulations (bottom). The images show example of flow-velocity streamlines from both CMR and computational fluid dynamics at the same time points through the cardiac cycle, showing good qualitative agreement as well as agreement of velocity magnitudes. Further quantitative comparison was carried out in this case study to show the goodness of the agreement between computational and experimental data (i.e. validation of the computational model).

#### **4.3. Testing devices: the need for compliant 3D models**

Experimental models can not only incorporate 3D-printed models but can also be used to test devices, e.g. a stent deployed into the 3D model itself. The system can, as per 4.2, be not only valuable for testing novel devices but also for generating validation data for computational model of the device itself. From a 3D printing perspective, in this case, it is interesting to consider what options are available in terms of 3D printing materials. While for other experi‐ ments or other applications and conventional materials such as SLA resins might be appro‐ priate, in the case of deploying a device inside the 3D model, it may be desirable to implement the realistic compliance of blood vessels, allowing the device to adapt well to the anatomical implantation site. With advances in 3D printing technology, models such as silicone are becoming compatible with the printers, thus opening interesting avenues of research. One of the first 3D printing compatible compliant materials was a commercially available compound (TangoPlus FullCure) [27]. A study was carried out to evaluate the range of distensibility that can be implemented by models of varying thickness printed with such materials. The study showed that the material is suitable for implementing the distensibility of different arteries, in a range of thicknesses 0.7–1.5 mm. However, limitations of the materials include its fragility (i.e. tearing under pressure) and cost (more expensive that printing the same model in a rigid resin or nylon). The models for the compliance tests were 3D-printed segments of the de‐ scending aorta of a healthy volunteer, approximating a cylinder for ease of testing. Other anatomies were also 3D printed with the compliant compound. For instance, a study was carried out to test the implantation of a PPVI stent (see Section 3.1) in a patient-specific implantation site. By testing the device in a compliant model, it was possible to appreciate the way in which the stent adapted to the shape of the implantation site and deformed according to the anatomy (**Figure 7**). As per the validation framework (see Section 4.2), the same setting was also reproduced in silico [28].

scenario. The model was 3D printed with transparent rigid resin. Visualisation data were acquired with 4D CMR flow, which generates exquisite images of flow streamlines within the volume of interest over the cardiac cycle. An age-matched control model (i.e. no CHD) was included in the study for comparison purposes. Furthermore, the model was replicated in silico in a multi-scale model including the same 3D domain (i.e. TGA or control aortic arch) and a lumped parameter network summarising the remainder of the circulation. Qualitative and quantitative comparison of flow and pressure data was carried out in order to validate the computational model, demonstrating good agreement between the two for both TGA and

**Figure 6.** A model of the aortic arch 3D printed for visualization CMR experiments (top) and meshed for computation‐ al simulations (bottom). The images show example of flow-velocity streamlines from both CMR and computational fluid dynamics at the same time points through the cardiac cycle, showing good qualitative agreement as well as agreement of velocity magnitudes. Further quantitative comparison was carried out in this case study to show the goodness of the agreement between computational and experimental data (i.e. validation of the computational model).

Experimental models can not only incorporate 3D-printed models but can also be used to test devices, e.g. a stent deployed into the 3D model itself. The system can, as per 4.2, be not only valuable for testing novel devices but also for generating validation data for computational model of the device itself. From a 3D printing perspective, in this case, it is interesting to consider what options are available in terms of 3D printing materials. While for other experi‐ ments or other applications and conventional materials such as SLA resins might be appro‐

**4.3. Testing devices: the need for compliant 3D models**

control scenarios. An example for this modelling paradigm is shown in **Figure 6**.

130 New Trends in 3D Printing

**Figure 7.** Reconstruction of patient-specific right ventricular outflow tract from CT data ('3D volume') which is 3D printed using a compliant rubber-like material ('compliant model'); a stent is inserted inside the 3D model for testing ('model + stent') and additional data can be acquired by replicating the 'model + stent' system computationally. Stent diameter at extremities = 40.7 mm.

Nowadays, TangoPlus FullCure is not the only option for 3D printing compliant materials for simulating arterial models. For example, another commercially available compound, namely Heart Print Flex [29], represents a valid material for 3D printing compliant cardiovascular models. Considerations of 3D printing compliant models include the wall thickness, which may represent a limit in terms of the printer resolution and in turn could restrict the range of distensibilities that can be implemented. Robustness of the models would also be desirable if these were to be used for hydrodynamic applications, e.g. testing devices in deformable models under pulsatile flow.

