Imaging of Heterogeneous Materials with a Turbo Spin Echo Single-Point Imaging Technique Steven D. Beyea, Bruce J. Balcom, 1 Igor V. Mastikhin, Theodore W. Bremner,* Robin L. Armstrong, and Patrick E. Grattan-Bellew² MRI Centre, Department of Physics, and *Department of Civil Engineering, University of New Brunswick, P.O. Box 4400, Fredericton New Brunswick, Canada E3B 5A3; and ² Institute for Research in Construction, National Research Council of Canada, Ottawa Ontario, Canada K1A 0R6 Received September 8, 1999; revised February 1, 2000 A magnetic resonance imaging method is presented for imaging of heterogeneous broad linewidth materials. This method allows for distortionless relaxation weighted imaging by obtaining mul- tiple phase encoded k-space data points with each RF excitation pulse train. The use of this method, turbo spin echo single-point imaging-(turboSPI), leads to decreased imaging times compared to traditional constant-time imaging techniques, as well as the ability to introduce spin–spin relaxation contrast through the use of longereffective echo times. Imaging times in turboSPI are further decreased through the use of low flip angle steady-state excitation. Two-dimensional images of paramagnetic doped agarose phan- toms were obtained, demonstrating the contrast and resolution characteristics of the sequence, and a method for both amplitude and phase deconvolution was demonstrated for use in high-reso- lution turboSPI imaging. Three-dimensional images of a partially water-saturated porous volcanic aggregate (T 2L  200 ms,  1/2  2500 Hz) contained in a hardened white Portland cement matrix (T 2L  0.5 ms,  1/2  2500 Hz) and a water-saturated quartz sand (T 2  300 ms, T * 2  800 s) are shown. © 2000 Academic Press Key Words: MRI; single-point; SPI; turboSPI; concrete; porous materials. INTRODUCTION The use of nuclear magnetic resonance (NMR) for studying and characterizing bulk properties of porous media has been demonstrated by many authors (1–5). The extension to spa- tially resolved methods using MRI has demonstrated the dif- ficulties involved in obtaining good signal-to-noise ( S / N) high- resolution images of highly heterogeneous media (1, 6 –10). Porous materials frequently possess low fluid contents and short spin–spin relaxation times, both of which contribute to poor-quality NMR images (6). In addition, the difference in magnetic susceptibility between the pore fluid and the solid matrix leads to a large distribution of magnetic fields within the porous materials and correspondingly broad NMR linewidths (2). Images of such porous media have been obtained using spin-echo (SE) (8, 9), gradient echo (9), and -echo-planar (-EPI) (10, 11) imaging methods. Images acquired using all of these methods exhibit, to varying degrees, distortion and resolution loss in the frequency encode direction. Image distortion and resolution loss due to susceptibility heterogeneities can often be overcome through the use of very strong frequency encoding gradients (8 –12). Acquiring image data in the presence of strong gradients, however, comes at the expense of increased acquisition bandwidth, leading to a pro- portional decrease in S/ N. The optimum acquisition bandwidth when frequency encoding, in the absence of diffusion, is N(  1/2 ), where  1/2 is the inhomogeneously broadened line- width. This is the minimum bandwidth necessary to have the frequency width of a pixel equal to that of the natural line- width. Sarkar et al. (9) discuss the requirement of doing as many as 50 signal averages in order to accurately interpret two-dimensional SE images of porous sintered glass disks. This problem is compounded when imaging in the presence of molecular diffusion. It is well known that diffusion effects can be minimized through the use of stronger gradients applied for shorter periods of time (13, 14). This requires a frequency encoding acquisition bandwidth greater than the optimal value required to resolve the image, leading to a further decrease in S/ N. Pure phase encoding techniques are often useful for imaging highly heterogeneous samples, due to the fact that images acquired with this technique do not contain artifacts due to chemical shift, susceptibility variations, or imperfect B 0 shim- ming (13–20). However, while pure phase encode methods are excellent at obtaining high-resolution “artifact-free” images, it is generally accepted that these methods are slow (21), as they require N 2 excitations for a 2D image of N 2 pixels. This is typically true, however, only because traditional spin-warp (frequency-phase) (22) methods are routinely used on systems with intrinsically large NMR signals and long signal lifetimes, and the subsequent images are usually acquired with little or no signal averaging (14, 15). When significant signal averaging is necessary, such as in high-resolution imaging and imaging of heterogeneous materials (e.g., porous media), it can be shown that pure phase encoding is equally time efficient. 1 To whom correspondence should be addressed. Journal of Magnetic Resonance 144, 255–265 (2000) doi:10.1006/jmre.2000.2054, available online at http://www.idealibrary.com on 255 1090-7807/00 $35.00 Copyright © 2000 by Academic Press All rights of reproduction in any form reserved.