An interactive computational pipeline to investigate ventricular hemodynamics with real-time three-dimensional echocardiography and computational fluid dynamics

Abstract

Blood flow within the ventricle can provide important information on the performance of the heart. The determined blood flow structures are used to extract flow biomarkers to quantify cardiac function. Patient-specific computational fluid dynamics (CFD) models that import segmented ventricular deformations from non-invasive imaging data for an individualized hemodynamical analysis are often used. However, tedious pre-processing of those geometries is often necessary and decisions on the modeling of the valve and the surrounding vessels have to be made on an individual level. This leads to a lack of reproducibility and usability of the existing computational models. In this work we introduce IP-HEART - an interactive and open-source computational pipeline to perform geometry processing for CFD models of ventricular blood flow. We showcase its use on real-time three-dimensional echocardiography data of three patient datasets from two different clinical centers. We outline how different modeling assumptions of the mitral valve can be easily implemented and quantify their effect on CFD simulations. The results correspond well with clinical data on transvalvular Doppler ultrasound recordings and distinct flow features such as mitral jet and diastolic vortex formation can be observed. The pipeline is accompanied by an extensive video tutorial and freely available code for further use.