3D printed ascending aortic simulators with physiological fidelity for surgical simulation

Abstract

Introduction

Three-dimensional (3D) printed multimaterial ascending aortic simulators were created to evaluate the ability of polyjet technology to replicate the distensibility of human aortic tissue when perfused at physiological pressures.

Methods

Simulators were developed by computer-aided design and 3D printed with a Connex3 Objet500 printer. Two geometries were compared (straight tube and idealised aortic aneurysm) with two different material variants (TangoPlus pure elastic and TangoPlus with VeroWhite embedded fibres). Under physiological pressure, β Stiffness Index was calculated comparing stiffness between our simulators and human ascending aortas. The simulators’ material properties were verified by tensile testing to measure the stiffness and energy loss of the printed geometries and composition.

Results

The simulators’ geometry had no effect on measured β Stiffness Index (p>0.05); however, β Stiffness Index increased significantly in both geometries with the addition of embedded fibres (p<0.001). The simulators with rigid embedded fibres were significantly stiffer than average patient values (41.8±17.0, p<0.001); however, exhibited values that overlapped with the top quartile range of human tissue data suggesting embedding fibres can help replicate pathological human aortic tissue. Biaxial tensile testing showed that fiber-embedded models had significantly higher stiffness and energy loss as compared with models with only elastic material for both tubular and aneurysmal geometries (stiffness: p<0.001; energy loss: p<0.001). The geometry of the aortic simulator did not statistically affect the tensile tested stiffness or energy loss (stiffness: p=0.221; energy loss: p=0.713).

Conclusion

We developed dynamic ultrasound-compatible aortic simulators capable of reproducing distensibility of real aortas under physiological pressures. Using 3D printed composites, we are able to tune the stiffness of our simulators which allows us to better represent the stiffness variation seen in human tissue. These models are a step towards achieving better simulator fidelity and have the potential to be effective tools for surgical training.

Introduction

Surgeons require a highly technical skill set that is traditionally acquired through intensive clinical training. For several years now, surgical training has included some form of medical simulation that allows a trainee to repeatedly practise techniques in a controlled environment. When compared with traditional approaches, simulation avoids the inherent risk of harming patients while resulting in improved knowledge retention, confidence and management skills of surgical trainees.^{1 2}

Establishing the cost–benefit of using high-fidelity (HF) or low-fidelity (LF) simulators in medicine is determined based on learning objectives. Motor skills literature shows that novice learners benefit from using simpler models especially when practising skills like repetitive suturing.^{3 4}

LF simulation is therefore seen when smaller, more specific tasks are required and is useful for repetitive practise by the trainee until the skill is fully grasped. A disadvantage of LF simulation models is the difficulty of the trainee to fully comprehend the concept or patient reaction seen in actual medical scenarios.⁵

In contrast to LF, HF simulation more closely reproduces real-life scenarios including the mimicking of human reactions, environmental situations and disease processes that trainees can encounter in healthcare but can navigate through without any risk to real patients. In open heart surgery, these reproductions would include tissue similarities and physiological likeness. Although HF simulation most accurately mimics real scenarios, the use of LF simulation is usually seen as the starting point for training with the progression to HF as the trainees’ skills, knowledge and training requirements become more complex, a strategy known as ‘progressive fidelity’.^{6 7}

Training for complex situations that are encountered in cardiac surgery can benefit from using HF simulators.⁸ Unfortunately, current vascular simulators remain simple with basic tubular shapes and made of synthetic materials or require the use of animal or cadaver tissue that lack pathological fidelity.^9–12 New three-dimensional (3D) printing technologies may be used to produce complex and/or patient-specific geometries quickly,¹³ although constraints exist in the variety of material properties available in commercial printing materials.

Optimal cardiovascular simulators require that physiological properties of blood vessels, such as how the tissue is deformed by the dynamics of blood flow or how it sutures and how it cuts, need to be considered.^{13 14} This is especially pertinent when training residents to put a patient on, or wean from, cardiopulmonary bypass (CPB) where the surgeon requires visual feedback from the distension and recoil of the aorta to safely cannulate and cross-clamp the aorta. In this regard, an adequate surgical simulator would be perfused with pulsatile flow and exhibit realistic pressure–diameter fidelity for a variety of pathological states. For example, ascending aortic aneurysms increase tissue stiffness when compared with normal ascending aortas caused by elastin degradation and an increased collagen synthesis.^15–17 This increased stiffness alters the deformation of aneurysmal aortas and the touch and feel of the tissue to the surgeon.

