Semi-Automatic Vortex Extraction in 4D PC-MRI Cardiac Blood Flow Data using Line Predicates

Abstract

This indicates an improperly closing aortic valve and supports the decision whether or not to implant an artificial valve.Abstract-Cardiovascular diseases (CVD) are the leading cause of death worldwide. Their initiation and evolution depends strongly on the blood flow characteristics. In recent years, advances in 4D PC-MRI acquisition enable reliable and time-resolved 3D flow measuring, which allows a qualitative and quantitative analysis of the patient-specific hemodynamics. Currently, medical researchers investigate the relation between characteristic flow patterns like vortices and different pathologies. The manual extraction and evaluation is tedious and requires expert knowledge. Standardized, (semi-)automatic and reliable techniques are necessary to make the analysis of 4D PC-MRI applicable for the clinical routine. In this work, we present an approach for the extraction of vortex flow in the aorta and pulmonary artery incorporating line predicates. We provide an extensive comparison of existent vortex extraction methods to determine the most suitable vortex criterion for cardiac blood flow and apply our approach to ten datasets with different pathologies like coarctations, Tetralogy of Fallot and aneurysms. For two cases we provide a detailed discussion how our results are capable to complement existent diagnosis information. To ensure real-time feedback for the domain experts we implement our method completely on the GPU.

VesselMap: A web interface to explore multivariate vascular data

“…van Pelt et al [22] used various techniques to depict the blood flow and associated characteristics in different styles, together with an evaluation to measure the value of those visualization styles. Köhler et al [23] extracted vortices in blood flow data sets using line predicates and highlighted the corresponding regions. van Pelt et al [24] proposed to semi-automatically place and align a probe in the blood flow field, which serves as a seeding basis.…”

Fuzzy feature tracking: Visual analysis of industrial 4D-XCT data

“…Köhler et al [13] published an approach for the extraction of vortex flow in the aorta and pulmonary artery incorporating line predicates. This technique evaluates time-resolved and spatial phase-contrast magnetic resonance imaging data (4D PC-MRI) and facilitates reliable measuring of 3D flow for qualitative and quantitative analysis of the patient-specific hemodynamics.…”

Motion-aware stroke volume quantification in 4D PC-MRI data of the human aorta

“…Line predicates are a threshold-based filtering technique for integral lines such as pathlines, which are commonly used to visualize blood flow. Line predicates were used by Born et al [4] and Köhler et al [18] to extract qualitative flow features such as vortices. Carnecky et al [7] provide a special flow data preprocessing to improve noise robustness of the employed λ 2 vortex criterion.…”

“…Flow calculation The GPU is utilized to integrate the full set of pathlines [10] and to derive the λ 2 -criterion that is required for the line predicate-based vortex extraction [18]. Velocity vectors u ∈ R 3 in the 4D flow field V at the spatiotemporal position x = (x, y, z, t) T are obtained using quadrilinear interpolation.…”