A numerical framework for preprocedural prosthetic valve positioning and hemodynamic evaluation

Figures

• 

  Fig. 1. Left: volume rendering of the left heart from cardiac CT images together with the heart valve prosthesis. Right: geometrical representation of the left ventricle including papillary muscles and large-scale trabeculation of the ascending aorta and aortic heart valve. The inline figure illustrates the spatial resolution of the triangle mesh of the heart valve housing used in the simulations

• 

  Fig. 2. Eulerian (box) and Lagrangian (cardiac structure) domains. A three-element Windkessel (3WK) boundary condition was applied at the aortic outlet (red circle), while zero-pressure boundary conditions were applied at all other outer faces of the Eulerian grid. The figure also illustrates the three levels of adaptive mesh refinement, with the finest resolution located near the cardiac structures

• 

  Fig. 3. Virtual valve implantation. From left to right: baseline (in vivo), translated 6 mm ventricularly, rotated 14°, and rotated 27° around an axis lying in the aortic annular plane (orthogonal to the LVOT centerline). Baseline position shown in white; yellow surface indicates new valve position

• 

  Fig. 4. Upper panel: The prosthetic heart valve was virtually implanted using CT images for alignment. Then, the contact area between the cardiac structure and the valve—the landing zone—was extracted and used as input for an alignment algorithm to obtain the optimal rigid-body motion of the valve. Lower panel: The valve leaflets were segmented in the closed position and deformed into the open position. The leaflet position was then parameterized from closed to open, and in the flow simulation, their motion was determined by the valvular pressure gradient

• 

  Fig. 5. Volume rendering of velocity at maximum flow acceleration, peak flow rate, and maximum flow deceleration for the four different valve orientations. Regions of low velocity are not shown to improve visualization of major flow structures, and the valve leaflets have been omitted in the renderings for clarity

• 

  Fig. 6. The velocity magnitude normal to planes in the ascending aorta for the four valve orientations at peak flow rate

• 

  Fig. 7. The valvular pressure drops for the four different valve orientations (left y-axis) and valve leaflet position (open/closed, right y-axis)

• 

  Fig. 8. Time course of the power loss from viscous dissipation for the four different simulation cases. Dashed black line indicates valve leaflet position (0 = fully closed, 1 = fully open)

• 

  Fig. 9. Visual validation of boundary deformation at end-systole. Left: three-chamber (3CH) view. Right: short-axis (SAX) view. The grayscale background shows the original CT image, and the red contours represent the deformed surface geometry obtained from the registration-based motion extraction

• 

  Fig. 10. Time history of the valvular pressure drop for the four different grid resolutions. For visibility reasons, the Y-axis range is capped to 65 [mmHg], even though the maximum value of the coarse grid is at 110 [mmHg]

• 

  Fig. 11. Visualization of velocity magnitude (upper row) and out-of-plane vorticity (lower row) for four different levels of background grid resolution during maximum flow acceleration

• 

  Fig. 12. Sensitivity analysis of valve opening and closing time duration. Left: valve position as a function of time for the four evaluated time durations. Right: Resulting pressure gradient over the valve for the time durations