The creation of a new generation of HF dynamic aortic simulators suitable for mock flow circulation would therefore need to provide realistic deformation of the ascending aorta to enhance cardiac surgical training. This study investigates the ability of polyjet 3D printing technology to create distensible synthetic tissue with tunable material properties that can be used to mimic the physiological distensibility of human aortic tissue for surgical training. The pressure-mediated deformations of the simulators were compared with intraoperative patient echocardiography (ultrasound) biomechanical values.¹⁸ By using a typical imaging modality in the procurement of our measurements, we aim to demonstrate not only simulator fidelity but also the ability to use point-of-care imaging technologies with our simulators.

Materials and methods

Geometries and materials of the 3D printed aortic simulators

The 3D printed ascending aortic models were developed by computer-aided design with a combination of two software suites: Solidworks (Dassault Systèmes, Waltham, USA) and Rhinoceros 3D (Robert McNeel & Associates, Seattle, USA), and then 3D printed with a multimaterial Connex3 Objet500 printer (Stratasys, Eden Prairie, USA). Model geometry and material composition were each varied in two configurations resulting in four simulator variations (figure 1). The model geometries were (1) a straight tube with an inner diameter of 25 mm and length of 110 mm, and (2) an idealised aortic aneurysm, containing a sinus of Valsalva, a spherical aneurysm of 40 mm diameter and an aortic arch. Each model had a wall thickness of 2 mm. These geometries were designed based on typical aortic dimensions and wall thickness to represent both an idealised non-dilated aorta (tube) and an anatomical idealised aortic aneurysm with curvature through the aortic arch. The idealised aneurysm was designed based on average anatomical dimensions of typical aneurysmal aorta as seen in our previous studies.^18–22 The normal diameter of the ascending aorta is between 21 mm and 34 mm, with a wall thickness of 2 mm. An aortic aneurysm is represented by an aortic diameter measuring with at least a 50% increase from the original diameter.²¹

Figure 1.

Three-dimensional printed aortic simulators were designed with two geometries: an idealised aortic aneurysm (A, B) and a cylindrical tube (C, D) each with a wall thickness of 2 mm. For each model type, two material variants were investigated: pure elastic TangoPlus (A and C) or TangoPlus with embedded ridged VeroWhite fibres (B and D).

To replicate the compliant (elastin) and stiff (collagen) composite nature of the human aorta, two material variants were produced for each geometric configuration (figure 1). The first variant was homogeneous elastic TangoPlus Fullcure 930 (Stratasys, Eden Prairie, USA), while the second variant used the same TangoPlus Fullcure 930 structure with embedded rigid zigzag fibres composed of VeroWhite (Stratasys, Eden Prairie, USA). VeroWhite is a rigid opaque photopolymer with a reported modulus of elasticity (stiffness) of 2000–3000 MPa, while TangoPlus is a soft rubber-like material with an estimated modulus of elasticity of approximately 0.5 MPa (based on the values presented in this study).²³ These materials were selected because they were known to be echogenic and have previously demonstrated haptic, strength and isotropic similarity to human tissue.¹⁴ Previous work by our group has shown that the number of VeroWhite fibre layers, the distance between the fibres, as well as the pattern and shape of the fibres have an impact on the composite material stiffness.¹⁴ We therefore used a 3D printed composite containing VeroWhite fibres embedded in TangoPlus. The width of the fibres was 100 µm and the space between fibres 10 mm. Two orthogonal layers of sinusoidal fibres with an amplitude of 2 mm and frequency 1 rad/mm. Orthogonal layers prevented local anisotropy which is present if the fibres are aligned in a single axis as we have previously shown.¹⁴

Mechanical analysis by echocardiography

A water-fed pulsatile flow loop was used to mimic the physiological environment of the aorta to make echocardiographic measurements of stiffness (figure 2). In this set-up, a piston pump (Cardioflow MR 1000, Shelley Medical, Toronto, Canada) and resistance valve were used to generate a 0.5 Hz sinusoidal pressure wave within the model that oscillated between 120 mm Hg and 80 mm Hg. Pressure was recorded from a transducer (86A, TE Connectivity Measurement Specialties) that was connected to a port at the model’s inlet. All subsequent echo measurements were made with the simulators submerged in a water-filled box that permitted the transmittance of ultrasound. Furthermore, all 3D printed simulators were tested within a week of printing.

Figure 2.

Perfusion system set-up for the echocardiographic measurements of the aortic simulator. (A) A GE Vivid E95 echo machine captures a cross-section of the simulator suspended in a shallow bath of water. (B) A closed-loop piston pump generates a sinusoidal pressure wave that oscillates between 120 mm Hg and 80 mm Hg with the pressure measured by a transducer in the model’s inlet. 3D, three-dimensional.

