3D printed polymers for transseptal puncture simulation

Abstract

Purpose

This study aims to examine the efficacy of 3D printing polymers to realistically mimic the fossa ovalis (FO) – a cardiac structure that is positioned in the interatrial septum – for simulating transseptal puncture (TP) procedures.

Design/methodology/approach

Specimen of different materials and thicknesses were prepared and tested simulating the puncturing of the FO using a 12F ID/14F OD TP catheter. Force and tenting length results were compared with literature references. Scanning electron microscope images of the specimen were acquired to observe the perforated region as well as the structure of the specimen in the unaffected area.

Findings

Significant changes in the properties of interest were registered for the same material on specimens tested at different moments after fabrication. Suturable vessel wall polyjet material offers the best characteristics to replicate FO anatomy and mimic its behavior to puncturing.

Originality/value

The originality of the work lies in the test of different commercial materials. Similar works are already present in literature but specifically designed materials are used. Demonstrating that off-the-shelf materials could be viable for FO specimens would simplify the design of realistic TP simulators in the future.

Citation

Buonamici, F., Stomaci, T. and Carfagni, M. (2024), "3D printed polymers for transseptal puncture simulation", Rapid Prototyping Journal, Vol. 30 No. 11, pp. 289-302. https://doi.org/10.1108/RPJ-12-2023-0428

Plain Language Summary

Percutaneous catheter-based interventions in SHD treatment present key advantages when compared to their competitors. Trans-catheter Mitral Valve Repair and LAAO, for example, are replacing traditional surgery in most situations. When access to the LA or ventricle is required from the RA, TP is the technique of choice. The learning of this procedure for new operators is mainly performed on living patients under the supervision of a mentor. Training on dedicated devices could be a suitable way to fasten the learning curve of this procedure in a risk-free environment; at the moment several training devices and simulators for TP exist as literature prototypes or commercial tools. However, the haptic feedback for the surgeon for most of these devices is left to the qualitative sphere. The scope of this article is to investigate if commercial available, off-the-shelf materials for additive manufacturing could be suitable to produce FO simulacra which could convey a quantitatively correct force feedback to the operator when used in a TP training device. To this purpose, specimen in different combinations of materials and thickness have been tested simulating the puncturing of the FO. Puncturing force and tenting length results have been compared with literature references.

1. Introduction

The current trends in structural heart disease (SHD) intervention are directed toward percutaneous, catheter-based techniques, which are less expensive and present a swifter and easier recovery with respect to traditional surgical procedures (Boukhris et al., 2020; Ghiasvand et al., 2023; Tabata et al., 2019). Transseptal puncture (TP) is a noninvasive, percutaneous medical procedure during which a catheter is inserted in the heart’s left atrium (LA) from the right atrium (RA) by puncturing through the fossa ovalis (FO), a region of the inter atrial septum (IAS) characterized by reduced thickness (Almendarez et al., 2022; Klimek-Piotrowska et al., 2016; Morais et al., 2017; O’Brien et al., 2017; Russo et al., 2019). TP is used in several procedures that share the LA as region of interest, for example, left atrial appendage occlusion (LAAO), transcatheter mitral valve repair, pulmonary vein isolation (Alkhouli et al., 2016). The preferred access to the patient’s venous circulation system is the femoral vein at the groin, although examples of access from the superior vena cava are presented in the literature (Kato et al., 2017).

The TP kit is usually composed of a transseptal sheath, an introducer and of a transseptal needle. Different shapes of the terminal part of the sheath, as shown in Figure 1, have been developed to address different atrial features and clinical needs (Alkhouli et al., 2016). Each shape empowers the surgeon with a specific trajectory to reach different regions of the RA and FO with different orientations of the needle and other instruments conveyed through the sheath.

Catheter guidance is realized through several different real-time imaging techniques: fluoroscopy, intra cardiac echography (ICE) or trans-esophageal echography (TEE) are all used to assess the correct catheter path (Morais et al., 2017; Radinovic et al., 2016). Once the IAS is reached, the haptic feedback of the surgeon, arising from the contact between the catheter’s tip and the LA wall, supports the localization the FO. After the FO is located, the surgeon applies a delicate push, named tenting, which further helps the surgeon to gather anatomical data, and then the needle is advanced through the catheter, piercing the FO. The TP procedure is hence completed by advancing the dilatator and sheath through the initial puncture. Figure 2 offers a schematic representation of the procedure’s result.

