A primer on computational simulation in congenital heart disease for the clinician

Abstract

Interest in the application of engineering methods to problems in congenital heart disease has gained increased popularity over the past decade. The use of computational simulation to examine common clinical problems including single ventricle physiology and the associated surgical approaches, the effects of pacemaker implantation on vascular occlusion, or delineation of the biomechanical effects of implanted medical devices is now routinely appearing in clinical journals within all pediatric cardiovascular subspecialties. In practice, such collaboration can only work if both communities understand each other's methods and their limitations. This paper is intended to facilitate this communication by presenting in the context of congenital heart disease (CHD) the main steps involved in performing computational simulation -from the selection of an appropriate clinical question/problem to understanding the computational results, and all of the "black boxes" in between.We examine the current state of the art and areas in need of continued development. For example, medical image-based model-building software has been developed based on numerous different methods. However, none of them can be used to construct a model with a simple "click of a button." The creation of a faithful, representative anatomic model, especially in pediatric subjects, often requires skilled manual intervention. In addition, information from a second imaging modality is often required to facilitate this process. We describe the technical aspects of model building, provide a definition of some of the most commonly used terms and techniques (e.g. meshes, mesh convergence, Navier-Stokes equations, and boundary conditions), and the assumptions used in running the simulations. Particular attention is paid to the assignment of boundary conditions as this point is of critical importance in the current areas of research within the realm of congenital heart disease.Finally, examples are provided demonstrating how computer simulations can provide an opportunity to "acquire" data currently unobtainable by other modalities, with essentially no risk to patients.To illustrate these points, novel simulation examples of virtual Fontan conversion (from preoperative data to predicted postoperative state) and outcomes of different surgical designs are presented. The need for validation of the currently employed techniques and predicted results are required and the methods remain in their infancy. While the daily application of these technologies to patient specific clinical scenarios likely remains years away, the ever increasing interest in this area among both clinicians and engineers makes its eventual use far more likely than ever before and, some could argue, only a matter of [computing] time.

A comparison of outlet boundary treatments for prevention of backflow divergence with relevance to blood flow simulations

“…Models for the first two cases were analytically defined and constructed by lofting together circles. For the patientspecific model in case three, starting from the X-ray computed tomography (CT scan), the geometry of each artery is identified and segmented [34]. By lofting these segments through the centerline of the vessels, the solid model is created.…”

Virtual surgeries in patients with congenital heart disease: a multi-scale modelling test case

“…The geometrical three-dimensional model of the pre-operative anatomy, representing the BCPA, was created from MR images using SIMVASCULAR (http://simtk.org) [9] and is depicted in figure 1b.…”

Inverse problems in reduced order models of cardiovascular haemodynamics: aspects of data assimilation and heart rate variability

“…In recent years, computational fluid dynamics (CFD) models, large parts of which comprise lumped parameter models, has been employed to aid clinical decision-making and development of novel surgical strategies [36,37]. For individual case-by-case clinical decision-making in single-ventricle patients, however, the estimation of lumped model parameters while taking into account the uncertainty in clinical measurements is necessary.…”