## **5. Facilitating doctor-patient communication**

Communication between cardiologists/surgeons and patients with CHD and/or families of children with CHD is a delicate issue and an example of communication between an expert and a non-expert subject in which the information being exchanged is both sensitive and complex [30]. Studies have shown that understanding CHD and overall knowledge of the defect can be an issue in both patients, particularly adolescents, and their families [31]. Important information, such as the name of the primary diagnosis, has been shown to be difficult to retain in some instances, which could represent an issue for reporting information to other medical professionals caring for the patient and the patient closer community. Also, knowledge of the defect and overall appreciation of the anatomy can have an impact on lifestyle adjustments, for both the patient and the carer. In this light, improvements in communication with novel tools can be highly desirable. 3D models printed with rapid prototyping technology have been advocated to be potential communication tools; however, evidence in this context is lacking. As part of the translational work at the Centre for Cardio‐ vascular Imaging, models were purposefully manufactured for communicating with patients and families. A questionnaire-based study [32], in particular, was designed to assess parental knowledge of CHD, randomising parents into two groups: a) with 3D model and b) routine consultation without model. Models of a range of CHD were manufactured in white nylon, as a neutral choice to approach the parents. Examples of 3D models from this study are shown in **Figure 8**. The study provided some interesting findings. It was noted that using 3D models during clinical consultations in the setting of cardiac transition clinic lengthened consultation time by 5 minutes on average, possibly indicating a more thorough explanation. Clinicians rated the models as 'very useful' and reported that generally parents interacted well with the models. Parents themselves greatly appreciated the models and commented positively on their features. However, short-term knowledge, quantified by open-end answers but also through the use of diagrams and keywords (**Figure 9**), did not appear to improve across the two groups. When two blinded cardiologists were shown the parental responses to the questionnaire, they were unable to identify the primary diagnosis of the child in still approximately 40% of the cases. On the contrary, feedback provided insight into some interesting features of the models from the parents' perspective. Parents reported that medical images (e.g. echocardiography images, often times shown during consultation) are 'meaningless', while a 3D model 'makes it all so much easier for the non-expert to understand'. Parents reported some degree of shock in realising the patient-specific nature of the model ('[…] seeing the model of my son's heart was quite a shock') but understood and appreciated the value of the model for communicating anatomical features of the CHD ('[…] once I was used to the idea that the model I was looking at was what my son's heart actually looked like, it was a very useful tool for the doctor to illustrate exactly what was wrong with his heart').

these were to be used for hydrodynamic applications, e.g. testing devices in deformable models

Communication between cardiologists/surgeons and patients with CHD and/or families of children with CHD is a delicate issue and an example of communication between an expert and a non-expert subject in which the information being exchanged is both sensitive and complex [30]. Studies have shown that understanding CHD and overall knowledge of the defect can be an issue in both patients, particularly adolescents, and their families [31]. Important information, such as the name of the primary diagnosis, has been shown to be difficult to retain in some instances, which could represent an issue for reporting information to other medical professionals caring for the patient and the patient closer community. Also, knowledge of the defect and overall appreciation of the anatomy can have an impact on lifestyle adjustments, for both the patient and the carer. In this light, improvements in communication with novel tools can be highly desirable. 3D models printed with rapid prototyping technology have been advocated to be potential communication tools; however, evidence in this context is lacking. As part of the translational work at the Centre for Cardio‐ vascular Imaging, models were purposefully manufactured for communicating with patients and families. A questionnaire-based study [32], in particular, was designed to assess parental knowledge of CHD, randomising parents into two groups: a) with 3D model and b) routine consultation without model. Models of a range of CHD were manufactured in white nylon, as a neutral choice to approach the parents. Examples of 3D models from this study are shown in **Figure 8**. The study provided some interesting findings. It was noted that using 3D models during clinical consultations in the setting of cardiac transition clinic lengthened consultation time by 5 minutes on average, possibly indicating a more thorough explanation. Clinicians rated the models as 'very useful' and reported that generally parents interacted well with the models. Parents themselves greatly appreciated the models and commented positively on their features. However, short-term knowledge, quantified by open-end answers but also through the use of diagrams and keywords (**Figure 9**), did not appear to improve across the two groups. When two blinded cardiologists were shown the parental responses to the questionnaire, they were unable to identify the primary diagnosis of the child in still approximately 40% of the cases. On the contrary, feedback provided insight into some interesting features of the models from the parents' perspective. Parents reported that medical images (e.g. echocardiography images, often times shown during consultation) are 'meaningless', while a 3D model 'makes it all so much easier for the non-expert to understand'. Parents reported some degree of shock in realising the patient-specific nature of the model ('[…] seeing the model of my son's heart was quite a shock') but understood and appreciated the value of the model for communicating anatomical features of the CHD ('[…] once I was used to the idea that the model I was looking at was what my son's heart actually looked like, it was a very useful tool for the doctor to

under pulsatile flow.

132 New Trends in 3D Printing

**5. Facilitating doctor-patient communication**

illustrate exactly what was wrong with his heart').

**Figure 8.** Samples of models (not to scale) that are 3D printed in white nylon for assessing parental knowledge of CHD in a questionnaire-based study. All models are reconstructed from CMR data.

**Figure 9.** Cardiologists used 3D-printed models in discussing the anatomy and the CHD with the parents and the pa‐ tients during their consultation. Parental knowledge was assessed before and after the consultation to assess possible improvements in short-term knowledge after having seen and manipulated the 3D model.

This line of work touches on important principles of patients & public involvement and engagement (PPI/E) in research. PPI, according to the INVOLVE national advisory group (part of the National Institute of Health Research), can be referred to as research that is carried out with/by the public, not about/for them, e.g. prioritising research topics, taking part in steering groups, undertaking research or disseminating research findings [33]. The importance of developing PPI is linked to the fact that research stemming from a PPI framework is likely to be more ethical and more relevant for the patients. The area of 3D printing cardiovascular models can be conducive to developing PPI, by actively involving patients' representatives in the process of improving not only model features or prioritising their use but also the com‐ munication around CHD and in turn promote a PPI framework through a true conversation with patients and their families (**Figure 10**).

**Figure 10.** Engagement activities with patients and families can be carried out by a multidisciplinary team exploring concepts such as making a 3D heart (left: patient with CHD making a 3D heart with Play Doh®). Developing a conver‐ sation with the public can in turn inform better usage of 3D models and technology for communication purposes. Right image courtesy of Stephen King.