According to literature data, the TP learning curve requires >50 cases to achieve acceptable efficiency (Quinn et al., 2021). Training sessions on actual patients, under mentors’ supervision are the main teaching method, even for unexperienced operators. The procedure presents serious risks and could cause dangerous clinical complications, e.g. cardiac perforation or pericardial effusion (Salghetti et al., 2017). Moreover, apprenticeship training has become increasingly challenging because of the growing disparity in numbers between trainees and teachers, limitations to consecutive working hours for physicians (McCarthy, 2014; West et al., 2007) and the sophisticated interventional techniques.

Cadaveric and animal models have been used extensively to enhance teaching. This methodology presents several drawbacks, such as ethical and legal concerns, high costs, need for controlled harvest and conservation, non-reusability, possibility that they do not accurately represent human anatomy and pathophysiology and the need for specialist facilities (Cheung et al., 2014).

As per similar clinical procedures (Abri and von Ballmoos, 2022; Aeckersberg et al., 2019; Marzola et al., 2021; Middleton, 2017; Waran et al., 2014), training on simulacra, phantoms and mannequins could be a viable solution to accelerate the learning process of the TP, especially for novice users. The main advantages of this approach are the training in a controlled and safe environment for trainees (Mussi et al., 2020), where the same steps or complete procedure can be repeated until they are well acquired, the complete absence of danger or clinical complications for patients (Mussi et al., 2021), reduced costs and supply difficulties when compared to cadaveric or animal specimen. In addition, on-demand, specific pathologic cases can be represented on the phantoms or mannequins, confronting the trainee with increasingly difficult levels of the procedure.

The use of physical phantoms in TP instruction and training is documented in the specialized literature, as described by a recent review on the subject (Stomaci and Buonamici, 2024); moreover, commercial TP training devices are also available on the market. Examples of these devices are LAACS Station (Biomodex, Quincy, MA, USA) and ANGIO Mentor Transseptal Simulator Module (SurgicalScience Simbionix Simulators, Göteborg, Sweden).

Considering the academical research, the primary technological process for phantom creation is additive manufacturing, whether for direct fabrication or for molding creation for silicone casting, starting from standard patient diagnostic imaging. Additive manufacturing technologies are fundamental for this application, as they enable design complexity, the production of patient-specific parts and the use of digital materials able to replicate the characteristics of anatomical elements. The training experience realism is achieved through simulation of image guidance technologies (ICE, TEE or Fluoroscopy) and forces feedback for the operator; to this latter purpose, tissue-mimicking materials are used for the FO and IAS phantoms.

However, even in the most sophisticated devices (James et al., 2020; Thompson et al., 2021; Zimmermann et al., 2020, 2021), a quantitative examination of the FO mimicking materials’ mechanical properties in comparison to the cardiac tissue seems an open problem.

The purpose of the activity presented in this article is to evaluate different off-the-shelf materials, considering AM photopolymers, to investigate the optimal choice for the FO realization in TP simulators in terms of realistic puncturing force and haptic feedback for the user. Similar activities are present in literature (Bezek et al., 2020); the contribution to the state of the art of the present work lies in the testing of a wider specter of materials used in different technological processes all suitable for the creation of TP training devices. An experimental test campaign was designed and carried on, simulating the tenting and puncturing of the FO during the TP, measuring force (puncturing force [PF]) and displacement (tenting length [TL]) during the test and comparing them with literature reference values. For each of the selected materials, two different thicknesses of the specimen were investigated, with the aim of understanding which combination of material and thickness could provide the most satisfactory result. Moreover, the study has also investigated time-related property changes in the characteristics of interests of the material; this factor, never considered in literature, is significant considering the known poor stability of several 3D printing polymers and resins over time.

2. Material and methods

The materials tested are Formlabs “Elastic 50A” (E50A), “Vessel Wall” (VW) and “Suturable Vessel Wall” (SVW) materials from Stratasys J850 Anatomy printer.

Elastic 50A is an acrylic-based transparent, flexible resin for stereolithography (SLA), with a post-cured shore hardness of 50A and a UTS of 3.23 MPa. Applications of similar materials are reported in literature for the manufacturing of vascular phantoms (Alawneh et al., 2023; Stomaci et al., 2023).

