Mesh generation for finite element simulation of biomedical domains has emerged as a key open problem to be addressed by the scientific community. Building representative models of organ systems that can provide accurate simulations is a cross-cutting issue requiring domain expertise from both biologists and computational scientists. Often these two groups have attacked the problem from independent viewpoints. In particular, a number of software packages for automatic mesh generation have been developed from the computational fields that provide various levels of control for geometric 'quality' of the meshes they create. However, there is a divide between these geometric quality measures and the desired properties a mesh should have to achieved robust biomedical simulation results.

Our focus for this work is to help bridge the gap between these two communities by investigating mesh generation within the pipeline from acquiring physical data to analysis of simulation results. The main goal is to better understand which properties of meshes have the most impact and how varying them translates to effects on the simulation. We take an empirical point of view. Many simulations require domain-dependent meshes that are catered to the particular simulation type. While we narrowed the focus for this work to the study of electrical simulations of the heart, in particular modeling ischemia using bidomain simulations, we hope to learn lessons that can be applied to biomedical simulations on multi-material domains in a general sense.