(Gradient- and Spin-Echo) MR Imaging: a New Fast Clinical Imaging Technique’

Magnetic Resonance Imaging

David A. Feinberg, PhD, MD #{149}Koichi Oshio, MD, PhD

GRASE (Gradient- and Spin-Echo) MR Imaging: A New Fast Clinical Imaging Technique’

A novel technique of magnetic reso- W HILE magnetic resonance (MR) ben of image sections and in S/N, the nance (MR) imaging, which corn- imaging has improved rapidly hybrid RARE technique currently has bines gradient-echo and spin-echo oven the last decade, it is remarkable a four- to 16-fold reduction in imag- (GRASE) technique, accomplishes that multisection two-dimensional ing time from that of spin-echo imag- T2-weighted multisection imaging in Fourier-transform spin-echo imaging ing with one excitation and has tissue drastically reduced imaging time, (1) with T2 weighting has remained a contrast similar to that of the spin- currently 24 times faster than spin- standard of reference for routine clini- echo technique. In hybrid RARE im- echo imaging. The GRASE technique cat MR imaging studies of the body aging, the rapid application of a large maintains contrast mechanisms, high and head. Since the earliest studies in number of 180#{176}radio-frequency (RF) spatial resolution, and image quality 1982, there has been a progressive pulses produces considerably higher of spin-echo imaging and is compati- reduction in image acquisition time. deposition of RE energy in the human ble with clinical whole-body MR sys- Improvements in MR-imager hard- body at the standard absorption rate tems without modification of gradi- ware and pulse-sequence design have (SAR). Although SAR increases with ent hardware. Image acquisition time permitted faster imaging times by ne- shorter time intervals between RE is 18 seconds for 11 rnultisection body ducing the number of signals aver- pulses, the speed of RARE imaging is images (2,000/80 [repetition time aged (NSA) used to raise the signal- ultimately limited by the time require- msec/echo time msec]) and 36 sec- to-noise ratio (S/N) in images. ments of section-selective 180#{176}RF onds for 22 brain images (4,000/104). Conjugate synthesis of data (half Fou- pulses during which signal cannot be With a combination of multiple Hahn acquired. rier or NSA = Y2) yields nearly an- spin echoes and short gradient-echo other factor-of-two reduction in imag- In echo-planar imaging (5), signals trains, the GRASE technique over- ing time by taking advantage of a are produced by rapid switching of comes several potential problems of natural symmetry in the k-space data gradient polarity in place of the echo-planar imaging, including large slower selective 180#{176}RF pulses. In this to halve the number of phase-en- chemical shift, image distortions, and way, echo-planar imaging and its coded signals (2,3). For clinical imag- signal loss from field inhomogeneity. modern variants, modulus-blipped ing, it is significant that the conjugate Advantages of GRASE over the echo-planar single-pulse technique synthesis method produces tissue RARE (rapid acquisition with relax- (MBEST) (6) and Instascan (Advanced contrast, chemical shift, and spatial ation enhancement) technique in- NMR, Woburn, Mass) (7), both involv- resolution identical to that of spin- dude faster acquisition times and ing use of a single 180#{176}RF pulse, can echo imaging with one excitation. lower deposition of radio-frequency produce an image in less than 100 Whereas the conjugate synthesis power in the body. Breath holding msec. At high magnetic field method produces an expected 35%- during 18-second GRASE imaging of strengths, methods of echo-planar 41 % reduction in image S/N, this S/N the upper abdomen eliminates respi- imaging produce a large amount of is acceptable for many T2-weighted ratory-motion artifacts in T2- chemical shift on the phase axis of examinations. weighted images. A major improve- images, typically 10-15-pixel misregis- ment in T2-weighted abdominal An alternative approach for fast tration between water and fat. This imaging is suggested. T2-weighted imaging is the rapid ac- problem can be circumvented by us- quisition with relaxation enhance- ing fat-suppression methods. These ment (RARE) technique (4). The hy- Index terms: Abdomen, MR studies, 70.1214 single-shot images, however, have brid RARE technique, also called fast Brain, MR studies, 13.1214 #{149}Magnetic reso- lower spatial resolution and S/N than nance (MR), comparative studies . Magnetic spin-echo imaging, reduces imaging spin-echo images. This can be im- resonance (MR), echo planar . Magnetic reso- time by performing phase encoding proved by using multiple excitation nance (MR), rapid imaging . Magnetic reso- both in trains of multiple spin echoes nance (MR), technology and during multiple cycles (repetition times [TRs]) of signal excitation. Radiology 1991; 181:597-602 While there is a reduction in the num- Abbreviations: CPMG = Carr-Purcell-Mei- boom-Gill, GRASE = gradient echo and spin echo, MBEST = modulus-blipped echo-planar single-pulse technique, NSA = number of sig- I From the Department of Radiology, Harvard Medical School, Brigham and Women’s Hospital, nals averaged, RARE = rapid acquisition with Boston. Received April 18, 1991; revision requested June 1; revision received June 25; accepted July relaxation enhancement, RF = radio frequency, 1. Address reprint requests to D.A.F., Department of Radiology, New York University Medical Cen- SAR = standard absorption rate, S/N = signal- ter, 560 First Ave. New York, NY 10016. to-noise ratio, TE = echo time, TR = repetition