VW and SVW are commercial denominations of specific combinations of materials available for Polyjet printers; they are part of Digital Anatomy™ material suite – a technology proprietary of Stratasys and implemented in their most advanced Polyjet printers (J750 and J850 systems). Polyjet is an established technology in the manufacturing of flexible anatomical phantoms for cardiac applications (Biglino et al., 2013; Hong et al., 2021; Vukicevic et al., 2022). Both materials are a combination of Agilus 30 and Tissue Matrix resins and have been developed to replicate the characteristics of specific anatomical tissues – i.e. walls of cardiovascular structures. While official characteristics of this class of materials are not disclosed by Stratasys, their shore hardness is surely higher than 30A, considering that this is the minimum hardness achievable by the combination of materials that are mixed to obtain the VW and SVW materials.

Several articles in the literature have studied flexible 3D printed materials properties. A comprehensive and recent review of the properties of flexible materials with a specific focus on biomedical applications is provided by Grab et al.(2024). The authors discuss the structure and morphology of 3D printed specimens printed in Elastic 50A and Agilus 30 thanks to the execution of scanning electron microscope (SEM) images. Agilus 30 strain rate and its correlation with build orientation were studied in Abayazid and Ghajari(2020). Its basic mechanical properties are documented by Stratasys (Stratasys Ltd, 2021) and report a tensile strength of 2.4−3.1 MPa, an elongation at break of 220%–240% and a polymerized density of 1.14–1.15g/cm³. Fractographic analyses performed through the acquisition of SEM pictures of specimens produced with Agilus 30 and tested in a compression and tensile state are presented in Tee et al.(2020).

In the present study, microscopic images of the upper surface of the fabricated specimens were acquired using a ZEISS EVO 40 SEM to investigate the characteristics of the superficial structure of the material. Specimens were prepared with a gold sputter coating applied using a Quorum Q150R ES machine. The results, depicted in Figure 3, illustrate the structural differences between specimens fabricated using SLA and those made with polyjet technology. VW and SVW specimens exhibit similar superficial patterns characterized by the repeated visible deposition of individual drops. In contrast, the surface of the E50A specimen is smoother and more uniform, demonstrating a more constant surface texture. Manufacturing defects were not observed for SVW and VW, while some defects were visible on the observed E50A specimens in the shape of circular craters with 0.4–0.6 mm diameter. Such defects, visible in Figure 4, are caused by uncured resin covering the part at the end of fabrication, which occasionally leads to the inclusion of air pockets; however, they do not affect the structural integrity of the specimen, as they only cause a local excess of material with respect to the designed geometry; accordingly, they were considered not relevant for the present study.

IAS and FO anatomy present a great anatomical variability, which is furthermore influenced by the clinical conditions, sex and age of the patient (Qadir et al., 2019; Tomlinson et al., 2008). As a result, two reference values of 0.7 and 1 mm of thickness, described in literature (Elagha et al., 2020; Howard et al., 2015) were chosen for the specimen thickness. It should also be noted that, with the proposed materials and manufacturing technologies, preliminary tests showed that realizing a specimen or a component with a thickness below 0.7 mm would be extremely challenging; moreover, the fabrication of more complex simulators could require the integration of the FO anatomy within a full heart anatomy. Accordingly, it is important to select a thickness that can satisfy the fabrication constraints imposed by the selected 3D printing processes.

ASTM F1306-21 was taken as reference for the test design, as done in similar works (Bezek et al., 2020). A total of 15 35 mm round specimen were prepared for each combination of material and thickness (Table 1) with the corresponding fabrication technology. For each specimen, a unique ID was assigned based on the material and thickness; all the specimens of each material were realized from the same batch. Printing parameters recommended by producers were used in each case. Discs were laid flat on the build plate to reduce curling and to not introduce a preferential tearing direction during puncturing because of the material anisotropy that can be caused by layering. Polyjet specimens were fabricated using a layer thickness of 27 µm. A 100 µm layer was used for the Elastic 50A material. Standard printing and curing parameters responsible for the solidification of the resins were used in both systems, adopting standard process profiles associated with the materials. PreForm v3.32 software by Formlabs and GrabCAD Print v1.89 were used to slice and manage the printing process.

The tests were conducted using a MTS 810 (MTS, Eden Prairie, MN, USA), universal testing machine. A TP catheter (anterior curve, 12F ID/14F OD) was used to replicate the tenting and piercing of the FO; a custom-made clamping system was designed and realized through additive manufacturing to hold the specimen and the TP needle and catheter in position (Figures 5 and 6). The specimen was fixed using a plate/backplate mounting with four screws. The catheter was held in puncturing position using a bracket, which was designed considering the instrument curvature, to assure a perpendicular position between the elements. Again, plate/backplate mounting with screws solution was used. The test was performed with the whole catheter fixed in the bracket, and the hollow needle protruding in a fixed position from the sheath.