A GE Vivid E95 echocardiographic machine (GE Healthcare, Chicago, Illinois, USA) was used to obtain a cross-sectional view of the simulator for three cycles of the pressure waveform to ensure a single non-truncated cycle. Each measurement used a 2.3/4.6 MHz cardiac probe (M5Sc) placed on the surface of the tubular model or on the belly of the aneurysm, respectively.

Analyses were performed on the cross-sectional view of the simulators following the methods described by Emmott et al. ²⁴ The calliper function in GE EchoPac (GE Healthcare, Chicago, Illinois, USA) was used to measure the maximum and minimum cross-sectional diameters (D_Max and D_Min, respectively). The β Stiffness Index was used as a surrogate measure of vessel stiffness and is defined as the following²⁵:

$${\displaystyle \beta \mathrm{S}\mathrm{t}\mathrm{i}\mathrm{f}\mathrm{f}\mathrm{n}\mathrm{e}\mathrm{s}\mathrm{s}=\mathrm{I}\mathrm{n}({\mathrm{P}}_{\mathrm{M}\mathrm{a}\mathrm{x}}/{\mathrm{P}}_{\mathrm{M}\mathrm{i}\mathrm{n}})/([{\mathrm{D}}_{\mathrm{M}\mathrm{a}\mathrm{x}}-{\mathrm{D}}_{\mathrm{M}\mathrm{i}\mathrm{n}}]);}$$ (Equation 1)

where P_Max and P_Min are the maximum and minimum pressures that were recorded by the transducer for a single period.

Mechanical analysis by biaxial tensile testing

Equi-biaxial tensile testing was used to evaluate the ex vivo material properties of aortic tissue.¹⁷ For each of the 3D printed aortic simulators, a 1.5×1.5 cm² testing square was isolated from the simulator at the equivalent location of the echocardiographic measurement. The testing squares were attached to an ElectroForce TestBench (TA Instruments, New Castle, Delaware, USA) with 4-0 hooked silk sutures in a bath of distilled water at room temperature. Each sample underwent cyclic loading–unloading to 40% strain at 0.4 mm/s for 10 preconditioning cycles followed by three data cycles at a displacement rate of 0.1 mm/s. Two mechanical indices were calculated from these data: (1) stiffness as defined as the incremental modulus evaluated at 40% strain and (2) the energy loss. Stiffness is a point measure of the rigidity of a material and describes the material’s resistance to deformation. Energy loss is a viscoelastic measurement of a material’s ability to function as an elastic capacitor to store and return elastic energy. Each of these parameters is commonly reported for human aortic tissue.15 24 26 27

Human ascending aortic aneurysm biomechanics

Informed consent was obtained from 21 patients with aortic aneurysms receiving elective aortic replacement surgery at the Royal Victoria Hospital (Montreal, Quebec, Canada). Expansion and recoil of the ascending aorta was measured by transoesophageal echo at the time of surgery and followed the methodology presented by Emmott et al. ²⁴ Blood pressure was obtained from an invasive radial artery trace. A specimen of the resected aorta was obtained and stored in normal saline at 4°C for subsequent tensile testing. From these human data, equivalent β Stiffness Index and tensile measurements were made following similar protocol. However, tensile measurements on human tissue were made in a bath of normal saline at 37°C to maintain an environment similar to the intraoperative measurements.

Statistical analysis

Data are presented as interquartile box-whisker plots with individual measurements presented as unique dots with N=5 replicates for all geometry and material configurations of the aortic simulators. Multiple comparison analyses were done using one-way or two-way analysis of variance (ANOVA) with Tukey post-hoc tests. P values of <0.05 were considered significant.

Results

This study assessed the physiological fidelity of 3D printed aortic simulators with different anatomy and material types. Two variants of ascending aortic anatomy were produced: (1) a simplified straight-tube aorta (‘tube’) and (2) an aortic aneurysm (‘aneurysm’), complete with an aortic sinus, a concentric aneurysm and curvature through the aortic arch. For each anatomical model, two variants of materials were selected to produce either a relatively compliant (TangoPlus, ‘tango’) or stiff wall (TangoPlus with embedded VeroWhite fibres, ‘tango+fibres’).

