Systematic Review on the Use of 3D-Printed Models for Planning, Training and Simulation in Vascular Surgery

The use of 3D-printed models in simulation-based training and planning for vascular surgery is gaining interest. This study aims to provide an overview of the current applications of 3D-printing technologies in vascular surgery. We performed a systematic review by searching four databases: PubMed, Web of Science, Scopus, and Cochrane Library (last search: 1 March 2024). We included studies considering the treatment of vascular stenotic/occlusive or aneurysmal diseases. We included papers that reported the outcome of applications of 3D-printed models, excluding case reports or very limited case series (≤5 printed models or tests/simulations). Finally, 22 studies were included and analyzed. Computed tomography angiography (CTA) was the primary diagnostic method used to obtain the images serving as the basis for generating the 3D-printed models. Processing the CTA data involved the use of medical imaging software; 3DSlicer (Brigham and Women’s Hospital, Harvard University, Boston, MA), ITK-Snap, and Mimics (Materialise NV, Leuven, Belgium) were the most frequently used. Autodesk Meshmixer (San Francisco, CA, USA) and 3-matic (Materialise NV, Leuven, Belgium) were the most frequently employed mesh-editing software during the post-processing phase. PolyJet™, fused deposition modeling (FDM), and stereolithography (SLA) were the most frequently employed 3D-printing technologies. Planning and training with 3D-printed models seem to enhance physicians’ confidence and performance levels by up to 40% and lead to a reduction in the procedure time and contrast volume usage to varying extents.


Introduction
Vascular surgery, including both open vascular surgery and endovascular surgery, requires a high degree of precision and skill to ensure the best possible outcomes for patients.Recent advancements in vascular surgery have focused on enhancing precision and skill through the development of smaller guidewires for navigating complex vessels with greater accuracy, and through the introduction of thinner sutures to facilitate more delicate surgical techniques.Moreover, surgeons require extensive anatomical knowledge and proficient technical skills.According to the latest guidelines from the European Society of Vascular and Endovascular Surgery, surgical simulation plays a pivotal role in skill enhancement [1].Traditional training methods like cadaveric dissection or simulators have limitations in terms of skill acquisition and cost [2].
Advances in medical technology have revolutionized the way surgeons are trained and how surgical procedures are planned and executed [3].The integration of 3D-printed Three authors independently conducted the data extraction and evaluated the quality of the studies (AC, AM, and CM).Controversies were resolved by a fourth author (PP), who did not participate in the aforementioned process.The extracted data included the first author, year of publication, study type (retrospective, prospective, or feasibility study), surgical planning and training methods, surgical performance and self-confidence of residents/surgeons, vascular disease classification, details of 3D printing (materials, imaging software, types of printing, printed models, printing time, and model costs), preprinting imaging modalities (computed tomography/magnetic resonance with or without contrast medium), anatomical area of surgical training, complications, and effectiveness of the planning (technical success).Data that could not be inferred were labeled as "not extractable" (NE) or "not reported" (NR), as appropriate.The quality of the studies was assessed using the Mixed Methods Appraisal Tool (MMAT) [6].The records were assessed independently by two reviewers (PP, CBM) using a rating of "good" (≥6 "Yes" answers), "moderate" (4-5 "Yes" answers) or "low" (1-3 "Yes" answers) quality agreed between them.We focused on the applications of 3D-printed models and their outcomes in vascular surgery.Therefore, the main aspects and outcomes considered were as follows:

•
Diagnostic imaging technique; • Image processing and post-processing software; • 3D-printing technologies and materials; • Feasibility of 3D-printing technology application in vascular surgery; • 3D-printed models in vascular training; • 3D-printed models in vascular planning.

Definitions
The planning was defined as the selection of the most suitable material and the exploration of various possible treatments and/or strategies for different vascular diseases and patients.This planning process should rely on the analysis of 3D-printed custom-made models.Training, on the other hand, was described as the simulation of interventions by residents or surgeons using these 3D models (with or without subsequent, direct application on the patient).This simulation is intended to enhance their technical skills.Both planning and training were highlighted as tools to enhance the competence and confidence of vascular surgeons.We interpreted performance as the enhancement of technical skills and self-confidence as the optimal security during operations.

Software for 3D Model Generation
The process of generating 3D patient-specific models starts with the acquisition of patients' CT images in the form of DICOM datasets (Figure 2).