597 cycles (multiple TRs) that lengthen imaging times. Other stringent re- quirements of echo-planar imaging include static magnetic fields with good homogeneity (better than 0.3 ppm for head imaging with a 1.5-T -i system) to avoid signal loss and image Phase error ottleld (nhomogenetty T..... distortions resulting from inhomoge- and themlcal shift. neity-induced phase errors. All echo- it1 2 Si s2s3 s4s5s6 s7s8s9 planar imaging variants require strong magnetic gradients with fast rise times, which to date has limited RF&slgnal ir/2 JVJ’tAJ& JVJ’\lJ(J& their application to a handful of re- search centers. It is significant that the subsecond imaging time of echo-planar techniques eliminates artifacts (Read out) f\_/\ I\_[\ caused by both respiratory and car- ‘k_I diac motion. In this article, a novel technique of MR imaging, which combines gradi- Gy tt fi ent echo and spin echo (GRASE) tech- (Phase) vu vu vu niques, enables T2-weighted abdomi- nat imaging in tess than 18 seconds. This is sufficiently fast to permit Gz n n breath holding during image acquisi- (slice select) tions to eliminate respiratory-motion Figure 1. Pulse-sequence diagram of the GRASE technique. An echo train is formed with the artifacts, currently not possible in din- Carr-Purcell-Meiboom-Gill (CPMG) sequence with selective 180#{176}RE pulses (rr, ‘rr, . . .) and ical T2-weighted spin-echo studies of multiple signal readout gradient (Gx) reversals. With the multiple 180#{176}pulses, phase errors the abdomen performed with acquisi- due to field inhomogeneity and chemical shift are greatly reduced in magnitude and are peri- tion times of 4-8 minutes. odically reproduced throughout the echo train. In echo-planar imaging, with the absence of 180#{176}pulses there would be a large accumulation of phase errors and chemical shift (dashed line). The 24 echoes are phase-encoded differently by the gradient pulses in the phase-encoded direction (Gy). In each excitation cycle (TR), the strength of certain Gy pulses are MATERIALS AND METHODS changed (down-directed arrows) to offset the phase position in k space, while the pairs of negative Gy pulses maintain constant large jumps in the k-space trajectories. (The k-space lo- GRASE Technique cation of signals [sl-s9] is shown in Fig 2.) By using refocusing pulses (up-directed arrows), the net accumulated Gy phase shift is returned to zero before each 180#{176}RF pulse. Here, three The technique of combining gradient readout gradient reversals (NCR = 3) are shown, and only three of the eight RE refocused spin- echoes with spin echoes is shown in the echo periods (N5) are shown. pulse-sequence diagram (Fig 1). A train, or number of RF refocused spin echoes (NSE), is produced by using the CPMG spin-echo sequence. Centered about each spin echo, read (frequency) axis as a result of fre- small phase increments are imposed dun- a number of gradient-recalled echoes (NGR) quency sensitivity through the echo train. ing each 180#{176}-180#{176}interval along the total are produced by switching the polarity of In GRASE imaging, the chemical shift is echo train. Subsequent computer reorder- the readout gradient. The speed advan- proportional to the time interval of the ing or interleaving of the signals in k space tage over standard spin-echo imaging is short echo train with NCR echoes, whereas (Fig 2) effectively removes the peniodicity proportional to the total number of echoes in echo-planar imaging, the chemical shift of field inhomogeneity errors on the phase pen 90#{176}excitation and equals NGR X N5. is proportional to the longer time of the axis. The effective echo time of the image is the gradient-echo train composed of 64-128 In this way, GRASE phase encoding or- time at which the origin of k space (zero echoes. den is significantly different from that of order phase encode) is sampled, near the The GRASE pulse sequence is more RARE and echo-planar imaging, which middle of the echo train. complicated than a simple combination of continuously increase their phase-encod- In GRASE imaging, the use of 180#{176}RE spin echo with gradient echo, since the ing steps as a function of time in the echo pulses to nutate on reverse the position of relatively small field-inhomogeneity phase train (Fig 2). Similar to the hybrid RARE spins leads to the nulling of field inhomo- errors recur identically in each group of technique, multiple excitations are per- geneity errors at the Hahn spin-echo time NCR echoes between successive 180#{176}RE formed with small incremental phase