The upper sledge was moved downward at a constant speed of 25.4 mm/min, again, following ASTM F1306-21, and the force between the catheter and the specimen was measured with a U4000M (Maywood Instruments Limited, Basingstoke, England) piezoelectric 25 kgf load cell. Sampling rate was performed at 20S/s.

For each run, the initial condition to start the proper test was set by performing an initial approach of the upper sledge and monitoring the force value on the load cell. A load of 0.01 N read by the sensor was taken as initial value for the test. Each test was stopped when the catheter completely pierced the specimen.

To have an insight on how material properties could change over time, as curing resins used in Formlabs SLA and in Stratasys’s Polyjet have time-sensitive mechanical characteristics, tests were divided into three different sessions: at fabrication, after 30 days and after 45 days. All the specimens were fabricated in the three days before the first test session date. Specimens were properly stored in sealed containers with controlled levels of humidity, at room temperature and sheltered from sunlight. A total of 5 specimens out of 15 for each material were tested in each session; accordingly, in the following text, runs identified by numbers from #1 to #5 belong to the first session of tests, numbers from #6 to #10 identify the second one and numbers from #11 to #15 identify the third one. Tests were conducted at room temperature and the specimens were tested in a random order.

3. Results

Signals acquired during the entire experiment were aligned taking as reference a PF value of 0.2 N. This was deemed necessary to reduce alignment errors caused by fluctuations in the force value registered by the sensor because of vibrations. Accordingly, the results presented hereafter consider a displacement of 0 mm associated with a PF of 0.2 N. Moreover, a third-order one-dimensional median filter (1-D median filtering – MATLAB medfilt1 – MathWorks Italia, 2024) was applied to all signals to eliminate spikes. The effect caused by the filter is visible in Figure 7, where it is applied on one of the signals.

The typical shape of each signal showed a first phase where the curve is convex and increasing, followed by a second phase characterized by a linear behavior until the PF maximum is reached. In all tests, this corresponded to the moment when the catheter completely pierced the specimen. Each signal was cut where the PF reached its maximum value to remove its tail, which did not contain useful information for this study. Figure 8 describes the typical shape of the signals acquired during the experiment.

The results of the tests for all the samples are depicted in Figure 9 to provide a general overview of the specimens behavior. In Figure 9, each color represents a different type of specimen; similar colors identify specimens fabricated in the same material with a different thickness. The violet and green continuous lines mark references values taken from literature – i.e. an average PF value measured in (Howard et al., 2015) using swine hearts in puncturing tests and another value measured in Manavi et al.(2023) using lamb hearts. Vertical reference value for the TL is taken from Howard et al.(2015) and measured on human subjects. Dashed lines identify the deviation standard areas for the reference values. Figure 10, Figure 11 and Figure 12 present the results of each material. The legend of each graph reports the values of the maximum PF and the corresponding TL for each run. Dashed lines and red dots are used for the runs of the first session; continuous lines and blue dots are used for the second session; dotted lines and green dots are used for the third one.

Aggregated results describing the characteristics of the ending point of the test (i.e. complete piercing of the specimen) are presented in Table 2. Table 2 presents the mean, median, standard deviations and interquartile range values calculated for the peak PF and corresponding TL. Such values have been computed for the 1st, 2nd and 3rd sessions runs and for the all set. Each sample was tested for normality; SVW 1.0 TL (1st session) E50A 1.0 PF (1st and 2nd sessions) measurement distributions were not verified as normal, accordingly different metrics are suggested and used for subsequent analyses.

The analysis of Table 2 allows the evaluation of the statistical dispersion of the results among the three situations described in sessions 1 through 3. Time-related effects are clearly observed for the Elastic 50A material for the distributions of PF values registered in the three sessions, while in other cases such effects are more hidden. To fully characterize this aspect, independent t-tests were performed on PF and PTL values comparing the first session with the second one and the second with the third one using a standard 5% p-value threshold. The resulting values are summarized in Tables 3 and 4. Mann–Whitney U tests were used to compare samples whose normality hypothesis was rejected.

