Table 4.

Clinical application of 3D printed models in simulation of surgical or interventional procedures.

Author	Study Design	Sample Size and Participants	Original Data Source	Application in CVD	Image Processing Software	3D Printer/ Printing Parameters	3D Printing Material	Key Findings
Fan et al. [54]	A mix of retrospective and prospective study	Retrospective group of 72 patients with LAA occlusion guided by 3D TEE imaging. Prospective group of 32 patients with LAA occlusion guided by 3D printed models.	3D transesophageal echocardiography (TEE)	3D printed models in LAA device selection and assessment of procedural safety and efficacy.	Mimics 19.0 (Materialise, Leuven, Belgium)	Objet350 Connex 3 (Stratasys). Hollow wall thickness: 1 mm. Printing resolution: 32 μm.	Agilus A30 Clear (Stratasys)	The implantation success was 100% and 93.1% in retrospective and prospective groups. The prospective group with the use of 3D printed models achieved shorter procedural time, few devices used for the procedure (p < 0.05 for all) with no procedure complications. There were no major adverse cardiac events or mortality in the prospective group at a mean follow-up of 9.4 months, while there were three major adverse events and nine deaths in the retrospective group at a mean follow-up of 3 years.
Hell et al. [55]	Cross-sectional study	Twenty-two patients underwent LAA occlusion	TEE and CT	Prediction of device size and compression for LAA occlusion	3D Slicer 5–6 h, 20 EUR	Ultimaker 2 (Ultimaker, B.V., Geldermalsen, The Netherlands)	Silicon rubber	There was 95% agreement between 3D printed model-based sizing and the finally implanted device sizes, while the agreement between CT- and TEE-based device sizes was only 77% and 45%, respectively. The 3D printed models correlated well with the prediction of device compression (p = 0.003).
Li et al. [56]	Randomized controlled trial	Twenty-one patients in the 3D printing group with 3D printed model guiding occlusion device selection. Twenty-one patients in the control group with an occlusion device selected by TEE, CT and angiography.	TEE and CT	LAA occlusion device selection.	Mimics 17.0 (Materialise, Leuven, Belgium)	NR	NR	The occlusion procedure was successful in both groups. The procedure time, contrast volume, and costs were 96.4 ± 12.5 vs. 101.2 ± 13.6 min, 22.6 ± 3.0 vs. 26.9 ± 6.2 mL, and 12,671.1 vs. 12,088.6 USD for 3D printing and control groups, respectively. The radiation dose was significantly lower in the 3D printing group than that in the control group (561.4 ± 25.3 vs. 651.6 ± 32.1 mGy, p < 0.05).
Conti et al. [57]	Case-control study	Twenty patients: 6 with LAA leak and 14 control patients without LAA leak.	CT	3D printed models in LAA size to prevent LAA leak.	ITK-SNAP0	Form 2 Desktop printer (Formlabs, Inc., Somerville, MA, USA)	NR	Compared to 3D printed models, the device sizes based on traditional imaging analysis were unestimated in 11 patients (55%), agreed with the implanted sizes in 7 (35%) and overestimated in 2 (10%) cases. Of 8 cases with an implant device of 22 mm, 75% of the 3D printed models matched the implanted size.
Goitein et al. [58]	Cross-sectional study	Twenty-nine patients underwent LAA occlusion.	TEE and CT	3D printed models to size LAA occluder. Two types of occluder were used: AMPLATZER: n = 12 WATACHMAN: n = 17	Comprehensive Cardiac Analysis (v 4.5, Philips Healthcare)	Objet (Rehovot, Israel)	TangioPlus FLX930 (Stratasys, Germany)	A high correlation was found between 3D printed models and the inserted device size for the AMPLATZER occluder (p = 0.001) but was a poor correlation between 3D printed models and the WATCHMAN device (p = 0.203).
Torres and Luccia [59]	Cross-sectional study (prospective controlled single-center trial)	Control group: 5 residents operated on 30 patients; Training group: 5 residents operated on 25 patients.	CT	Twenty-five aneurysms were 3D printed for training and simulation of EVAR procedure when compared to the traditional training approach.	TeraRecon iNtuition Unlimited (Aquarius v 4.3, TeraRecon, CA, USA)	Connex350 (Stratasys), Formlabs Form 1 and MakerFiveot	Five materials used for 3D printing: Rubber FLX930 Plastic RGD810 TangoPlus +Vero Clear Shore 60 Resin and PLA in silicone	Use of the3D models printed with flexible resin and silicone produced the best results. Patient-specific training based on 3D printed models reduced fluoroscopy time by 30% (33 vs. 48 min), total procedure time by 29% (207 vs. 292 min), the volume of the contrast medium by 25% (65 vs. 87 mL) and time for cannulation by 52% (3 vs. 6 min) when compared to the control group (p < 0.05 for all), respectively.
