Page 217 - Artificial Intelligence for Computational Modeling of the Heart

190 Chapter 6 Additional clinical applications

the trends in QRS shortening during pacing, and showed promis-
ing accuracy in terms of absolute QRS duration prediction. Ex-
tensions of this work could include evaluation on larger cohort,
prediction of additional electrophysiology and hemodynamics
biomarkers.

6.2 Aortic coarctation

6.2.1 Introduction
Coarctation of the aorta (CoA) is a congenital disease that ac-
counts for 8–11% of cardiovascular diseases [414]. It consists of
a narrowing of the aortic lumen, usually located near the aortic
arch, that obstructs the flow, causing a significant pressure drop
along the narrowing, and less blood being delivered to the lower
part of the body. If left untreated, it can cause hypertension, early
coronary artery disease, heart failure, stroke, aneurysm formation
and decreased life expectancy [415–417]. Treatment of CoA aims
at reducing the pressure drop and restoring normal flow condi-
tions to the lower part of the body. Treatment is typically per-
formed percutaneously by placing a stent or by balloon angio-
plasty, or surgically [418,419]. In older children and particularly
in teenagers and adults, catheter-based intervention is becoming
more prevalent and is typically the standard of care. Although the
occlusion ratio can be accurately measured from image data, e.g.
MRI or transthoracic echocardiography, accurate clinical decision
making requires hemodynamic assessment that can be performed
only invasively through pressure measurements [420]. Specifically,
treatment is recommended in case of a resting state peak-to-peak
trans-coarctation pressure drop of 20 mmHg or more [421]. The
most accurate technique for measuring pressure drop is intravas-
cular catherization.
Given the invasive nature of cardiac catheterization, and asso-
ciated costs and risks, efforts have been made to develop non-
invasive methods for computing the trans-CoA pressure drop
from medical images. A previously reported approach consists
in using flow velocity information extracted with Doppler ultra-
sound or phase contrast MRI (PC-MRI), in combination with fluid
dynamics equations, to estimate pressure as a function of velocity
[422–424]. This approach tends to overestimate the pressure drop,
and is not suitable for older children or adults due to the posterior
location of the descending aorta. Although PC-MRI is able to pro-
vide significantly higher quality velocity measurements [425,426],
the imaging time is typically 10–15 minutes [427], too high for a
routine clinical environment. A simple non-invasive alternative