Mechanical comparison of aortic simulators by echocardiography

An ultrasound imaging modality was used to measure the relative apparent stiffness (β Stiffness Index) when the aortic simulators were perfused under physiological pressures of 120/80 mm Hg (figure 3). Intraoperatively, blood pressure is controlled and the target is normotensive; therefore, measurements were preformed under normal physiological pressures. Models with embedded VeroWhite fibres had higher β Stiffness Index values compared with the equivalent models composed of pure TangoPlus (p<0.001, two-way ANOVA) (figure 3). No statistical difference was measured in β Stiffness Index between the straight tube and aneurysmal geometries (p=0.062, two-way ANOVA). Post-hoc multiple comparisons test showed that for each geometry, there was a significance difference between the models with tango and tango+fibres (tube: p=0.020; aneurysm: p=0.001) and between the aneurysm, tango and the tube, tango+fibres (p<0.001).

Figure 3.

Mechanical comparison of pressurised three-dimensional printed aortic simulators by echocardiography-measured β Stiffness Index. Statistical analysis by two-way ANOVA: significant overall effect of material composite, p<0.001, and not significant (n.s.) overall effect of simulator geometry, p=0.06. Intergroup differences were measured using a Tukey post-hoc multiple comparisons test with significance indicated as *p<0.05, **p<0.01 and ***p<0.001. Mechanical comparison of aortic simulators by tensile testing. ANOVA, analysis of variance.

Biaxial tensile testing of specimens from each simulator was used to assess the planar mechanical properties—incremental modulus (ie, stiffness) and energy loss—for the different material composites and geometries (figure 4). Models with embedded VeroWhite fibres had significantly higher stiffness and energy loss as compared with models of pure TangoPlus for either the tubular or aneurysmal geometries (stiffness: p<0.001; energy loss: p<0.001, two-way ANOVA). The geometry of the aortic simulator did not statistically affect the stiffness or energy loss (stiffness: p=0.221; energy loss: p=0.713, two-way ANOVA). Post-hoc multiple comparisons showed that there was a significant difference between all tango and tango+fibres (p<0.001) variants for both stiffness and energy loss, regardless of the geometry of the model.

Figure 4.

Mechanical comparison of pressurised three-dimensional printed aortic simulators planar biaxial testing: (A) incremental modulus at 40% strain (ie, stiffness) and (B) energy loss. Statistical analysis by two-way ANOVA: effect of material composite, (A) p<0.0001, (B) p<0.0001; effect of simulator geometry, (A) p=0.2, (B) p= 0.7. Intergroup differences were measured using a Tukey post-hoc multiple comparisons test with significance indicated as ***p<0.001. ANOVA, analysis of variance.

Mechanical comparison of aortic simulators with human tissue

As the geometry did not alter the mechanical measurements, we proceeded with the aortic aneurysm geometry to compare our aortic simulators with human tissue. Human ascending aortic aneurysm tissue was obtained from 21 patients undergoing aortic replacement surgery at the Royal Victoria Hospital in Montreal, Quebec.

Comparison of β Stiffness Index between human tissue and the material composite variants of the aortic simulator by one-way ANOVA (p<0.001) demonstrated that the aortic simulator with pure TangoPlus had a similar β Stiffness Index to human tissue (p=0.345) (figure 5). The simulators with embedded fibres, however, were stiffer than human tissue with statistically higher β Stiffness Index values (p<0.001). However, the fibre-embedded variant of the aortic simulator exhibits data point overlap with human tissue in the latter’s top quartile range suggesting that it can replicate human tissue that is exceedingly stiff (figure 5).

Figure 5.

Mechanical comparison of three-dimensional printed aneurysmal aortic simulators and human ascending aortic aneurysmal tissue by echocardiography-measured β Stiffness Index. Statistical analysis by one-way ANOVA: effect of grouping, p=0.0002. Intergroup differences were measured using a Tukey post-hoc multiple comparisons test with significance indicated as ***p<0.001. ANOVA, analysis of variance; n.s., not significant.

Discussion

Cardiac surgery requires high-level cognitive and technical skills that are developed through intensive clinical training. Recently, medical simulation has been introduced as another form of training to hone these skills.² However, the benefit to using cardiovascular simulators is limited due to the current lack of fully immersive open cardiac simulators which mimic the complex physiological processes and the disease state of human tissue seen in cardiac surgery. Our study looked at leveraging novel 3D printing technologies to create more realistic models of the ascending aorta that can be perfused with pulsatile flow and exhibit biomechanical fidelity with human tissue.