(8), eliminating image distortions and loss pulses, as shown in Figure 1 . If phase en- shifts to fill in additional areas of k space. of S/N. Small field-inhomogeneity errors coding is continuously incremented along Multiple sections are imaged during each evolve within each group of N(;R echoes the total echo train, as in echo-planar im- TR cycle of GRASE echo-train excitation. during the relatively short time between aging or the RARE technique, there would The acquisition time (T) of the GRASE each gradient-recalled echo and the center be modulation of chemical shift and T2* technique can be directly expressed as T = of their respective spin echo. In a similar on the phase axis of k space. After Fourier- [TR x (NL X NSA)/(NGR X N SE)] + TR, way, the modern echo-planar imaging transform image reconstruction, these where NL number of acquired phase- techniques, MBEST (6) and Instascan (7), modulations would result in severe ghost- encoded image lines. The additional TR is involve use of a single 180#{176}RE pulse, but ing artifacts in the image. required to establish steady-state magneti- with greater field-inhomogeneity error, This problem is eliminated with use of a zation. This initial excitation also produces proportional to half the time of the gradi- discontinuous phase-encoding order dun- a template of echoes without phase en- ent-echo train typically composed of 64- ing the echo train. As more fully described coding, which can be used for T2* magni- 128 signals. elsewhere (9), phase encoding sweeps tude normalization and gradient-echo Common to both GRASE and echo-pla- through three large increments of k space timing corrections in the phase-encoded nan imaging, chemical shift is greaten on during each group of NCR echoes. Holding echoes. In the presented imaging expen- the phase axis of the image than on the these large phase increments constant, ments in which NL 192, NcR 3, and

598 #{149}Radiology November 1991 ky

s4 s7 I III III s2 uu III S5& Ii’ II I

III I I I kx III s3 II L s6 I___ s9 L

a. b. c. Figure 2. The k-space trajectory of (a) the GRASE technique, (b) RARE technique, and (c) echo-planar imaging techniques, including MBEST and Instascan. Numbers on the left side in a-c correspond to the time order of the signal in the echo train. The phase-encoded signals (horizontal arrows) reverse their direction from right to left during the negative-polarity readout gradients (Fig 1). In the GRASE technique (a), the tra- jectony of the first group of three signals scans over nearly the entirety of k space, with identical trajectories for subsequent groups of signals except for a slightly displaced starting position between each group. In RARE imaging (b) and echo-planar imaging (c), the k-space trajectories are continuously displaced on the phase axis sequentially in time (thick arrows), unlike GRASE imaging. Multiple excitation cycles fill in k space by interleaving signals in both GRASE and hybrid RARE imaging. ky = phase axis, kx = frequency axis.

