Page 232 - Artificial Intelligence for Computational Modeling of the Heart

Chapter 6 Additional clinical applications 205

• Systemic circulation: peak aortic systolic pressure, end-
diastolic aortic pressure, left ventricular end-systolic and end-
diastolic volumes, left ventricular ejection time.
• Pulmonary circulation: peak pulmonary artery systolic pres-
sure, end-diastolic pulmonary artery pressure, right ventric-
ular end-systolic and end-diastolic volumes, right ventricular
ejection time.
The personalized measures of interest determined after run-
ning the personalization are:
• Time-independent quantities:
• Systemic circulation: dead volume of the left ventricle, time
at maximum left ventricular elastance, systemic resistance,
systemic compliance, ratio of proximal to distal resistance
of systemic circulation.
• Pulmonary circulation: dead volume of the right ventricle,
time at maximum right ventricular elastance, pulmonary
resistance, pulmonary compliance, ratio of proximal to dis-
tal resistance of pulmonary circulation.
• Time-dependent quantities:
• Systemic circulation: aortic, left ventricular and left atrial
pressures, left ventricular volume, aortic flow rate, and left
ventricular pressure-volume loop.
• Pulmonary circulation: pulmonary artery, right ventricu-
lar and right atrial pressures, right ventricular volume, pul-
monary artery flow rate, and right ventricular pressure-
volume loop.
The fully automatic optimization-based calibration method
is formulated as a numerical optimization problem, the goal of
which is to find a set of parameter values (initial volume in the
close loop, left ventricular dead volume, time at maximum left
ventricular elastance, systemic resistance, systemic compliance,
ratio of proximal to distal resistance of systemic circulation, dead
volume of the right ventricle, time at maximum right ventricular
elastance, pulmonary resistance, pulmonary compliance, ratio of
proximal to distal resistance of pulmonary circulation) for which
a set of objectives is met (left ventricular average, minimum, and
maximum pressure, left ventricular minimum and maximum vol-
ume, aortic valve ejection time, right ventricular average, min-
imum, and maximum pressure, right ventricular minimum and
maximum volume, pulmonary valve ejection time). The number
of parameters to be estimated is set equal to the number of ob-
jectives, and, thus, the parameter estimation problem becomes a
problem of finding the root for a system of nonlinear equations.
To solve the system of equations, we use the dogleg trust region
method [151].