Four-dimensional phase contrast magnetic resonance angiography: potential clinical applications

Abstract

Unlike other magnetic resonance angiographic techniques, phase contrast imaging (PC-MRI) offers co-registered morphologic images and velocity data within a single acquisition. While the basic principle of PC-MRI dates back almost 3 decades, novel time-resolved three-dimensional PC-MRI (4D PC-MRI) approaches have become increasingly researched over the past years. So-called 4D PC-MRI includes three-directional velocity encoding in a three-dimensional imaging volume over time, thereby providing the opportunity to comprehensively analyze human hemodynamics in vivo. Moreover, its large volume coverage offers the option to study systemic hemodynamic effects. Additionally, this offers the possibility to re-visit flow in any location of interest without being limited to predetermined two-dimensional slices. The attention received for hemodynamic research is partially based on flow-based theories of atherogenesis and arterial remodeling. 4D PC-MRI can be used to calculate flow-related vessel wall parameters and may hence serve as a diagnostic tool in preemptive medicine. Furthermore, technical improvements including the availability of sufficient computing power, data storage capabilities, and optimized acceleration schemes for data acquisition as well as comprehensive image processing algorithms have largely facilitated recent research progresses. We will present an overview of the potential of this relatively young imaging paradigm. After acquisition and processing the data in morphological and phase difference images, various visualization strategies permit the qualitative analysis of hemodynamics. A multitude of quantitative parameters such as pulse wave velocities and estimates of wall shear stress which might serve as future biomarkers can be extracted. Thereby, exciting new opportunities for vascular imaging and diagnosis are available.

Copyright © 2011 Elsevier Ireland Ltd. All rights reserved.

Figures

Fig. 1

Principle of phase contrast MRI. (A) Stationary (static) tissue experiences no net magnetization after exact reversal of magnetic field gradients. Moving spins, however, will experience different field gradients which will not cancel out. The remaining magnetization can be detected in form of a small net phase. (B) Subtraction of phase information before and after gradient reversal thus results in a phase difference proportional to the velocity of the underlying motion. In (C) the transfer to 3-directioanl velocity encoding, as commonly performed in 3-dimensional imaging, is outlined. Note, however, that multiple multidirectional encoding schemes are available , .

Fig. 2

Flow chart describing the components of a PC HYPR Flow processing chain for cranial imaging. The strength of this imaging and post-processing approach is the combination of advantages each technique has. The high temporal resolution of the whole volume time-resolved contrast-enhanced MRA (tr CE-MRA) using radial undersampling (VIPR) is complemented by the high spatial resolution of the undersampled phase contrast data (PCVIPR). Note that principally a different high spatial resolution constrained image could be used. The simultaneously acquired phase contrast velocity data can be used to display flow patterns applying streamlines and particle traces, to calculate volume flow, and to estimate derived vessel wall parameters. In our clinical routine we achieve serial images with a spatial resolution of 0.68 × 0.68 × 0.68mm and a sub-second temporal update rate. PC HYPR Flow Imaging courtesy of Yijing Wu, PhD; hemodynamic visualization courtesy of Ben Landgraf, BS, University of Wisconsin.

Fig. 3

Detailed view of a thalamic arteriovenous malformation (AVM) visualized with PC HYPR Flow (same patient as in figure 2). In (A) the whole brain CE-MRA with sub-second temporal resolution is displayed. (B) is the simultaneously available high-resolution PC angiogram (PC VIPR). Procession with HYPR results in a high spatial and temporal resolution (C) with the simultaneous option to perform hemodynamic analyses (D). In (D), visualization (EnSight, CEI, Apex,NC) with streamlines was color-coded enhancing the flow from and to the AVM. Angiographic images courtesy of Yijing Wu, PhD; hemodynamic visualization courtesy of Ben Landgraf, BS, University of Wisconsin.

Fig. 4

(A) 3D angiographic reconstruction from PC VIPR data and (B) streamline visualization of a 4 mm paraclinoid aneurysm. Streamlines are displayed during early diastole. The visualization is based on the velocity information derived from a PC HYPR Flow acquisition. ICA = internal carotid artery, MCA – middle cerebral artery, LT – left, Rt = right. Visualization courtesy of Steve Kecskemeti and Ben Landgraf, BS, University of Wisconsin - Madison.