3.1 Visual analysis

To further study the puncture area, shape and characteristics of the regions of interest were analyzed with microscopic images, which are presented in Figure 13. The two main classes of materials demonstrated distinct characteristics in their perforation margins. The E50A material exhibited a clean, continuous margin created by the needle. In contrast, SVW and VW materials displayed jagged and discontinuous margins, particularly noticeable at the 1.0 mm thickness. The most striking result from the image analysis is the difference in the propagation of the puncturing margins. In the E50A material, the perforation edge tears, causing the propagation of the perforation margin into areas not directly reached by the needle. Conversely, SVW and VW materials do not exhibit tearing; instead, their margins typically form an almost complete circle with a non-cut sector acting like a hinge during needle passage. On the other hand, polyjet materials displayed a damaging effect in the immediate surroundings of the hole, characterized by a series of circumferential lacerations of limited extent (100–200 µm) because of the extreme straining of the material.

4. Discussion

This investigation has gathered interesting data, summarized by Figure 9 and further presented in Figure 10 through Figure 12, regarding the behavior of 3D printed specimens during a perforation test that aims at simulating the conditions of a real TP procedure. The most important and revealing information remains the value of the peak PF and corresponding TL, measured in each case when the specimen was completely pierced by the catheter. This information is directly interpretable and can be compared with reliable literature data. A proper measurement of the puncturing force magnitude required to pierce a human FO is not reported in the literature. Tests performed perforating swine FO specimens (Howard et al., 2015) reported a value of 1.97 N ± 0.40N. In contrast, Manavi et al.(2023) report a value of 5.48 N registered as mean value over seven tests. Among the two, data obtained on swine specimens could be considered as more reliable considering the anatomical similarities often exploited for research purposes (Kobayashi et al., 2012). This discrepancy, likely because of the use of different animal models, underscores the need for more comprehensive studies to establish a reliable reference.

Concerning the TL values measured in the experiment, Figure 14 shows a situation that is not so revealing, as most samples are characterized by a non-neglectable statistical distribution value which needs to be considered. First, like the PF, the literature offers a limited perspective on values that can be used to evaluate the quality of the obtained results. Howard et al.(2015) presents a value of 9.3 mm ± 0.3 mm evaluated on humans and a value of 9.8 ± 0.14 mm measured on swine specimens. The TL measurement, however, greatly depends on the moment considered as starting condition: such factor – which is present both in the surgical room and in the synthetic setup used in the present study – needs to be considered. Eichenlaub et al.(2020) report a FO protrusion value of 6.8 ± 2.5 mm measured intraoperatively without completing the perforation.

The second goal of the study was to gain insights into the material performance and their suitability for TP simulation over time. On this respect, results were only partially valid because the small sample size did not allow, in many cases, to confidently assert the existence of effects because of the time elapsed between manufacturing and testing. Although some t-tests or Mann–Whitney U tests produced p-values below 0.05, indicating statistically significant differences between the samples, the low statistical power in many cases suggested that these results should be interpreted with caution. This highlights the need for larger sample sizes to draw more reliable conclusions about the impact of time on the properties measured.

In the following paragraphs, results are discussed focusing on the performances of each material and addressing all the aspects documented in the previous section: PF values, TL values, aging of the material and visual analysis of the perforation margins.

As Table 2 and Figure 14 clearly show, the E50A specimens present the PF values most distant from the reference force of 1,97 N; accordingly, considering also the fact that a further reduction in thickness cannot be hypothesized because of the fabrication process, this material is certainly not suitable for the desired application. Considering the PF value reported in Manavi et al.(2023), however, E50A at 0.7 might present suitable characteristics to correctly represent the desired force. It should be noted, however, that valid results could be achieved by using VW material and tweaking the thickness to obtain a higher PF. E50A showed a significant degradation in its mechanical properties among the first two sessions; as Table 3 shows, the data collected in the three sessions describe populations of samples with different statistical descriptors. A similar behavior is maintained in all the specimens fabricated with the same material and different thicknesses. An analogous effect is observed also for the TL values, with the t-test for the E50A 1.0 mm that rejected the null hypothesis for the TL data collected in the first two sessions. More importantly, E50A specimens showed high dispersions among PF and TL values (especially in 3rd session), proving it difficult to obtain reliable performances.

