Thoracic aortic phantoms may be an alternative to accurately characterize flow, velocity, wall stiffness and pressure gradient in controlled experiments without the need to expose patients to risky conditions. Recently, 3D printing technology has emerged as a very innovative technique to produce patient-specific anatomical replicas with great precision. Therefore, the aim of this work was to generate a one to one replica of the aorta of a patient with a kinking and to compare the hemodynamic parameters with the ones obtained from patient's PC-MRI and cardiac catheterization.

In vivo study: It was performed in a combined MRI/Catheter interventional suite¹, equipped with a 1.5 T Intera MRI scanner and a BT Pulsera cardiac radiography unit (Philips). The patient was a 25 years old man (64 kg and 175 cm) with a repaired aortic coarctation with an end-to-end anastomosis and a tortuous arch post coarctation dilatation of the descending aorta (DAo). The MRI protocol included a breath-hold 3D CE-MRA to image the aorta (spatial resolution = 1.3x1.3x1.8 mm, TR/TE = 4.4/1.3 ms, flip angle = 40°, cardiac phases = 2). The gadolinium-based contrast agent was injected intravenously by hand at a dose of 0.2 mmol/kg (Magnevist, Berlex Laboratories). After the CE-MRA acquisition, the hemodynamic information was acquired. The flow, velocity and wall stiffness information was obtained by free-breathing 2D through-plane flow at the level of the ascending aorta (AAo) (I) and DAo after the kinking (VI) (Table 1) (spatial resolution = 1.17x1.17 mm, TR/TE = 4.7/3.1 ms, flip angle = 15°, Venc = 350 m/s, temporal phases = 80-100). Just after the flow acquisition, a 15-20 sec breath-hold was performed to simultaneously register the catheter pressures in the AAo (I) and Diaphragmatic aorta (DiaphAo) (VII).

In vitro study: The 3D CE-MRA data was used to create a geometric solid model of the aorta using the custom software CRIMSON. The aorta model and a mold were built with a 3D-printer (Leapfrog Creatr HS) using PLA filament (Poly Lactic Acid). A space of two millimeters was generated between the mold and the aorta in order to generate a uniform wall and apply a translucent silicone (Smooth-on). The kinking phantom was assembly in an acrylic box and connected to MRI compatible pulsatile pump setup^2,3. MRI were performed on a 1.5 T MR-system (Philips) using a 4-channel body coil and retrospective cardiac gating. 3D PC-MRI was acquired covering the whole phantom and 2D PC-MRI were acquired in the AAo (I) and DAo after the kinking (VI) for flow, velocity and wall stiffness analysis (acquired and reconstructed spatial resolution of 1.79x1.83x1.80 mm³ and 0.89x0.89x0.90 mm³, acquired and reconstructed temporal resolution of 52.2 ms and 35 ms, field of view of 200x200x114 mm, TR/TE of 6.5/3.8 ms, flip angle of 6.5°, Venc of 130 cm/s, 25 time frames, and 127 slices). The phantom was equipped with a catheterization unit to measure the pressure gradient along the aorta (two Arrow catheters) in the AAo (I) and the DiaphAo (VII). Analyses were performed with the commercial software GTFlow 2.2.15 (Gyrotools LCC).

The figure 1 shows the sequence to build the phantom starting with the CE-MRA in the patient and ending up with the silicone mold and the PC-MRI in the phantom. A geometry comparison can be visualized in table 1. Hemodynamic parameters of the patient and the phantom are summarized in table 2. Similar hemodynamic parameter values were achieved for the heart rate, aorta flow split, peak flow, peak velocity, diastolic pressures and wall stiffness in the AAo. Lower stroke volume and cardiac output were obtained. Parameters that were different were the systolic pressures, systolic pressure gradient, flow and pressure waveforms and the wall stiffness in the DAo. Flow and pressure curves of the patient and the phantom are depicted in figure 2. We have built a one to one replica of a patient with a kinking from MRI and cardiac catheterization data. Most hemodynamic parameters were similar between the patient and the phantom. Probably, the combination of lower cardiac output and a lower wall stiffness of the DAo in comparison with the patient limited the pressure gradient found in the phantom. The cardiac output is regulated by the flow pump, which we can not further increase; however, we expect that by increasing the viscosity of the silicone and the thickness of the wall we can further refine our silicone one to one model. We have built a one to one replica of the aorta of a patient with a kinking obtaining similar hemodynamic parameters from PC-MRI and cardiac catheterization.