Page 94 - Artificial Intelligence for Computational Modeling of the Heart
P. 94

64  Chapter 2 Implementation of a patient-specific cardiac model




                                         using the following closed-form expression (Ω defines the com-
                                         putational domain; |Ω| is its volume ) [99]:

                                                               λ   1
                                                    φ e (x,t) =         [v(y,t) − v(x,t)]dy.   (2.13)
                                                             1 + λ |Ω|  Ω
                                            In case the extra-cellular potential is computed on a Cartesian
                                         grid, as done with the LBM method, it can be easily mapped back
                                         onto the original epicardial mesh using tri-linear interpolation.

                                         Boundary element model of torso potentials
                                            The boundary element method (BEM) is commonly used to
                                         map extracellular potentials onto a torso mesh by solving the
                                         Laplace equation ∇· R T ∇φ = 0 with a Neumann boundary condi-
                                         tion R T ∇φ · n = 0 on the torso mesh S B , and the Dirichlet bound-
                                         ary condition φ = φ e on the epicardium S H .The parameterR T
                                         denotes the tissue-dependent conductivity tensor.
                                            Green’s second identity writes     (A∇B − B∇A) · ndS =
                                                                            S

                                           (A B − B A) · dV for a volume V, its boundary surface S and
                                          V
                                         normal vector n,where A and B in this equation are scalar func-
                                         tions of position. By defining A as the product of the electric
                                         potential and the isotropic conductivity, and B as the term 1/r,
                                         where r is the vector from a particular integration point to the po-
                                         sition under investigation, Green’s second identity can be used to
                                         analyze voltages on the surface of a conducting volume [222].
                                            A personalized model that includes variable thoracic cavity
                                         conductivity would be the ideal setup for accurate ECG simu-
                                         lations. However, in a first approximation, one may assume an
                                         isotropic conductivity between epicardium and body surface.
                                         Therefore, the potential at any point x of the thoracic domain can
                                         be formulated as follows:

                                                   1       r · n      1        r · n  ∇φ e · n
                                            φ(x) =       φ b   dS B −       φ e    +         dS H .
                                                   4π       r  3     4π         r  3    r
                                                       S B               S H
                                                                                               (2.14)
                                         Here φ e denotes the previously computed extracellular potentials
                                         at the epicardium, while φ b are the unknown body surface poten-
                                         tials on the torso. If the surfaces S B (torso) and S H (epicardium)
                                         are discretized into triangular meshes, the problem can be formu-
                                         lated as the solution of a linear system of equations [222]:

                                                          P BB φ b + P BH φ e + G BH Γ H = 0,
                                                         P HB φ b + P HH φ e + G HH Γ H = 0.
   89   90   91   92   93   94   95   96   97   98   99