The pressure–diameter relation of our simulators was measured using typical point-of-care echocardiography through the β Stiffness Index (equation 1) which is a typical stiffness index for human aortas.^{28 29} Simulators made of pure TangoPlus had statistically similar β Stiffness Index values to those measured on human ascending aortic aneurysms. Although statistically stiffer, simulators with embedded VeroWhite fibres demonstrated data point overlap with the stiffest 25% of human data from this cohort (figure 5). This shows that tuning the mechanical properties with the addition or removal of fibres can achieve a broad range of disease states. This is not limited to replicating the increase in stiffness associated with age or medial degeneration, as is common in ascending aortic aneurysms,¹⁷ but by changing the local fibre density, simulators should also be able to mimic the stiffening associated with calcification (porcelain aorta) and the suture line of anastomoses seen in reoperation.^{30 31}

The production of two geometric variants allowed us to test whether anatomical configuration appreciably changed the β Stiffness Index measurements. Ascending aortic aneurysms can exhibit a variety of morphologies—aneurysms, can be saccular or fusiform and can present anywhere from the aortic root to the aortic arch.³² We demonstrated that either a tubular or a complex spherical aneurysm exhibited the same β Stiffness Index measurements given the same pressure swing (figure 3), signifying that 3D printing aneurysms of different anatomical forms should not appreciably change the pressure–diameter relation in simulation under physiological pressures.

Achieving tissue-like pressure–diameter fidelity of a perfused aorta is centrally important when simulating CPB. Placing patients on CPB is universal in cardiac surgery and requires significant visual, haptic and pressure–distension feedback from the aorta to the surgeon to correctly perform this stage of surgery.³³ The stiffness of the aorta provides resistance when cannulating, cross-clamping and inserting the cardioplegia line. Mastery of these techniques avoids unnecessary tearing and bleeding that can extend the operation time and increase the risk of patient mortality/morbidity.

Tensile testing of specimens from the simulator wall was used to measure the mechanical properties of the simulators independent of the perfusion flow loop. Both tensile stiffness and energy loss were tunable in both simulator geometries by adding or removing embedded VeroWhite fibres from a TangoPlus base (figure 4). This result builds on proof of principle work we have done with planar 3D printed specimens to confirm that tunable multimaterial composites can be implemented into complex 3D geometries.¹⁴

A variety of materials have been used to replicate vascular models such as latex, polyurethane, silicone and polyvinyl alcohol.^34–37 Compliant 3D printed materials have already demonstrated their utility in medical simulation.³⁸ The overwhelming advantage to 3D printing versus traditional mould production and animal or human cadaver is the ability to rapidly print a desired structure, such as patient-specific anatomy from CT or MRI.¹³ In addition, by using a multimaterial composite, the mechanics can be selectively tuned to achieve global or regional differences in model behaviour and, if desired, can even generate material anisotropy (eg, circumferential vs longitudinal-oriented fibres). Collagen and elastin fibre orientation in the aortic wall leads to mechanical anisotropy (both stiffness and energy loss) and can change considerably with disease.^{27 39} Conversely, globally homogeneous and isotropic material-propertied tissue analogues are generally produced when preparing simulators from moulded or cured materials.^40–42 Furthermore, by introducing pulsatile flow through our 3D printed models, we are moving towards a more realistic and controlled environment where medical residents can practise complex procedures.

Conclusion

Cardiac surgery is moving towards advanced specialties that include complex aortic surgeries. Prophylactic ascending aortic repair to prevent dissection and rupture of an aneurysm is commonplace in most large cardiac centres. With this, the development of simulation materials closely mimicking aortopathies will assist trainees and experienced surgeons in improving skills and accurately refining techniques to address these issues.

In our study, we developed dynamic echo-compatible aortic simulators with the ability to reproduce pressure–diameter dynamics mimicking the properties of real aortas to better achieve physiological fidelity. Furthermore, by introducing pulsatile flow through our 3D printed models, we are moving towards a more realistic and controlled environment where clinicians can practise complex procedures. Future studies will examine the educational impact of these aortic simulators when used by surgical trainees and develop standardised surgical training modules to enhance medical simulation.

What is already known on this subject.

• Most vascular surgical simulators lack anatomical detail and/or are made from uniform materials that do not capture interpatient pathological variance in tissue material properties. Multicomponent three-dimensional (3D) printing addresses these shortcomings by providing a platform to rapidly generate anatomical models with physiological fidelity.

What this study adds.

• By using multicomponent 3D printing, we have developed simplified and anatomical simulators of the ascending aorta that can reproduce the distensibility of human aortas under physiological pressures. By selectively introducing a 3D printed composite fibre structure into the walls of our simulators, we can tune the simulators’ stiffness allowing for a better representation of stiffness variation seen in human aortic aneurysms.

Data availability statement

All data relevant to the study are included in the article or uploaded as supplemental information. Data available from primary author: Dr Ali Alakhtar at ali.alakhtar@mail.mcgill.ca (ORCD ID: 0000-0001-8326-3719).

Ethics statements

Patient consent for publication

Not required.