Feasibility of 3D-Printing Technology for Vascular Models
Six studies focused on demonstrating the feasibility of employing 3D-printing technology to create patient-specific 3D models for vascular surgery applications in a hospital setting [7,9,10,14,16,20].Among these, four created abdominal aortic aneurysm (AAA) 3D models [7,10,14,20].One study generated 3D models of visceral artery aneurysms [9] and one study explored various vascular anatomical regions, including the carotid artery [16].All six studies detailed the process used to generate the 3D-printed vascular models.Images of the CT-scanned 3D-printed models were used to statistically calculate the difference in the size and shape of the models compared to patient CTA images using different software (reported in Table 2).One study utilized MATLAB (MathWorks, Natick, MA, USA), a mathematical computing software [9], whereas another employed Syngo.Via (Siemens Healthineers, Erlangen, Germany), a medical-imaging software known for its advanced image analysis capabilities [7].One study utilized 3-matic (Materialise NV, Leuven, Belgium), a mesh-editing software, and also performed manual analysis using Vernier calipers [10].One study optimized the process methodology to prove the feasibility of creating complex geometries and small diameter vessels [16].
Five studies employed different methods to evaluate the 3D-printed model or the designed simulator.Four studies utilized questionnaires to gather operators' opinions, assessing both the realism of the developed system and the effectiveness of the training sessions [8,12,21,25].One involved the use of a rating scale to evaluate the realism and the operator perception of the setup [27].Three studies reported specific data on the dimensional accuracy of the 3D-printed models, comparing the 3D anatomical vascular measurements with real-life data [10,25,28].Göçer et al. and Nguyen et al. confirmed the measurements with a caliper [10,28].Torres et al. compared the measurement of the total length from the lowest renal artery to the iliac bifurcation on the 3D model with intraoperative measures (obtained with vessel-sizing catheters) [25].Among the 11 studies reviewed, 5 evaluated the effects of simulation-based training on residents [12,17,18,21,25].Specifically, three studies demonstrated improvements in residents' surgical performance and self-confidence following training [12,17,21,25].Self-confidence was assessed through the completion of questionnaires or surveys [12,21,25].Performance was evaluated based on objective parameters, such as the procedure duration, calculated during simulation [12,21] or during real interventions, comparing residents trained with and without the 3D model [25].One study documented the experience of a trainee performing anastomosis on a 3D-printed AAA model [17], while another involved deploying a bifurcated aortic endograft in a 3D-printed AAA model under fluoroscopy [18], guided by a consultant in both instances.One study only demonstrated that there is a correlation between the operator's training outcome and previous experience by comparing the performance of less-and more-experienced operators [11].

3D-Printed Models in Vascular Surgery Planning
Planning using 3D-printed models has been successfully employed in various vascular procedures, as documented in 8 out of 22 studies [13,15,17,[22][23][24][25][26].In one study, 28 surgeons retrospectively reviewed 6 cases of complex aortic aneurysm, assessing that the 3D model guided changes in 20% of the surgical planning, switching from EVAR to open repair, or from off-the-shelf solutions to custom-made aortic endografts, with self-confidence increasing in 40% of cases [15].Seven studies prospectively used 3D models for preoperative planning, including a total of 203 patients [13,17,[22][23][24][25][26].One demonstrated that residents training with patient-specific 3D models before EVAR increased their perioperative self-confidence and improved the objective surgery metrics, such as reduced fluoroscopy and procedure times, lower contrast volumes used, and quicker target vessel cannulation [25].Two studies reported an increase in self-confidence [17,22], with one also reporting a significative reduction in the operating time [22], in both cases without detailing the assessment methods.In two studies, the 3D-printed models were used to accurate position the fenestrations in fenestrated-EVAR (FEVAR) [23,26].One study compared the 3D in vitro test with numerical simulation (NS), finding equivalent accuracy but a shorter FEVAR device delivery time in favor of NS [26].The other reported that 20% of fenestrated custom-made stent-grafts were modified after testing prototypes in a 3D-printed aortic model [23].In one study using both virtual and 3D-printed models, it was reported that 3D models helped redefine the surgical approach and simulate device deployments.The study suggested the complementary use of these tools to enhance the depth perception and dynamic representation [24].One article reported the experience of using 3D models in abdominal vascular surgery with a robotic approach [13].Additionally, two out of seven articles stated that the procedures achieved successful technical outcomes and were completed without complications [22,23].

