Accelerated Neuro-MRA Using Compressed Sensing and Constrained Reconstruction

点击翻译按钮获取中文摘要

Abstract Magnetic resonance imaging (MRI) is the standard of care for most diagnostic neurological imaging, but with certain notable shortcomings. Cerebral aneurysms, still first often diagnosed either by sudden death or catastrophic hemorrhage, are best visualized with the resolution provided by computed tomography or digital substraction angiography (DSA). The speed of MRI is often not enough to visualize the arterial inputs and venous drainage of arterial-venous malformations (AVM). Ischemic stroke is the most common neurological disorder worldwide and intracranial arterial stenosis is a major risk factor for ischemic stroke. In order to improve confidence of diagnosis and provide early detection of the pathological changes in the cerebrovascular system, significant advances should be made towards spatial and temporal resolutions currently unavailable even with state-of-art MRA techniques. We have been developing acquisition and reconstruction methods that circumvent MRI shortcomings in speed and resolution to provide multi-dimensional physiological and anatomical information for neurovascular imaging. We have recognized that in order to achieve the required combinations of spatial and temporal resolution and signal-to-noise ratio we need to exploit the synergy of complementary advanced image acquisition and reconstruction techniques. Image estimation methods developed in our labs combine constrained reconstruction algorithms with non-Cartesian radial trajectories whose variable sampling density allows for both high quality extended scans and time-resolved imaging. We have already successfully developed the first generation of this technology known as the HYPR (HighlY constrained back Projection) family of imaging techniques, delivering substantial acceleration to the acquisition of serially acquired images. This proposal suggests a next generation of accelerated imaging technology for the comprehensive evaluation of vessel stenoses, aneurysms, and AVMs that will rival and surpass CT through the development of new image acquisition and reconstruction methods. These methods will utilize independent and symbiotic acceleration mechanisms of optimized radial trajectories, parallel imaging, and constrained reconstruction, including HYPR and advanced compressed sensing algorithms. These algorithms will also be supplied with data from novel highly accelerated acquisition methods: 1) a contrast-free inflow technique that eliminates the dispersion of contrast-enhanced bolus to provide superb arterial isolation, high resolution, and coverage; and 2) high quality time-averaged vascular image volumes to constrain reconstruction of time-resolved contrast- enhanced data. These methods will be evaluated in the treatment and tracking of AVMs, the evaluation of vascular stenoses and the evaluation of cerebral aneurysms. Successful completion would supplement the arsenal of tools used in stroke management as well.