Karkkainen et al. [60]	Cross-section study (prospective pilot study)	22 participants: 20 trainees: Group A: 13 experience in <20 EVAR Group B: 7 experience in >20 EVAR procedures.	CT	Use of a 3D printed AAA model in 20 trainees and two experienced operators to perform EVAR simulations	Mimics (Materialise, Leuven, Belgium	Objet500 Connex3 (Stratasys). Models were printed with three layers: 3 mm, 3 mm and 1 mm for rigid inner and flexible outer layers and luminal side, respectively.	VeroClear and Agilus	The mean procedure time was 37 ± 12 min for all 22 simulations. Experienced trainees completed the simulation procedures with significantly lower time and fluoroscopy time than inexperienced trainees (p < 0.05).
Kaufamann et al. [61]	Cross-sectional study	27 interventional radiology (IR) procedures with 54 3D printed models.	CT	Fifty-four vascular models were printed with clear and transparent flexible resin for comparison of IR procedures.	Image J	Form 3 (Formlabs). Printing resolution: 0.1 mm.	Standard clear (transparent but rigid) and flexible resin (transparent and flexible).	Of the 216 measurements in the aorta and aortic branch diameters, there were no significant differences in all measurements between the original CT and clear and flexible resin models (p > 0.05). Printing accuracy was excellent for both materials (<0.5 mm). Printing success was 85.2% and 81.5% for standard clear and flexible resin, respectively.
Sheu et al. [62]	Randomized controlled trial	Forty-nine medical students were enrolled, with 26 assigned to 3D printed vascular model and 23 to the control group (commercial simulator).	CT	3D printed model training medical students in performing femoral artery (FA) access.	TeraRecon Aquarium Intuition (TeraRecon Inc., Foster City, CA, USA)	Formlabs Form 2 SLA. The model was hollowed to 0.75 mm wall thickness.	Resin	Prior to simulation, 76.9% of trainees in 3D printing and 82.6% of trainees in the control group did not feel confident performing FA access. After the simulation, both groups agreed that the model increased their confidence in performing FA by 2 Likert points (p < 0.01). The confidence increase in the 3D printing group was non-inferior to that in the control group (p < 0.001).
Goodie et al. [63]	Cross-sectional study	Thirty medical students were invited to evaluate the efficacy of 3D printed vascular models.	CT and MRI	Five aorta and vascular models were created to simulate interventional radiology (IR) procedures.	Osirix Lite Library	Ultimaker 2 and Lulzbot Taz TM	PLA and NinjaTek Cheetah TM	3D printed models served as a supplementary tool to traditional teaching for simulation and rehearsal of IR procedures.
Yoo et al. [64]	Cross-sectional study	Eighty-one surgeons or surgical trainees were presented with the 3D printed models and subsequently performed simulated surgical procedures under guided supervision.	CT and MRI	Hands-on surgical training using 3D printed models of CHD (DORV and HLHS) in simulation of congenital heart surgeries.	Mimics (Materialise, Leuven, Belgium) Average cost per model: $60	Objet Connex 260 printer. A shell thickness of 1.2–1.8 mm was added to the outer surface of the segmented model.	TangoPlus FullCure resin and VeroWhite	Fifty attendees participated in the survey after training sessions. 3D printed models were considered as acceptable quality (88%) or manageable (12%) aid in surgical practice. Further improvements were suggested, including using material more akin to human cardiac valves.
Brunner et al. [65]	Cross-sectional study	Nineteen medical students and doctors participated in the hands-on training program.	CT	Hands-on training on simulation of interventional cardiology procedures on common CHD models.	Mimics (Materialise, Leuven, Belgium)	Agilista 3200W Polyjet 3D printer	Silicone rubber	Practicing on 3D printed models led to a significant reduction in the mean fluoroscopy time. All participants gave 3D printed models very positive ratings as a training tool for the simulation of interventional cardiac procedures.
Rynino et al. [66]	Cross-sectional study	Eleven models of aortic dissection cases were printed using different materials and distributed into four groups: autoclave 121 °C sterilization, plasma sterilization, gas sterilization, 105 °C autoclave sterilization.	CT	Effect of sterilization methods on the geometric changes of the 3D printed aortic template.	3D Slicer	Raise3D Pro 2 printer (Raise3D, Irvine, CA, USA) & Form 2 (Formlabs, Somerville, MA, USA). 1.5 mm was added to the segmented aortic wall.	Polylactic acid (PLA), nylon, polypropylene (PP), polyethylene terephthalate glycol (PETG), and a rigid and flexible photopolymer resin.	3D printed models made from PLA, PETG and PP were deformed during sterilization at high temperatures (autoclave 121 °C). However, 3D-printed models made with nylon or flexible and rigid resin did not undergo filament deformities during high-temperature sterilization. All mean geometry differences were less than 0.5 mm.

Abbreviations are the same as shown in Table 1.