Page 103 - Artificial Intelligence for Computational Modeling of the Heart
P. 103

Chapter 2 Implementation of a patient-specific cardiac model 73




                     ulate neighboring organs and the stiff pericardial sac. Finally, the
                     total boundary condition force is given by f b = f base  + f peri .

                     2.3.4 Putting it all together: a fast computational
                           framework for cardiac biomechanics

                        Combining all the forces derived in the previous sections and
                     distributing them to the nodes of the mesh leads to the final TLED
                     formulation. Time integration is done using the central difference
                     scheme:
                                                          1  2
                                   u(t + δt) = u(t) + δt ˙ u(t) + δt ¨ u(t).  (2.25)
                                                          2
                     Let us assume that the displacements u(t) and u(t −δt) are known,
                     and that the total nodal forces f(t) have been computed (including
                     internal and external forces). The central difference method then
                     yields a node-wise equation for computing displacements at the
                     next time step:

                                u i (t + δt) = A i f i (t) + B i u i (t) + C i u i (t − δt),  (2.26)

                                  1           2M ii         D ii    B i
                          A i =       ,  B i =    A i ,  C i =  A i −  .   (2.27)
                              D ii  +  M ii    δt 2         2δt     2
                               2δt  δt  2
                     As any explicit time integration, δt must be small enough to guar-
                     antee stability and accuracy. In particular, δt must be smaller than
                     the critical limit δt cr = L e /c,where L e is the smallest characteris-
                     tic element length in the assembly and c is the dilatational wave
                     speed of the material.

                     Description of the TLED finite elements algorithm
                        One of the advantages of TLED algorithm is that all the vari-
                     ables are referred to the initial configuration, which allows very
                     efficient updates during the main iteration loop. The algorithm
                     starts with a pre-computation stage to initialize all variables (Al-
                     gorithm 3). Before starting the main iteration loop, the solver and
                     constraints are initialized (Algorithm 4), followed by the main it-
                     eration, which consists in first calculating the net nodal force (Al-
                     gorithm 5) and then updating the displacement (Algorithm 6).

                     2.3.5 Evaluation of the TLED algorithm
                     Validation against analytical solution
                        As shown in [120], simple shear in different planes can be used
                     to verify the implementation of any constitutive law. Indeed, this
   98   99   100   101   102   103   104   105   106   107   108