Page 73 - Artificial Intelligence for Computational Modeling of the Heart
P. 73
2
Implementation of a
patient-specific cardiac model
Viorel Mihalef, Tommaso Mansi, Saikiran Rapaka,
Tiziano Passerini
Siemens Healthineers, Princeton, NJ, United States
Creating a patient-specific model of the heart starts with a set of
medical images, like for instance CT or MRI, to estimate a detailed
model of the patient’s heart anatomy. For bi-ventricular simula-
tions, the resulting anatomical model comprises the geometry of
the two ventricles, the orientation of myocardial fibers, and other
spatially-varying information such as location of scarred tissue.
For detailed computations of blood flow one could also include
models of the valves, which can be provided as 3D anatomies in-
tegrated geometrically with the myocardium, or as reduced mod-
els, with 0D functional representations. Atrial models should be
included either as 3D geometric models, if needed for whole-
heart computations (e.g. atrial electrical signal propagation or 3D
atrial hemodynamics), or as reduced (0D) models. The anatomical
model is used as computational domain for three solvers: cardiac
electrophysiology, tissue biomechanics, and hemodynamics (in-
tracardiac flow as well as flow in the arteries, the atria and the
veins, applied as boundary conditions to the electro-mechanical
model). Finally, clinical parameters are computed from the simu-
lated cardiac dynamics (e.g. stroke volume, ejection fraction, etc.).
These parameters are used as targets during model personaliza-
tion, but also as ways to assess the effect of the simulation scenario
(baseline or therapy) to investigate. Fig. 2.1 illustrates the main
blocks of the model presented in this chapter. Every component
is designed independently, with data flowing between each com-
ponent as illustrated by the black arrows.
For each component, various simplifying assumptions can be
made in order to allow for efficient personalization and com-
putation. Simplification decisions are often made based on the
clinical application. For instance, studying the effect of conduc-
tion delays may be achieved with a graph-based model of cardiac
electrophysiology, whereas complex arrhythmias demand a more
complex, phenomenological if not biophysical model. Conversely,
Artificial Intelligence for Computational Modeling of the Heart 43
https://doi.org/10.1016/B978-0-12-817594-1.00012-7
Copyright © 2020 Elsevier Inc. All rights reserved.
