Original Research Resolution Improvement in Thick-Slab Magnetic Resonance Digital Subtraction Angiography Using SENSE at 3T Paul E. Summers, PhD, * Spyros S. Kollias, MD, and Anton Valavanis, MD Purpose: To evaluate the use of parallel imaging (sensitiv- ity encoding [SENSE]) to improve spatial resolution and achieve sub-second temporal resolution in fluoroscopic contrast-enhanced, magnetic resonance digital subtrac- tion angiography (MR-DSA). Materials and Methods: A MR-DSA sequence was opti- mized on a 3-T scanner with respect to sampling bandwidth and SENSE acceleration factor subject to the constraints of half-second acquisition time and 0.6 1.2 mm in-plane resolution. MR-DSA with and without SENSE acceleration was then evaluated in patients with arterio-venous malfor- mations (AVMs). Results: Consistent with previously reported results and theory, SENSE factors greater than two and increasing sampling bandwidth both led to increasing image noise. Compared to lower resolution MR-DSA images with similar temporal resolution, the SENSE accelerated se- quence provided better spatial resolution without notable changes in the contrast enhancement of the vascular territories of the AVMs but was hampered somewhat in the late venous phases by a reconstruction artifact. Conclusion: SENSE acceleration of MR-DSA by a factor of two allows improved temporal or spatial resolution without significant loss of image quality. Signal-to-noise degradation associated with higher SENSE acceleration factors are likely to necessitate other approaches to fur- ther improving resolution in MR-DSA. Clinically, SENSE accelerated MR-DSA improves the non-invasive pre- and postoperative depiction of AVM flow dynamics. Key Words: magnetic resonance angiography; MR-DSA; parallel imaging; arterio-venous malformations. J. Magn. Reson. Imaging 2004;20:662– 673. © 2004 Wiley-Liss, Inc. REPEATEDLY ACQUIRING T1-WEIGHTED magnetic resonance (MR) images during the first passage of an injected, T1-shortening contrast agent, together with digital subtraction of an image acquired before the ar- rival of contrast, can be used to selectively image vas- cular structures without interference from overlying tissues (1). This technique, magnetic resonance digital subtraction angiography (MR-DSA), is widely used for extracranial MR angiography where volumetric data sets can be acquired over the course of a few seconds without contamination of the arterial phase images by venous structures (2– 4). The temporal resolution of such three-dimensional MR-DSA studies is, however, inadequate for visualizing the dynamics of contrast passage through the intracranial vasculature. This is particularly relevant in the diagnostic and follow-up imaging of high-flow arterio-venous malformations (AVMs). Notably, the gold standard for intracranial vascular studies—intra-arterial digital subtraction (IA-DSA)—is performed in projection mode with high temporal and in-plane spatial resolution but without depth discrimi- nation. The utility of IA-DSA indicates that depth infor- mation is not critical to the interpretation of temporally resolved images. Thus, a major reduction in imaging time can be made by acquiring projection images (two- dimensional MR-DSA) rather than three-dimensional datasets (5,6). Several previous reports have demon- strated two-dimensional MR-DSA with one-second temporal resolution for visualization of intracranial AVMs (7–10). This has typically involved the acquisition of 256 readout samples, and 200 phase encoding steps with a 5-msec repetition time. For a field of view (FOV) of 20 cm, sampling theory dictates that the spatial res- olution is limited to 1.75–2 mm. Thus, two-dimensional MR-DSA compares poorly with both the micron-scale spatial and tens of frames per second temporal resolu- Institute of Neuroradiology, University Hospital Zurich, Zurich, Swit- zerland. Contract grant sponsor: Swiss National Science Foundation; Contract grant number: 3200B0-100355. Parts of this work were presented as Poster 2263 at the 2003 meeting of the ISMRM. *Address reprint requests to: P.S., Institute of Neuroradiology, Univer- sity Hospital Zurich, Frauenklinikstrasse 10, 8091 Zurich, Switzer- land. E-mail: paul.summers@usz.ch Received August 21, 2003; Accepted May 24, 2004. DOI 10.1002/jmri.20156 Published online in Wiley InterScience (www.interscience.wiley.com). JOURNAL OF MAGNETIC RESONANCE IMAGING 20:662– 673 (2004) © 2004 Wiley-Liss, Inc. 662