Author + information
- Dimitrios Maragiannis, MD,
- Matthew S. Jackson, MEng,
- Stephen R. Igo, BSc,
- Su Min Chang, MD,
- William A. Zoghbi, MD and
- Stephen H. Little, MD∗ ()
- ↵∗The Methodist DeBakey Heart & Vascular Center, 6550 Fannin Street, SM-677, Houston, Texas 77030
Computed tomography (CT) provides high-resolution images of the aortic valve with clear localization of calcium deposition. Three-dimensional (3D) stereolithographic printing can be used to convert these data into a physical model (1,2). We hypothesized that patient-specific, multimaterial, 3D printed models could be created from clinical CT imaging data, and these models would accurately replicate both the anatomic and functional characteristics of severe aortic valve stenosis (AS).
We retrospectively selected imaging data from a pool of patients (N = 250) who had undergone both CT and Doppler echocardiographic studies before transcatheter aortic valve replacement. Electrocardiogram (ECG)-gated mid-systolic CT DICOM (Digital Imaging and Communications in Medicine) images were imported into anatomic modeling software (Mimics X64 15.0, Materialise, Leuven, Belgium) with which the anatomic regions of interest (left ventricular outflow tract [LVOT], aortic valve, and proximal ascending aorta) were isolated and calcified regions identified. By combining the calcified and noncalcified image datasets, we then created 3D fused-material physical models of this patient-specific anatomy. Calcified anatomic regions were printed using a rigid material (VeroWhitePlus RGD835, Stratasys, Rehovot, Israel), and all soft tissue structures (noncalcified cusp segments, LVOT, and ascending aorta) were printed using a rubber-like material (Objet TangoPlus FLX930, Stratasys).
For the functional assessment of these patient-specific AS models, we then coupled each model to our pulsatile flow imaging circuit, which has been previously described (3). In brief, the circuit incorporates a pulsatile pump, arterial compliance/resistance elements, a fill reservoir, and a water bath to facilitate ultrasound imaging. Pressures proximal and distal to the aortic valve construct were measured using high-fidelity pressure catheters (Millar, Houston, Texas). In-line ultrasonic flow transducers (Transonic Systems, Ithaca, New York) assessed transvalvular stroke volume flow. An initial flow condition for the model was chosen to replicate the stroke volume recorded during the clinical echocardiogram. A multifrequency transthoracic transducer and nonimaging probe (iE33, Philips Healthcare, Andover, Massachusetts) were used for functional imaging of the 3D model. Aortic valve area (AVA) (in cm2) was calculated by the Doppler continuity method such that AVADoppler = stroke volume/time velocity integral (TVI), where TVI was assessed by continuous wave Doppler across the printed valve construct.
In total, we created 3D printed AS models for 4 different patients and performed functional testing of each model under 7 different flow conditions. Each model demonstrated accurate reproduction of the calcific deposits within the LVOT, aortic cusps, and aortic root. In addition, the shape of the orifice area at the cusp tips was qualitatively very similar in comparison to the corresponding clinical CT. Ultrasound imaging properties of the functional construct were similar in quality to the clinical study. Spectral Doppler evaluation revealed similar signal quality and replication of clinically meaningful hemodynamic values. Mean Doppler AVA was 0.65 ± 0.15 cm2 (range 0.41 to 0.87 cm2) and mean Doppler gradient was 36.1 ± 14.7 mm Hg (range 12.6 to 61.3 mm Hg), and correlated well with catheter-derived AVA and mean gradient (r = 0.975 and r = 0.976, p < 0.001, respectively). For each patient model, the AVADoppler difference between the model and clinical echocardiography study was small (range 0% to 17% difference), with some of that variation being attributable to limitations of the LVOT Doppler method in defining the true LV stroke volume. These data suggest that the geometric valve area of the model was an accurate replication of the patients’ valve area, and that the ultrasound properties of the model were sufficient to permit diagnostic quality Doppler imaging (Figure 1).
Previous studies have created models of cardiac valves, aortic root, and various congenital structural defects from clinical echocardiographic or CT data (1,4,5). However, these anatomic models were not created to be functional constructs or to permit replication of pathological hemodynamic conditions. Severe AS may be an ideal target for 3D printing because the pathological condition is widely acknowledged to be a “fixed” obstruction with relatively immobile valve cusps.
In our preliminary work, we identified that the largest measureable AVA occurred during mid-to-late systole, which was most consistently imaged by CT as the 30% phase of the R-R interval. As such, each model was fabricated from CT imaging data representing only a single moment of the cardiac cycle. Because severe AS is a clinical condition with a relatively fixed AVA, the restricted temporal resolution of CT is potentially only a minor limitation to the modeling objective.
We demonstrated that 3D printed models can replicate both the anatomic and functional properties of severe degenerative AS. These full-scale models of specific patient anatomy and valve function can be created by combining the technologies of high spatial resolution ECG-gated CT, computer-aided design software, and fused dual-material 3D printing. The development of patient-specific models that accurately replicate both anatomic and functional characteristics may have multiple near-future applications.
Please note: Supported in part by the American Heart Associationhttp://dx.doi.org/10.13039/100000968 (14GRNT19030007), St. Jude Medical Foundationhttp://dx.doi.org/10.13039/100000895, and the Dunn Foundation for Research and Education. Dr. Little has received research support from Medtronic Structural and St. Jude Medical Foundation; and consulting fees from St. Jude Medical. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- American College of Cardiology Foundation