Page 225 - Artificial Intelligence for Computational Modeling of the Heart
P. 225

198  Chapter 6 Additional clinical applications




                                             where A inlet and r inlet are the inlet area and radius respectively,
                                             and ν is the kinematic viscosity of blood.
                                         2. For each of the computed flow rates, the pressure drop  P
                                             is estimated using the Young-Tsai model. If the maximum
                                              P (corresponding to the largest Reynolds number) is larger
                                             than the physiological threshold of 60 mmHg, all the Reynolds
                                             numbers are scaled down until the maximum  P becomes
                                             60 mmHg.
                                         3. If any of the computed flow rates is larger than 32 L/min, the
                                             computation corresponding to that flow rate is not performed.

                                         6.2.2.3 Pressure drop model for aortic coarctation
                                            The pressure drop model initially proposed by Young and Tsai
                                         [438] is formulated as a function of flow rate containing a linear
                                         and a quadratic term that account for the viscous and turbulent
                                         effects respectively. In its initial form, the Young-Tsai model is for-
                                         mulated as follows:
                                                                                   2

                                                              μK v    ρp 4  A 0       2
                                                       P YT =    3 Q +   2     − 1  Q ,         (6.8)
                                                              2πr     2A    A s
                                                                 0       0
                                                                        2

                                                                L a  A 0
                                                         K v = p 1       ,                      (6.9)
                                                                r 0  A s
                                                         L a = p 2 L s + p 3 r s ,             (6.10)
                                         where, μ is the dynamic viscosity of blood, r 0 and r s are the healthy
                                         and stenosed radius respectively, A 0 and A s are the healthy and
                                         stenosed areas, L s is the stenosis length and Q is the flow rate.
                                         Finally, the factors p 1 ...p 4 are regression parameters, and were
                                         determined by fitting the model to experimental data. Young and
                                         Tsai performed the experiments on idealized, in vitro, stenosed
                                         vessel geometries by applying different flow rates. The numerical
                                         values for these parameters are given in Table 6.3.
                                            Our work aims at improving the original Young-Tsai model
                                         by specializing it for aortic flow conditions. Thus, we explored
                                         three strategies: (i) a fine-tuning of the regression parameters, (ii)
                                         adding additional terms to Eq. (6.8) and finally (iii) coupling the
                                         resulting optimized model with a deep neural network.
                                            The resulting pressure drop model was formulated based on
                                         Eq. (6.8) and three additional terms as follows:

                                                     P =  P YT + p 5  P convective + p 6 E + p 7 B,  (6.11)

                                         where, p 5 ,p 6 and p 7 are the parameters that determine the contri-
                                         bution of the additional terms to the total pressure drop.
   220   221   222   223   224   225   226   227   228   229   230