Discussion
The rapid evolution of image-processing software and 3D printers has made patientspecific 3D models accessible and cost-effective tools, sparking increased interest in 3D printing for vascular surgery [21,33,34].These models have been evaluated for their potential to improve pre-operative planning and simulation in both open and endovascular procedures, aiding in complex vascular disease diagnosis and treatment pathways [23][24][25].Additionally, given the necessity of incorporating simulation-based training in residency programs and the challenge of the high costs associated with traditional simulators, there is growing exploration of 3D-printed models as tools to develop residents' surgical technical skills [1].Traditional training methods such as cadavers are non-reproducible due to the anatomical variability and limited availability; while they are useful for hands-on training, cadavers are less effective for procedural planning as they lack the ability to simulate specific patient anatomies [2].However, challenges remain in establishing standardized methodologies for both training and planning purposes [21,35,36].
The 22 selected studies provided a comprehensive overview of 3D-printing applications in vascular surgery, encompassing the technology, materials used, production times, costs, and the quality of imaging data utilized.They showed that the use of 3D-printed models is more frequent in aneurysmal pathology compared to steno-occlusive pathology.This is likely due to the current need for more planning in complex EVAR cases.Moreover, large vessels, and in particular AAA, are probably easier to print since there is no need to reproduce very narrow lumens or even occlusions as for PAD patients.In this context, patient-specific 3D-printed vascular models are used in simulations either independently or connected to a fluid pump to accurately replicate pulsatile circulation [8,10,11,17,27,29].Common 3D-printing technologies like FDM and SLA are cost-effective and primarily designed for single-material, single-color prints, offering a good balance between cost and precision.Advanced techniques like PolyJet technology can simultaneously use multiple materials and colors, resulting in superior print quality with a high geometric accuracy, but at a higher cost.The higher costs are due not only to different technologies and complexity but also to equipment and maintenance, which increases the overall cost of the printing services.
Additionally, the cost of 3D-printed models is closely related to the choice of materials, which depends on the anatomical features being reproduced.PolyJet technology uses polyurethane-based resins, offering a wide range of material choices and allowing for the replication of complex anatomical features.However, these resins are more expensive compared to the polylactic acid (PLA) or acrylonitrile butadiene styrene (ABS) used in FDM, or the methacrylate-based resins used in SLA.PLA and ABS are cost-effective options for high-resolution prints with choices in terms of the flexibility, hardness, and transparency.However, they have fewer material options and generally cannot simultaneously use multiple colors or materials in a single print.Therefore, it is essential to carefully evaluate the specific needs of the project or final application to achieve an optimal balance between quality and costs.
Rigid materials for 3D models are useful for understanding the anatomy of the vessels to be treated or for performing accuracy measurements.To plan and simulate an endovascular procedure, it may be more appropriate to adopt elastic, flexible, and transparent models to allow the direct visualization of the endovascular materials, such as guidewires or catheters [7,10,20,[23][24][25]27,28].Simulation of stent-graft deployment in a nonrigid model may also provide an assessment of the behavior and deformation of the aortic wall [26].The production time depends on the type of 3D model to be printed (dimensions and materials), as well as on the type of 3D-printing technology.However, in most of the analyzed articles, it was not clear if the reported time for creating the 3D model related to only the printing time or to all of the production process.
The accuracy of 3D-printed models has proven to be critical for surgeons making clinical decisions and studies have evaluated it using various methods.Alongside the printing technology, accuracy is highly dependent on the quality and resolution of the original CT imaging, preferably with submillimetric slice thicknesses.The choice of software used in the manufacturing process also plays a significant role.Open-source software offers accessible tools for CT image processing, .STL file generation and post-processing, while commercial software suites like those produced by Materialise provide advanced and semi-automatic features specifically designed for medical use.These include sophisticated algorithms for segmentation and model optimization, potentially leading to higher-quality outcomes.
Moreover, 3D-printing technology is mostly used to train residents in endovascular procedures like EVAR, lower-extremity arterial interventions, and embolization techniques.Trainees can familiarize themselves with the numerous materials and different techniques in vascular and endovascular surgery.
It was reported that pre-operative planning was changed after the evaluation of 3Dprinted models, particularly when complex endovascular techniques such as the chimney-EVAR were required [22].The 3D-printed models were also reported to be a useful aid for modifying custom-made grafts: in some cases, the design model was modified from its original prototype based on the surgeon's instructions based upon the measurements performed on the 3D-printed model [23].The 3D models can also be helpful in reducing the dose of contrast medium, as well as the fluoroscopy and operating time [22,25].In fact, the surgeon will likely have already acquired the tactile sensations regarding stenosis or vessel tortuosity of the specific patient.In some articles, the model is tested in training under X-rays [11,17,18,20,24,28].To avoid the risks of radiation exposure, the use of transparent models is advisable, which are used by most of the authors.Transparent models can be used in association with a light-emitting diode (LED) light and a camera connected with a screen to mimic the operatory room environment and the two-dimensional fluoroscopy images [12,25,27].By adopting this method, the 3D model is projected onto the screen in two dimensions, and the result is very similar to what is obtained in the angiography suite.The pre-operative use of 3D-printed models was also deemed crucial in robotic surgery, where tactile feedback is absent [13].
Furthermore, 3D-printed patient-specific anatomical models can reproduce in detail all the anatomical structures that can identify possible variations or anomalies and sometimes aid in anticipating surgical risks.These models, by providing tactile feedback, enhancing 3D visualization, and improving understanding of anatomy, have been shown to improve residents' skills and increase their self-confidence [12,17,18,22,25].Furthermore, planning and practicing surgical procedures using patient-specific 3D-printed models has been shown to improve operative performance.This has the potential to reduce peri-operative complications, thereby enhancing patient safety [22,25].
The main limitation of this systematic review is that the included papers presented highly heterogeneous data about the materials, methods and purposes, and they were therefore challenging to compare.Data regarding the operating times are often lacking or incomplete [22].Thus, even though it was reported as an advantage of the use of 3Dprinted models, it was not possible to perform a quantitative synthesis of the operating time reduction.Despite the majority of the articles reporting a reduction in operative complications, the type of complication is rarely described [17,23,26,32].And, above all, it is not reported if any complication was possibly related to the 3D-printed model.Finally, there is currently no standard for evaluating patient-specific 3D-printed vascular models, even though all the included articles reported positive evaluation of the 3D model.
The refinement and personalization of 3D-printed models hold immense potential in the foreseeable future.Customization of 3D-printed anatomical replicas based on patientspecific data already offers the promise of tailoring training and surgical planning to each individual case, enhancing the precision in surgical procedures.Moreover, advancements in materials and printing techniques may lead to more anatomically accurate and physiologically responsive models, enabling a closer simulation of the dynamic conditions encountered during vascular surgeries [37].Further research in this area promises to not only advance surgical education but also to improve patient outcomes and safety in the field of vascular surgery.