As understandable from Table 2 and Figure 14, VW at a thickness of 0.7 mm and SVW at a thickness of 0.7 mm and 1.0 mm provide results comparable with the literature refence value for the PF [taking (Howard et al., 2015) as reference]. As expected, in all cases, increasing the thickness of the specimen increases the force for completing the puncture. SVW at a thickness of 0.7 mm presents a mean peak PF across all the samples of 1.77 N with a standard deviation of 0.27 N. The same material, with a thickness of 1.0 mm has registered a mean peak PF value of 2.15 ± 0.33N. The VW at 0.7 mm thickness measured a mean peak PF value of 2.20 ± 0.40N. Accordingly, considering the reference value provided by all three combinations of material and thickness, they might be eligible for the desired final application and are considered of interest for further tests and analyses.

Considering time-induced effects, data does not allow to draw conclusions on any specific effect when comparing 1st and 2nd sessions for polyjet materials. Conversely, the past 15 days of aging tested in the 3rd session seem to have had a heavier effect. Table 4 highlights, in general, a significant difference between the situation described at the 1-month mark and at 45 days after fabrication, imputable to the material degradation process, which could have produced an elasticity loss. Differences in the PF value measures were observed for the VW material at both 1.0 mm thickness [p = 0.0090 (1−β = 0.9050)] and 0.7 mm [p = 0.0011 (1−β = 0.9980)]. Notable differences were observed even for the PF values of the SVW material, even though the computed p-values and statistical power do not allow to draw definitive conclusions without performing additional tests. Similar results were obtained for the TL values, with overall high differences which were ultimately assessed for the SVW (both thicknesses) and VW at 0.7 mm.

It is important to note, when evaluating TL results, that the PF threshold used to synchronize signals may play a vital role in influencing the global TL values computed for all the specimens. Indeed, the choice of a higher starting PF value removes a longer starting portion of the signal and ultimately changes the value of the measured TL with respect to the ideal TL (which should be measured from the first contact point between needle and specimen). However, a smaller PF threshold carries the risk of a poor synchronization between different signals, owing to fluctuations of the signal caused by vibrations and reduces the potential effect caused by the chain of measurement. Under such considerations, this study identified SVW as the material most faithful with respect to the reference value reported in Howard et al.(2015). The statistical distribution observed in the global results (Figure 14) is evidently higher than the anatomical distribution; such aspect needs to be carefully evaluated to produce a membrane capable of performing in the desired way. With respect to SVW, such element can be attributed to time-induced variations in the material, as Table 4 highlights. Specifically, it could be beneficial to limit the time window for the application of the SVW material to a more restrictive period to avoid possible variations in the designed TL value and completely avoid the [30–45] days window.

From the analysis of the perforation margins performed thanks to the SEM acquisition depicted in Figure 13, a clear distinction between the two fabrication technologies emerges. SLA-fabricated material, represented by E50A, is prone to tearing, with perforation margins propagating beyond the direct impact area of the needle. This effect is arguably one of the causes of the higher PF confidence interval observed in Figure 14, as it was observed mainly in specimens tested in the 3rd session. In contrast, polyjet materials, such as SVW and VW, exhibit a more controlled perforation behavior; this makes SVW and VW preferable materials, allowing a consistent and predictable response to perforation.

Finally, there is abundant room for further progress in determining if the shape of the PF/TL curve might play a significant role in assuring a valid haptic response for the simulation of puncturing operations. With this respect, Figure 9 provides a good overlook at the different paths followed by the specimens to reach their final PF/TL point. The general trend followed by all signals follows the main characteristics described in Figure 8 and a good repeatability is observed.

5. Conclusions

The present study describes a testing methodology and the results obtained for the characterization of 3D printing materials to be used for the FO simulacrum in TP training phantoms. The obtained results show that, between the analyzed off-the-shelf materials, VW and SVW from Stratasys J850 Digital Anatomy Printer present characteristics that make them eligible for use in the considered application; indeed, their PF results are in line with the literature reference. While the present study has considered only a couple of plausible thicknesses for FO simulation, fine tuning of PF and PF/TL curve trend could be achieved by tweaking the thickness value or even making it not constant across the section of the specimen.

The design of specimens characterized by a non-constant thickness could allow changing the resulting combinations of PF/TL; for example, a thicker central area could provide additional resistance to the perforation while contributing minimally to the elasticity of the specimen and, ultimately, to the TL. This approach could be a valid strategy to overcome the limitations imposed by the material and reduce the variability of the properties of interest, leveraging the design complexity allowed by an additive manufacturing process.

