Manipulation of vessel geometry via surgical intervention can have unpredictable consequences with regards to blood flow due to variation in specific patient anatomy. It is therefore imperative to devise a methodology to test the efficacy of treatments prior to surgical intervention. Studies have shown that 4D flow MRI can be used for detailed in-vivo assessment of vascular hemodynamics with 3D visualization and quantification of blood flow[1]. Advances in rapid prototyping (3D printing) have provided the ability to convert 3D imaging data into real patient-specific replicas quickly and at low cost[2]. This study was conducted to introduce a novel application of 3D printing to develop an in vitro circuit for blood flow studies. Using 4D flow MRI, hemodynamic characteristics in a 3D printed model of aortic coarctation is demonstrated and compared to in vivo findings.

3D contrast-enhanced (CE) MRA and 4D flow MRI were obtained from a 41-year-old patient with bicuspid aortic valve (BAV) and coarctation of the aorta (AoCo). Flow visualization and quantification was carried out using velocity mapping (Figure 1A-C).

DICOM files containing the MR results were imported into a segmentation software (Mimics®, Materialise). A digital model of the aorta was generated based on the 3D CE-MRA data and exported as an STL file (Figure 2A). The model was combined with standardized size 1” flange fittings at both ends for integration into the flow circuit (Figure 2B). A 1:1 replica made out of Acrylonitrile-Butadiene-Styrene (ABS) was generated using a 3D printer (Makerbot Replicator® 2X, max. build size 246 x 152 x 155 mm3, layer resolution 100µm) with settings at 0.2mm layer height, 75% infill and five-layer shells (Figure 2C). The model was then combined with vinyl and standard PVC pipe tubing (Figure 2D) to allow for integration into the experimental setup.

Schematic of the flow circuit is illustrated in Figure 3. A clinical Ventricular Assist Device-VAD[4],[5] (MEDOS VAD-Pump-Chamber, MEDOS Medizintechnik AG, Stolberg, Germany; chamber size 60 mL) was used in an MR-compatible flow circuit placed directly inside a 3 Tesla MR system (TRIO, Siemens, Germany). Pulsatile VAD flow was generated using pulsatile pneumatic pressure created by a VAD control unit (MEDOS). A triggering system controlled via National Instruments® Lab View application was used to simultaneously trigger the scanner and the VAD for synchronized gating. Fluid in the circuit was tap water containing 1.08 mmol/L Gadolinium to enhance signal. 4D flow MRI was conducted with 2mm3 voxel size.

In vivo: velocity streamlines showed formation of flow jet patterns with a strong right-handed helix in both the ascending aorta (AAo) and distal to the coarctation (Figure 1B and C). Quantitative analysis revealed highest velocity through the coarctation (2.35 m/s) as well as a high velocity flow jet along the postero-lateral wall of the ascending aorta (DAo) secondary to BAV with fusion of right- and left-coronary leaflets.

In Vitro: 3D printing took approximately 5 hours for a total cost of $10US. Computer time for segmentation, 3D modeling and fitting design was about 90 minutes. 4D flow analysis of the in vitro study again included 3D flow visualization using 3D streamlines. In addition, flow quantification was performed using planes at various sections of the aortic model (Figure 4A). As demonstrated in Figures 4A-D flow characteristics inside the 3D printed model showed vortex formation in the arch and helical flow in the Dao, but no vortex in the AAo. Velocity and flow quantification showed peak velocity of 0.7 m/s and 155 ml/s respectively.

Flow through the 3D printed patient-specific model seems to strongly correlate with the in vivo 4D flow MRI findings. This is indicative of how much blood vessel geometry can affect hemodynamics. As mentioned before, flow characteristics inside the 3D printed model had vortex formation in the arch and helical flow in the Dao, but no vortex in the AAo. We believe the absence of the AAo vortex is due to lack of BAV in the 3D model. Moreover, the velocities in the 3D printed model were drastically lower than in vivo findings. We believe this is due to low stroke volume of the VAD (60ml) compared to an adult human left ventricle (80-100ml). Lastly, the model itself can be enhanced by substituting flexible (compliant) plastics for rigid ABS as well as integrating the branches (subclavian and carotids). The results were encouraging that combined with 4D flow MRI, 3D printing might have the potential to enable better in-vitro simulation for intervention planning such as graft repair and testing of post-operative blood flow characteristics, which may facilitate procedure planning and surgical simulation.