From patient-specific data to multiscale hemodynamics simulations: the challenge of boundary conditions.

Computer fluid dynamics is now a common tool to explore hemodynamics in patientspecific simulations, either to better understand a specific physiopathology, explore several physiological states or to create a representative hemodynamics model preoperatively to explore several virtual surgical options. In this context, the specification of boundary conditions is key because pressure and flow within the modeled domain are driven by the interplay between geometry (form) and hemodynamics boundary conditions (function). However, these boundary conditions are rarely the measured variables. We thus show how one can go from patient-specific clinical data (MRI and catheterization) to simulation input parameters, including modeling assumptions and the impact of both on simulation results. To take into account the rest of the circulation outside of the three-dimensional modeled domain, a number of reduced order models exist. We explain how Windkessel models and more involved lumped parameter models can be calibrated, and discuss their predictive aspects. The specification of boundary conditions in patient-specific simulations can also create numerical challenges for which recent developments are explained. Applications include preoperative modeling and virtual surgical options in the Fontan palliation of single ventricle congenital heart disease, under rest and exercise.