Techniques to reduce off-resonance image artifacts in rapid MRI.

摘要

Among modern medical imaging modalities, magnetic resonance imaging (MRI) is one of the slowest. Although restricted in part by natural properties of atomic nuclei, MRI scanning speed is limited by its inherent serial sampling paradigm and inefficiencies in the sampling trajectory, the path through frequency space along which the spectral data are acquired. Most MRI studies employ trajectories that under-utilize the scanner hardware and prolong the scan.;Fast trajectories such as spirals and echo planar patterns can reduce scan time several fold. Compared to other acceleration methods, rapid MRI trajectories more generally applicable and can be implemented without additional hardware or signal modeling assumptions. However, they are sensitive to the signal phase errors generated by off-resonance, the inevitable variation in resonance frequency across the imaging volume. Images they produce tend to contain corruptive artifacts which have mostly limited their usage to niche research applications. Yet they remain relevant to the future of MRI. With high field imaging comes the promise of finer resolutions and faster scans, which may not be realizable without fast trajectories.;This work introduces new techniques for robust off-resonance artifact reduction and correction. We present an improved outer volume suppression (OVS) sequence to suppress signal and allow imaging with a reduced field-of-view (rFOV). Our OVS design is tolerant of field inhomogeneity and has short duration, unlike other methods. The rFOV mitigates the image blurring artifacts characteristic of spiral trajectories. We derive a new solution to the Bloch equations with an interesting property that is utilized in the design.;We also present a new algorithm for automatic deblurring of images acquired with spiral trajectories. This is the first autofocus method for MRI that does not optimize a focus metric. Instead, it estimates the underlying off-resonance using signal physics and a piecewise linear processing framework. Evaluations using fine-resolution spiral imaging show that our method can outperform current automatic methods, and is comparable to non-automatic approaches wherein the off-resonance is known.;We also show how the automatic off-resonance correction algorithm can be modified to accommodate alternate trajectories. We investigate its application to reduce the geometric distortion artifacts caused by off-resonance in echo planar imaging. Initial tests show the reduction in distortion using our approach is similar to non-automatic methods.;Ironically, rapid trajectories may be the tortoise in the long race to improve MRI. Their immediate potential to shorten scans has kept them in the running, but they are encumbered by the need for robust mitigation of off-resonance artifacts. Using the proposed techniques, rapid trajectories may become the front-runner in more MRI protocols.