Fig. 5

Stenosis of the transverse sinus (white arrows). (A) shows a 3D angiographic view based on the PC VIPR data performed with MIMICS Innovation Suite (Materialise, Ann Arbor, MI; blue = venous, red = arterial). For hemodynamic analyses, color-coding was achieved based on pressure differences (B) and velocity information (C). All data was derived from a singular PC HYPR Flow acquisition. A marked drop in pressure across the stenotic segment and elevated velocities with the stenosis can be appreciated. Visualization courtesy of Ben Landgraf, BS, University of Wisconsin – Madison.

Fig. 6

(A) Visualization of a color-coded PC angiogram of the upper abdomen (diaphragm to renal arteries) performed with MIMICS Innovation Suite (Materialise, Ann Arbor, MI). The venous system is coded in blue, the arteries in red, and the portal vein in yellow. For the latter, a streamline visualization using EnSight (CEI, Apex, NC) can be appreciated in (B). Clearly, the potential of large volume coverage and the feasibility to analyze singular vessels of interest or multiple structures and their interdependencies can be appreciated. Visualization courtesy of Ben Landgraf, BS, and Eric Niespodzany, MS, University of Wisconsin - Madison.

Fig 7

A typical visualization pathway of 4D PC- VIPR can include (A) the creation of an isosurface and placement of cutplanes that serve as either emitter planes for visualization options or as analysis planes for quantification. Vascular territories such as the aorta (B) or the pulmonary artery (C; early RV diastole) can be appreciated in singular views or at the same time (D-F). (D) – (F) shows the temporal evolution of flow patterns from the superior and inferior vena cava (SVC, IVC, respectively) and the right ventricular outflow tract (RVOT) during selected RV systolic (D, E) and diastolic (F) time points. Note the regurgitant flow volume (TR) through the tricuspid valve and the flow vortex in the right ventricular outflow tract (white arrowheads). Ao = aorta; DAo = descending aorta; RA = right atrium. Visualization courtesy of Michal Markl, PhD, University Hospital Freiburg, Germany.

Fig 8

Schematic overview over the quantitative results that can potentially be extracted from 4D PC-MRI (based on previously publish results, not complete).

Fig. 9

Schematic illustration of a transit time method to analyze the pulse wave velocity (PWV) based on waveform characteristics. Here, the time to foot (TTF) is used by fitting a line to the upslope of the waveform. The point the fitted line creates crossing the baseline is used to analyze the time shift between the locations of the two analysis planes. The distance between those slices and the time between waveform characteristics is used for calculation of the PWV. Obviously, a high spatial resolution of the acquired flow information allows for a better fit. Similarly, a large distance between slice A and B provides a more accurate PWV estimation. Note that 3 acquisitions are needed for a 2D approach (slice A, slice B, sagittal plane of the aorta) whereas a 4D PC-MRI data volume contains all necessary information in a perfectly co-registered way.

Fig. 10

Wall shear stress (WSS) can be calculated from the acquired 4D PC-MRI data. For each acquired voxel, three-directional velocity information is available. In a 2D plane with velocity encoding in a single direction, Poiseulle’s law can be applied to estimate the shear stress magnitude at the vessel wall (WSS magnate symbolized by arrows 1 and 2). Following Poiseulle’s law, WSS is proportional to the blood flow Q (area times velocity), viscosity (μ), and inversely proportional to the radius (r) of the vessel. Since 4D-PC MRI encodes for all three directions of blood flow, not only the magnitude but also contributing circumferential and through-plane fractions to the WSS magnitude can be analyzed , . Hemodynamic visualization courtesy of Ben Landgraf, BS, and Eric Niespodzany, MS, University of Madison.

Fig. 11

Pressure difference map in a patient after surgery for coarctation. The pressure gradient can be visually appreciated by the color swap (left). On the top right, pressures over time (yellow line) as derived from the acquired 4D velocity field are plotted against pressure gradients derived from the simplified Bernoulli equation. The difference between MR and echocardiography can be related to both, an underestimation of velocity by MR due to volume averaging and to overestimation by echocardiography since maximum values in the center are measured. Image courtesy of Dipl.-Ing. Jelena Bock, University Hospital Freiburg, Germany.