Super-Resolved Spatially Encoded Single-Scan 2D MRI Noam Ben-Eliezer, 1 Michal Irani, 2 and Lucio Frydman 1 * Single-scan MRI underlies a wide variety of clinical and research activities, including functional and diffusion studies. Most common among these ‘‘ultrafast’’ MRI approaches is echo-planar imaging. Notwithstanding its proven success, echo-planar imaging still faces a number of limitations, partic- ularly as a result of susceptibility heterogeneities and of chemical shift effects that can become acute at high fields. The present study explores a new approach for acquiring mul- tidimensional MR images in a single scan, which possesses a higher built-in immunity to this kind of heterogeneity while retaining echo-planar imaging’s temporal and spatial perform- ances. This new protocol combines a novel approach to mul- tidimensional spectroscopy, based on the spatial encoding of the spin interactions, with image reconstruction algorithms based on super-resolution principles. Single-scan two-dimen- sional MRI examples of the performance improvements pro- vided by the resulting imaging protocol are illustrated using phantom-based and in vivo experiments. Magn Reson Med 63:1594–1600, 2010. V C 2010 Wiley-Liss, Inc. Key words: single-scan imaging; super-resolution; ultrafast MRI; spatial encoded imaging; susceptibility compensation; EPI The last decades have witnessed a continuous growth in the use of single-scan MRI, both for clinical and research applications (1,2). These ‘‘ultrafast’’ protocols play an essential role in experiments demanding high temporal resolution like functional MRI (3–5); they also constitute integral components in high-dimen- sionality experiments such as diffusion tensor imag- ing (6). Foremost among the sequences enabling the acquisition of MR images in a single scan stands echo-planar imaging (EPI) (7), with its many different variants (8,9). EPI relies on a single excitation of all spins within the volume to be examined, followed by repetitive gradient oscillations that scan, in a single continuous acquisition, large regions of the image conjugate ( ~ k-space) domain. The rð ~ r Þ spin density pro- file being sought is then retrieved by a numerical Fou- rier transform (FT) of the digitized information. Not- withstanding their real-time image-gathering capabilities, EPI-based protocols are still challenged by the relatively long data sampling times that they involve. These are ca. an order of magnitude longer than those typically involved in multiscan MRI and ex- pose the protocol to progressive temporal artifacts aris- ing from susceptibility variations, from unfavorable shimming conditions, or from chemical shift heteroge- neities. These in turn put practical limitations to the organs and/or conditions that can be studied using ultrafast MRI protocols. By contrast to EPI’s reliance on contributions aris- ing simultaneously from the entire sample, we have recently begun exploring the consequences of relying on a progressive spatial encoding of MR images. Cen- tral in the development of these new experiments is the spatiotemporal manipulation of the spin interac- tions, a concept that originated from a search for methods capable of delivering arbitrary multidimen- sional MR spectra in a single scan (10,11). The gener- ality of the ensuing approach eventually led to sev- eral new routes for executing single-scan multidimensional MRI (12–16). Contrary to EPI, these new MRI methods were found to be local in nature in the sense that, at each instant, the spins’ signal S(t) becomes proportional to the density profile rð ~ r Þ within a limited region of the volume of interest. It fol- lows that spatially encoded methods do not require FT processing for delivering their imaging information: the signal’s magnitude, |S(t)|, is the image being sought. We and others have discussed elsewhere how this property, which in turn is closely linked to experi- ments put forward by Kunz and Pipe decades ago (17– 19), allows spatially encoded MRI to cope efficiently with field inhomogeneities and to deal simultaneously with multiple sites possessing different chemical shifts (14–16). These advantages, however, were also found to materialize at a decreased efficiency in terms of the spatial encoding’s use of the acquisition time variable, which usually results in spatial resolution penalties. In this study, we introduce a way of solving these defi- ciencies. Figure 1 illustrates the ensuing benefits with an in vivo example, showing how super-resolution algorithms (SR) and spatially encoded methods can be combined to retrieve high-quality two-dimensional (2D) MR images. The resulting single-scan data possess spatial and temporal resolutions comparable to those achieved by EPI, but a much higher immunity to fre- quency-dispersing artifacts. The following paragraphs discuss how SR algorithms (20,21)—which find wide- spread use in microscopy scenarios to ‘‘break’’ wave- length-imposed diffraction limits (22), as well as in compensating motion-related artifacts (23)—couple in a natural way to the time-dependent spin evolution involved in spatially encoded MRI. We then present ways for executing and processing this kind of experi- ment and demonstrate the multiple advantages of the resulting method when attempting to carry out single- scan MRI under a variety of conditions that challenge conventional EPI’s usual capabilities. 1 Chemical Physics Department, Weizmann Institute, Rehovot, Israel. 2 Computer Sciences Department, Weizmann Institute, Rehovot, Israel. *Correspondence to: Lucio Frydman, Ph.D., Chemical Physics Department, Weizmann Institute, 76100 Rehovot, Israel. E-mail: lucio.frydman@ weizmann.ac.il Received 9 September 2009; revised 9 December 2009; accepted 6 January 2010. DOI 10.1002/mrm.22377 Published online in Wiley InterScience (www.interscience.wiley.com). Magnetic Resonance in Medicine 63:1594–1600 (2010) V C 2010 Wiley-Liss, Inc. 1594