Influence of image artifacts on image-based computer simulations of the cardiac electrophysiology.

Myocardial infarct patients have an increased risk of scar-based ventricular tachycardia. Late gadolinium enhanced magnetic resonance (MR) imaging provides the geometric extent of myocardial infarct. Computational electrophysiological models based on such images can provide a personalized prediction of the patient's tachycardia risk. In this work, the effect of respiratory slice alignment image artifacts on image-based electrophysiological simulations is investigated in two series of models. For the first series, a clinical MR image is used in which slice translations are applied to artificially induce and correct for slice misalignment. For the second series, computer simulated MR images with and without slice misalignments are created using a mechanistic anatomical phantom of the torso. From those images, personalized models are created in which electrical stimuli are applied in an attempt to induce tachycardia. The response of slice-aligned and slice-misaligned models to different interval stimuli is used to assess tachycardia risk. The presented results indicate that slice misalignments affect image-based simulation outcomes. The extent to which the assessed risk is affected is found to depend upon the geometry of the infarct area. The number of unidirectional block tachycardias varied from 1 to 3 inducible patterns depending on slice misalignment severity and, along with it, the number of tachycardia inducing stimuli locations varied from 2 to 4 from 6 different locations. For tachycardias sustained by conducting channels through the scar core, no new patterns are induced by altering the slice alignment in the corresponding image. However, it affected the assessed risk as tachycardia inducing stimuli locations varied from 1 to 5 from the 6 stimuli locations. In addition, if the conducting channel is not maintained in the image due to slice misalignments, the channel-dependent tachycardia is not inducible anymore in the image-based model.

Fluid-structure interaction study of the edge-to-edge repair technique on the mitral valve.

The effect of functional mitral regurgitation has been investigated in an anatomically sized, fluid-structure interaction mitral valve model, where simulated correction has been performed by applying: (1) edge-to-edge repair with annuloplasty and (2) edge-to-edge repair only. Initially defined in an open unstressed/corrected configuration, fluid-structure interaction simulations of diastole have been performed in a rigid ventricular volume. Comparison of the maximum principal stresses (during diastole) in the normal and repaired models has shown that the magnitude of stress in the repaired scenarios is ~200% greater. The combined edge-to-edge and annuloplasty procedure was found to spread the induced stresses across the free margin of the leaflets, whereas without annuloplasty a localised stress concentration in the region of the suture was observed. Fluid flow downstream of the corrected configurations was able to achieve the same magnitude as in the normal case, although the flow rate was impaired. The maximum flow rate was found to be reduced by 44-50% with the peak flow rate shifted from the end of the diastole in the normal case to the start in the repaired cases.

Mitral valve dynamics in structural and fluid-structure interaction models.

Modelling and simulation of heart valves is a challenging biomechanical problem due to anatomical variability, pulsatile physiological pressure loads and 3D anisotropic material behaviour. Current valvular models based on the finite element method can be divided into: those that do model the interaction between the blood and the valve (fluid-structure interaction or 'wet' models) and those that do not (structural models or 'dry' models). Here an anatomically sized model of the mitral valve has been used to compare the difference between structural and fluid-structure interaction techniques in two separately simulated scenarios: valve closure and a cardiac cycle. Using fluid-structure interaction, the valve has been modelled separately in a straight tubular volume and in a U-shaped ventricular volume, in order to analyse the difference in the coupled fluid and structural dynamics between the two geometries. The results of the structural and fluid-structure interaction models have shown that the stress distribution in the closure simulation is similar in all the models, but the magnitude and closed configuration differ. In the cardiac cycle simulation significant differences in the valvular dynamics were found between the structural and fluid-structure interaction models due to difference in applied pressure loads. Comparison of the fluid domains of the fluid-structure interaction models have shown that the ventricular geometry generates slower fluid velocity with increased vorticity compared to the tubular geometry. In conclusion, structural heart valve models are suitable for simulation of static configurations (opened or closed valves), but in order to simulate full dynamic behaviour fluid-structure interaction models are required.