Page 234 - Artificial Intelligence for Computational Modeling of the Heart

Chapter 6 Additional clinical applications 207

Table 6.4 Results of the deep neural network for time-independent quantities on the test dataset.

Circulation Parameters MAPE [%] Pearson
correlation
Dead volume 3.87 0.9997
Time at max. elastance 0.12 0.9995
Systemic
Resistance 0.13 0.9999
Compliance 2.83 0.9895

Dead volume 13.48 0.9994
Time at max. elastance 0.10 0.9995
Pulmonary
Resistance 0.26 0.9997
Compliance 0.76 0.9985

Figure 6.12. Overall workflow of the proposed deep learning based model.

real-time computation of pressure-volume loops. The left ven-
tricular pressure-volume (PV) loop (see examples in Fig. 6.14L)
represents an efficient tool for understanding and characterizing
cardiac function. It contains information regarding stroke volume,
cardiac output, ejection fraction, myocardial contractility, cardiac
oxygen consumption, and other important measures of the heart
and the systemic circulation. For example, the extent of ventric-
ular remodeling, the degree of ventricular-arterial mismatching
[456], and the left ventricular end-diastolic pressure-volume re-
lationship [457] represent strong predictors of congestive heart
failure. Pathologies such as left ventricular hypertrophy, dilated
cardiomyopathy, aortic and mitral valve stenosis, and regurgita-
tion [137] are manifested in the PV-loop. (See Table 6.5.)