The experiment has hence proved that fabrication time and storing conditions of components fabricated with certain materials is an important factor that should be considered relevant when designing components that need to perform puncturing operations with strict tolerances in terms of required force. However, for the specific case considered, the degradation of the mechanical properties observed between the first and second test session does not prejudice entirely the capability of the most performing materials to produce a puncturing force in the desired range.

The main limitations of this study are the reduced number of materials tested and the use of only a single catheter size (anterior curve, 12F ID/14F OD) for the tests. One of the future steps of this research will involve testing more materials to widen the choice of technologies that could be used to manufacture FO simulacra for TP simulators. Additionally, it should be noted that the investigation into the relationship between time and measured properties yielded only partially valid results. A larger number of samples would have allowed us to more confidently assert the existence of time-related effects between the fabrication and testing phases. The limited sample size hindered our ability to conclusively determine the impact of time on the measured properties. Future works should address this aspect conclusively if required by the final application. It should be noted that this aspect is not vital for the final application, as the potential negative effect introduced by the aging of the material could be mitigated recurring to a shorter shelf life of the fabricated components.

An interesting direction could be to perform advanced analyses on the shape of the signals acquired during the experiment: the force feedback haptically registered by the surgeon while performing the procedure depends not only on the maximum PF and corresponding TL but also on the previous steps of the procedure. Indeed, the surgeon collects information on the patient’s anatomy from different sources during the real procedure: force feedbacks obtained pushing against different areas of the FO and surrounding septum (operation that is typically used to identify the FO), evaluating the loads perceived when tenting the FO (operation used to confirm FO identification and to gain intel on the specific position to puncture). On this aspect, the related literature does not offer any data useful to further deepen the analysis, and a proper methodology to evaluate this aspect needs to be drafted. Further studies could allow the design of task simulators capable of adequately representing a specific characteristic of the patient (e.g. a fibrous, thick interatrial septum) exploiting the gathered data to perform informed choices on the simulator’s design parameters (material, thickness).

As previously mentioned, TP procedure relies on the physician haptic feedback to be performed. The sensitivity of the human hand to loads in the 1–5 N range is reported in literature (Wheat et al., 2004), suggesting that differences in the perceived tenting force might alter the realism of the training experience. To further investigate this topic and to validate the material selection, the haptic feedback of the proposed materials will be evaluated with experienced TP operators through tailored experimental activities.

Figures

Figure 1

TP kit

Note: In detail – Swartz SL transseptal sheaths (a); Abbott transseptal kit (b); needle radiofrequency generator transseptal needle (c); Transseptal Crosser kit with different distal shapes (D,E). From (Alkhouli et al., 2016).

Figure 2

Transseptal puncture

Figure 3

SEM 50x images acquired on the puncturing surface of the fabricated specimens

Figure 4

Superficial defects of the E50A specimens

Figure 5

Specimen clamping system

Figure 6

Clamping system mounted on the test machine

Figure 7

Comparison of original (dark blue) and filtered signal (blue)

Figure 8

Typical signal acquired in the experiment and main phases

Figure 9

Entire set of signals acquired in the three sessions

Figure 10

Suturable vessel wall – 1.0 mm (left) and suturable vessel wall – 0.7 mm (right)

Figure 11

Vessel wall – 1.0 mm (left) and vessel wall – 0.7 mm (right)

Figure 12

Elastic 50A – 1.0 mm (left) and elastic 50A – 0.7 mm (right)

Figure 13

SEM images acquired at 100x magnifications of the puncturing region of the specimens

Figure 14

PF and PTL overall results across all 15 specimens with STD intervals

Table 1

Specimen resume

Material name	Producer	Thickness [mm]
Elastic 50A	Formlabs	0.70	1.00
Vessel wall	Stratasys	0.70	1.00
Suturable vessel wall	Stratasys	0.70	1.00

Source:

