Diffusion-Weighted MR Microscopy with Fast Spin-Echo z Chris F. Beaulieu, Xiaohong Zhou, Gary P. Cofer, G. Allan Johnson zyx A diffusion-weighted fast spin-echo (FSE) imaging sequence for high-field MR microscopy was developed and experimen- tally validated in a phantom and in a live rat. Pulsed diffusion gradients were executed before and after the initial 180" pulse in the FSE pulse train. This produced diffusion-related reduc- tions in image signal intensity corresponding to gradient zyxwvuts ("b") factixs between 1.80 and 1352 s/mm2. The degree of diffusion weighting was demonstrated to be independent of echo train length for experiments using trains up to 16 echoes long. Quantitative measurements on a phantom and on a live rat produced diffusion coefficients consistent with literature val- ues. Importantly, the eight- to 16-fold increase in imaging ef- ficiency with FSE was not accompanied by a significant loss of spatial resolution or contrast. This permits acquisition of in zyxwvutsr vivo three-dimensional data in time periods that are appropri- ate for evolving biological processes. The combination of ac- curate diffusion weighting and high spatial resolution pro- vided by FSE makes the technique particularly useful for MR microscopy. Key words: MA microscopy; fast spin-echo zyxwvuts (FSE); pulse se- quences. INTFiODUCTION Fast spin-echo (FSE) pulse sequences have recently been adapted to magnetic resonance (MR) microscopy (1). FSE is based on a Carr-Purcell-Meiboom-Gill (CPMG,2,3) echo train in which phase-encoding of individual echoes provides spatial mapping (4, 5). At high magnetic fields the Iecho trains are typically eight to 16 echoes long, resul ring in eight- to 16-fold reductions in imaging time compared with single-echo imaging. A key feature of FSE is that spatial resolution is zyxwvutsrq not significantly reduced de- spite the increase in efficiency (1). This makes FSE par- ticularly appealing for diffusion-related contrast manip- ulation at microscopic spatial resolution, especially for three-dimensional (3D) studies with large arrays. Diffusion imaging with FSE has several potential ad- vantages (6, 7). image contrast afforded by spatial varia- tions in water proton translational diffusion provides a sensitive parameter for monitoring pathologic processes in vivo (8-10). In stroke, for example, a change in the apparent diffusion coefficient (ADC) of brain water ap- pears to be the earliest MRI-detectable manifestation of tissue injury (9, 11, 12). Diffusion imaging with FSE could help study stroke and other processes with im- MRM 00:201-206 zyxwvutsrqpon (1993) From the Department of Radiology, Center for In Vivo Microscopy, Duke University Medical Center, Durham, North Carolina. Address correspondence to: G. A. Johnson, Ph.D., Duke University Medical Center, Department of Radiology, Box 3302, Durham, NC 27710. Received December 24, 1992; revised March 31, 1993; accepted April zyxwvuts 5, 1993. This research was supported by IC1 Pharmaceuticals, NIH Grant No. P41- RR-05059-02, and NIEHS Grant No. ROI-ES04187-04A1. 0740-3.1 94/93 $3.00 Copyrigiht 0 1993 by Williams & Wilkins All rights of reproduction in any form resewed. proved temporal resolution while maintaining the spatial resolution of single-echo images. While not as fast as echo-planar imaging (EPI; 13, 14), FSE does not suffer from the reduced spatial resolution or heightened sensi- tivity to magnetic susceptibility of EPI. Both of these €actors have a negative impact on high-field MR micros- copy. Diffusion-weighted FSE could provide an ideal combination of high spatial resolution and diffusion-re- lated contrast sensitivity (6, 7). A feature of FSE that is particularly important in MR microscopy is the potential to shorten acquisitions of 3D data to periads that are suitable for animal life-support and that are relevant to dynamic biological processes. The latter issue is important in studies such as stroke, in which long data acquisition times may average the effects of evolving ischemia over many hours. Three-dimen- sional FSE will permit acquisition of large data arrays with isotropic or near-isotropic voxels on live animals (1). This is a crucial step in producing volume-rendered data for examination of inherently 3D pathological pro- cesses (15). This report describes a diffusion-weighted version of FSE and demonstrates its accuracy in measuring diffu- sion coefficients on a phantom and on a normal live rat. Diffusion coefficients (D) were measured for varying echo train lengths (ETL) on a phantom containing stan- dard chemicals. Two-D quantitative as well as 3D diffu- sion-weighted data were obtained on a normal live rat. The results indicate that diffusion-weighted FSE will be a useful microscopic imaging technique owing to its spe- cial combination of high spatial resolution and accurate diffusion weighting. METHODS AND MATERIALS Pulse Sequence The pulse sequence is shown in Fig. 1. An initial 90° pulse was followed by a variable-length CPMG train of 180" pulses. A pair of pulsed diffusion-sensitizing gradi- ents (16) was placed along the slice-selection axis with the first of the pair immediately following the zyx goo slice- selection gradient and the second immediately prior to readout. The diffusion gradients were half-sine shaped to minimize eddy currents and gradient-induced vibra- tions. Crusher gradients with alternating polarity and de- creasing amplitude were used to spoil signals generated by imperfect 180" pulses. Either slice-selective 180° sinc pulses or nonselective 180" pulses, as shown, were effec- tive. The hard 180° pulse was 482 ps. The pulse program permitted placement of diffusion gradients along any axis or combination of axes. All of the studies shown here utilized slice-direction diffusion gradients. The phase- encoding axis shows the phase-encoding and rewinding performed for each echo. The scheme shown assigns the zero-amplitude phase-encoding gradient to the first echo. 203