Towards Absolutely Quantitative Phase Contrast Magnetic Resonance Imaging
Phase Contrast Magnetic Resonance Imaging (PC-MRI) is a non-invasive clinical imaging technique used primarily for measuring blood velocity and flow throughout the major blood vessels of the cardiovascular system. PC-MRI uses magnetic field gradients to impart zero phase to stationary spins and a non-zero phase to moving spins. The phase measurements provide quantitative information that is useful during the diagnosis and treatment of many cardiovascular diseases. Blood flow measurements obtained using PC-MRI hold an advantage over other techniques, namely echocardiography and catheterization, due to its ability to reduce lifetime radiation exposure, provide accurate and direct quantification of flow, and it's non-invasiveness. Despite decades of research, our ability to measure blood flow with PC-MRI is still hampered by quantitative inaccuracies leading to clinically significant errors, which dampens clinical enthusiasm for the technique. Nevertheless PC-MRI continues to be a compelling clinical technique because of the need to non-invasively measure flow in a wide range of clinical contexts. Frequently inconsistent PC-MRI measurements, however,continue to be a source of clinical frustration and in order for PC-MRI to become an absolutely quantitative measure of flow, both the accuracy and precision of these measurements must be improved. Herein an analysis of the effects of chemically shifted perivascular fat; time efficient velocity encoding; region-of-interest contouring; and the use of convex gradient optimization is conducted in an effort towards developing absolutely quantitative PC-MRI.
In Chapter 4 we explore the phase errors associated with chemically shifted perivascular fat. Stationary perivascular fat, which surrounds most vessels throughout the cardiovascular system, can impart a significant chemical shift-induced phase error in PC-MRI. This chemical shift error does not subtract in phase difference processing, unlike other off-resonance phase errors, but can be minimized significantly with proper parameter selection. The chemical shift induced phase errors largely depend on both the receiver bandwidth and the echo time (TE). The amount of chemically shifted fat pixels that shift into the vessel can be reducedby increasing the receiver bandwidth while the use of an in-phase TE (TEIN) will ensure that fat and water resonances are in-phase with slow flowing blood near the vessel wall, which minimizes the resulting errors in the calculated velocity. Computational simulations and both in vitro and in vivo experiments are used to show that the use of a high bandwidth and TEIN significantly improves intra-subject flow agreement compared to a more clinically standard low receiver bandwidth and the minimum available TE (TEMIN).
In Chapter 5 we explore a time efficient chemical shift reduction strategy. The minimum available TEIN at 3T field strength (TEIN,MIN = 2.46ms), however, may not be routinely achievable with standard flow-encoding methods. Hence, we developed a novel method for flow encoding in PC-MRI, which uses the slice select gradientand a time-shifted refocusing gradient lobe for velocity encoding. Velocity encoding with the slice select refocusing gradient (SSRG) enables the use of TEIN,MIN at 3T for time-efficient reduction of chemical shift-induced phase errors in PC-MRI, whereas this can't be achieved with bi-polar or flow compensated/flow encoded PC-MRI. In vivo measurements were acquired to show that PC-MRI measurements obtained using SSRG with a high receiver bandwidth and TEIN,MIN significantly improves intra-subject flow agreement compared to a conventional clinical sequence, which uses a low receiver bandwidth and TEMIN. This approach also increases temporal resolution and signal-to-noise ratio by 35% and 33%, respectively.
In Chapter 6 we explore time efficient velocity encoding and the capabilities of convex gradient optimization in PC-MRI in chapter 7. Conventional PC-MRI pulse sequences use time inefficient velocity encoding methods along with trapezoidal and triangular gradient lobes, which do not make optimal use of the available gradient hardware. Convex gradient optimization (CVX) can be used to minimize PC-MRI gradient waveform durations subject to both gradient hardware and pulse sequence constraints.CVX PC-MRI with TEIN,MIN provides more accurate measurements of blood flow and velocity through the reduction of chemical shift-induced phase errors and increased sequence efficiency, which can provide either higher spatial or higher temporal resolution.
Another potential source of error in PC-MRI flow quantification occurs during image analysis. In Chapter 8 we analyze the errors associated with ROI contouring in PC-MRI. PC-MRI blood flowmeasurements require a region-of-interest (ROI) to be manually contoured to encompass the vessel lumen, but this process is subjective and prone to error. A systematic analysis of ROI contouring was used to evaluate the impact of overestimating and underestimating the ROI size on PC-MRI flow measurements. ROIs that overestimate the vessel lumen/wall boundary contribute a lower magnitude total flow error compared to ROIs that underestimate the same boundary.
Reducing errors arising from chemically shifted perivascular fat, implementing time efficient velocity encoding, increasing spatiotemporal resolution through the use of convex gradient optimization, and careful analysis of ROI contours all help move us towards absolutely quantitative PC-MRI.