Produced by authors

Table 2

Results summary – omnia

Session	Metric	VW 1.0	VW 0.7	SVW 1.0	SVW 0.7	E50A 1.0	E50A 0.7
1st
PF [N]	Mean	3.44	2.00	2.46	1.78	7.50	6.05
	StD	0.15	0.45	0.18	0.40	0.12	0.24
	Median	3.35	2.18	2.50	1.64	7.46	6.16
	IQR	0.25	0.65	0.30	0.76	0.11	0.28
TL[mm]	Mean	14.09	11.27	10.43	8.83	13.28	12.79
	StD	0.79	1.77	0.27	1.31	0.10	0.83
	Median	14.06	11.87	10.58	8.79	13.33	12.52
	IQR	1.15	2.21	0.31	2.40	0.14	1.37
2nd
PF [N]	Mean	3.54	2.59	2.17	1.92	5.92	5.05
	StD	0.26	0.17	0.11	0.17	0.16	0.32
	Median	3.59	2.60	2.16	1.91	5.98	5.08
	IQR	0.40	0.22	0.13	0.22	0.11	0.43
TL[mm]	Mean	14.30	13.58	9.90	9.25	11.53	12.30
	StD	1.09	0.87	0.82	0.67	0.46	1.07
	Median	13.85	13.64	9.83	9.28	11.25	11.93
	IQR	1.35	0.70	1.16	0.81	0.82	1.67
3rd
PF [N]	Mean	2.91	2.00	1.84	1.61	7.27	8.00
	StD	0.31	0.21	0.31	0.13	1.83	1.94
	Median	2.90	1.97	1.72	1.58	7.11	6.98
	IQR	0.35	0.20	0.40	0.11	3.42	3.62
TL [mm]	Mean	12.64	11.15	8.13	7.77	10.93	11.90
	StD	0.77	0.80	0.86	0.61	1.45	1.59
	Median	12.44	10.96	7.74	7.79	10.64	11.15
	IQR	0.82	1.24	1.08	0.63	2.78	2.88
Global results
PF [N]	Mean	3.30	2.20	2.15	1.77	6.89	6.37
	StD	0.37	0.40	0.33	0.27	1.22	1.65
	Median	3.35	2.29	2.17	1.68	7.11	6.16
	IQR	0.57	0.56	0.38	0.40	1.51	1.18
TL [mm]	Mean	13.68	12.00	9.49	8.62	11.92	12.33
	StD	1.13	1.62	1.21	1.07	1.32	1.18
	Median	13.80	12.01	9.83	8.51	12.04	12.38
	IQR	1.48	2.56	2.02	1.51	2.08	2.24

Notes:

PF = puncturing force; TL = tenting length; normality of data was not verified for SVW 1.0 mm TL in 1st session; E50A 1.0 PF in 1st and 2nd sessions. Figures in italics were used to indicate the preferred statistical descriptor for such cases

Source: Produced by authors

Table 3

Results summary for significant differences among 1st session vs 2nd session

Results	VW 1.0 mm	VW 0.7 mm	SVW 1.0 mm	SVW 0.7 mm	E50A 1.0 mm	E50A 0.7 mm
PF (1st vs 2nd session) p-value	0.4728 (0.1593)	0.0262 (0.4391)	0.0139 (0.6180)	0.4918 (0.0782)	0.0079 (1. 000)	1.0265e-07 (0.999)
PTL (1st vs 2nd session) p-value	0.7379 (0.0659)	0.0307 (0.4424)	0.2222 (0.7777)	0.5465 (0.0727)	3.2951e-05 (1.0000)	0.4380 (0.1329)

Notes: p-values and statistical power = (independent t-tests; Mann–Whitney U tests were used for comparisons involving samples whose normality hypothesis was rejected – i.e. SVW 1.0 mm TL in 1st session, E50A 1.0 PF in 1st and 2nd sessions). Rejected null hypotheses according to a standard 5% significance level are marked in bold. PF = puncturing force; PTL = tenting length. Green cells have a statistical power measured for the test > 0.8

Source: Produced by authors

Table 4

Results summary for significant differences among 2nd session vs 3rd session

Results	VW 1.0 mm	VW 0.7 mm	SVW 1.0 mm	SVW 0.7 mm	E50A 1.0 mm	E50A 0.7 mm
PF (2nd vs 3rd session) p-value	0.0090 (0.9050)	0.0011 (0.9980)	0.0559 (0.9892)	0.0127 (0.6912)	0.5238 (0.7532)	0.0100 (1.000)
PTL (2nd vs 3rd session) p-value	0.0237 (0.5640)	0.0018 (0.9707)	0.0107 (0.8411)	0.0066 (0.8608)	0.4062 (0.4395)	0.6538 (0.0819)

Notes:

p-values and statistical power = (independent t-tests; Mann–Whitney U tests were used for comparisons involving samples whose normality hypothesis was rejected – i.e. E50A 1.0 PF 2nd session). PF = puncturing force; PTL = tenting length; Green cells have a statistical power measured for the test > 0.8

Source: Produced by authors