Page 68 - Artificial Intelligence for Computational Modeling of the Heart

38 Chapter 1 Multi-scale models of the heart for patient-specific simulations

solver. We end this part with a closer look at the third component,
the valve model.

Valve models and FSI
Valves, especially the atrio-ventricular ones (and any of them
when pathology is present) can introduce large flow disturbances
that need to be taken into account in certain circumstances. For
example, a proper integration of the mitral valve would induce a
posterior flow deflection, and create a natural circulation pattern
in the ventricle [184], which may not happen if the valve is not
correctly modeled. For modeling the valves one can distinguish
three approaches:
• dynamic: modeling the valve as a deformable structure and
computing its motion using FSI
• kinematic: using a kinematic model of the valve, with the ge-
ometry and movement of the leaflets prescribed from previous
measurements
• kineto-dynamic: combining the above two approaches.
The dynamic approach [88–95] requires a significant modeling
effort for the mitral valve, so that the constitutive and structural
properties of such a complex structure can be identified prop-
erly. On one hand, with an eye focused on clinical applications,
we are still far from being able to deploy patient-specific biome-
chanical mitral valve models. On another hand, FSI models have
shown their usefulness in the study of natural and prosthetic aor-
tic valves and also prosthetic mitral valves. This is because the
structure and kinematics of mechanical prosthetic valves is much
simpler than the real counterparts. Two recent review articles [96]
and [97] provide more extended summaries of the current status
on the patient-specific simulation of cardiac valves.
The kinematic approach uses pre-defined mesh sequences of
the (usually) mitral mesh opening-closing cycle, generated from
imaging sequences (usually CT or US, but MRI can also be used
with extra modeling effort). This approach is not useful if the in-
terest is to predict the motion of the leaflets themselves, but in cer-
tain clinical applications the kinematic model of the mitral valve
could be a viable approach. An example is LV thrombosis, where
the interest lies in the flow patterns in the left ventricle. Exam-
ples of using the kinematic approach can be found in [131], where
flow patterns are analyzed in presence of pathology. This can be
considered a 1-way FSI system, where the valve motion does not
respond to stress/pressure gradients.
The third approach combines the two previous types, and was
also used for some of the results reported in Chapter 2.In[98]the
authors proposed a reduced degree of freedom model of the mitral