Conclusions
Simulation-based training and planning using patient-specific 3D-printed models are gaining ground in vascular surgery, showing promising future prospects.These models, whether with or without a designed simulation setup, can be produced affordably and with high dimensional accuracy, allowing surgeons to accurately replicate real-life procedural challenges.In the foreseeable future, they have the potential to be recognized as essential tools for aiding in pre-operative planning and advancing residents' surgical technical skills.However, establishing standardized methodologies for generating and validating 3Dprinted models in terms of the accuracy and effectiveness is necessary for their integration as a clinical standard.

Figure 1 .
Figure 1.PRISMA flow diagram showing the process used to identify the included studies.PRISMA, Preferred Reporting Items for Systematic Reviews and Meta-Analyses.

Figure 1 .
Figure 1.PRISMA flow diagram showing the process used to identify the included studies.PRISMA, Preferred Reporting Items for Systematic Reviews and Meta-Analyses.

Figure 2 .
Figure 2. Entire process from CTA image acquisition to 3D model printing, reporting the most used technologies in our review for each step.

Figure 2 .
Figure 2. Entire process from CTA image acquisition to 3D model printing, reporting the most used technologies in our review for each step.

Figure 3 .
Figure 3. 3D-printing technologies and families.To note, 3D printers originally engineered for s gle-color, single-material applications (such as FDM, SLA, SLS, and ColorJet technologies) can adapted into machines capable of multi-color, multi-material printing by integrating supplement accessories.

Figure 3 .
Figure 3. 3D-printing technologies and families.To note, 3D printers originally engineered for singlecolor, single-material applications (such as FDM, SLA, SLS, and ColorJet technologies) can be adapted into machines capable of multi-color, multi-material printing by integrating supplementary accessories.

Table 1 .
Anatomical regions considered for 3D printing in the included studies.

Table 2 .
Software characteristics and utilization in 3D models' manufacturing process.

Table 3 .
The suppliers of the different printers.

Table 4 .
Specificities and characteristics of the 3D-printed model in each included study.

Table 5 .
Differences in 3D printed models' utilization in vascular surgery training.