N5E 8, the total data acquisition time is 2 for the GRASE and RARE tech- 1.6 pixels and on the frequency axis (9 x TR x NSA). mques (Fig 3). In all three images, the (vertical) is 0.8 pixel. phase axis is horizontal. The contrast Typical tong-TR head studies (Figs Methods among gray matter, white matter, and 5, 6) obtained with a pulse sequence cerebrospmat fluid is similar in these of 4,000/104 and NSA = 2, require a Experiments were performed with a images. The signal intensity of skin in 72-second imaging time; those ob- 1.5-T MR system (Signa; GE Medical Sys- the RARE image (Fig 3c) is noticeably tamed with NSA = 1 require a 36- tems, Milwaukee) using maximal gradient more intense than that in either the second imaging time. Comparison of strength of 1 C/cm (10 mT/rn). The read- GRASE or spin-echo image due to a the image quality and S/N can be out periods were 2 msec on 4 msec with 0.6 msec gradient ramp times for a respective higher signal of lipid in the RARE im- made at similar brain levels in these 18- or 24-msec interval between each pair age. Greater flow artifact is present NSA = 1 and NSA = 2 studies. The of 180#{176}RE pulses. Free-induction-decay across the central region of the spin- image quality is sufficient to demon- spoiler pulses were used on the read-gra- echo image (Fig 3b). strate many small radial arteries in the dient axis and are not shown in Figure 1. Quantitative measurements of tis- neocortex. By using the GRASE tech- The effective echo time (TE) is the time at sue signal intensity (pixel intensity) nique with NSA = 1, a multiple scle- which the origin of the k space is sampled: obtained from the comparative head rosis plaque is readily demonstrated 80 on 104 msec. study involving the spin-echo, RARE, in the right frontal white matter (Fig With NGR 3 and NSE 8, there is a to- and GRASE techniques are provided 6b) of a patient with multiple sctero- tat of 24 signals per 90#{176}excitation to give exactly 192 phase-encoded signals in eight in the Table. Comparison of tissue sis. Direct comparison by a neuroradi- excitations. The 192 x 256 k-space data set contrast is somewhat affected by the ologist of the number of multiple scle- acquired in this way is symmetrically zero- different TE values. To obtain similar rosis plaques in T2-weighted spin- filled to 256 x 256 before two-dimensional S/N in images, acquisitions of GRASE echo images with that in GRASE Fourier-transform image reconstruction. A and RARE images involved use of images at the same levels demon- standard multisection excitation scheme NSA = 2 and a signal readout period strated an equal number of plaques in (1) was performed by using selective RE of 4 msec, whereas acquisition of the two patients studied. excitations with frequency offsets. spin-echo images involved use of Sagittal GRASE images of the lum- Head and body images of the same NSA = ‘/2 and a signal readout period bosacral spine (Fig 7) demonstrated healthy volunteer were obtained. Also, of 8 msec. Although estimates of im- the myetographic effect of cerebrospi- two patients with a known radiologic and clinical diagnosis of multiple sclerosis un- age S/N, calculated from the ratio of nat fluid with a TR of 3 seconds. De- derwent imaging with the GRASE tech- tissue signal intensity to air signal in- generative disk disease was noted in nique at the time of their routine T2- tensity, are similar among the three the lumbar spine. weighted spin-echo studies. techniques, the accuracy of measure- ment is affected and limited by van- DISCUSSION ability in tissue structure. RESULTS Images of the abdomen (Fig 4) were The development of computed tomo- To compare tissue contrast, brain obtained in 18 seconds, during a sin- graphic (CT) technology for abdominal images were obtained with the gle breath hold. Therefore, respiratory imaging, not unlike that of MR imaging, GRASE, spin-echo, and RARE tech- motion artifact is totally absent. The required a large leap in image acquisition speed to move from the early head imag- niques, each with similar pulse se- renal arteries branching from the ab- ing times of 3-5 minutes to the present quences: 2,500/100-104 (TR msec/TE dominal aorta are well demonstrated imaging times of 1-3 seconds. CT body msec). For this comparison, the S/N without respiratory motion artifact. imaging is now of great clinical utility, was intentionally set to be nearly The contours of the liver, spleen, and while the problems of respiratory motion equal by using conjugate synthesis kidney are sharply defined. Chemical have not been overcome in routine clinical (NSA = V2) for spin-echo and NSA = shift on the phase axis (horizontal) is MR imaging. To date, respiratory motion

Volume 181 #{149}Number 2 Radiology #{149}599 a. b. C. Figure 3. Comparison of tissue contrast in (a) GRASE, (b) spin-echo, and (c) RARE MR images of the brain. The three multisection images were acquired with a TR of 2.5 seconds and a section thickness of 5 mm. (a) The GRASE image was obtained with NSA = 2 and an image time of 45 seconds for 13 sections. (b) The spin-echo image was obtained with conjugate synthesis (NSA = ‘/2) and an image time of 4.6 minutes for 25 sections of both proton-density and T2-weighted images. (c) The hybrid RARE image was obtained with NSA = 2 and an image time of 64 seconds for 13 sections. RARE and spin-echo images were acquired with a TE of 100 msec, while the GRASE TE differed slightly (104 msec) due to sequence timing differences.

artifacts have significantly limited T2- weighted spin-echo imaging, which otherwise has great sensitivity to abdominal pathologic conditions. To eliminate respi- natony artifact, GRASE imaging times of 18 seconds permit comfortable breath holding for a large portion of the patient popu- lation. Due to the discontinuous phase-encoding method of GRASE imaging, Hahn spin echoes coven the entire central third portion of k space where the highest signal energy occurs (8). For this reason, the tissue contrast of GRASE, in theory and practice, is indistinguishable from that of spin-echo imaging, determined predomi- nantly by these central gradient echoes that in fact are spin echoes. Given their similar contrast mechanism, GRASE and spin-echo imaging demonstrate multiple sclerosis plaques equally well, unlike low- flip-angle gradient-echo imaging, which is the GRASE technique more efficiently uses technique can be given in a simple expres- suboptimal for detection of some patho- the T2 relaxation period to refocus or sion for the average time of each phase-

logic conditions. ,, pump out” more signals pen acquisition encoded signal. The average time of a As demonstrated in the head studies time, this factor does not directly cause phase-encoded echo can be directly calcu- (Fig 3, Table), tissue contrast in GRASE S/N loss unless shorter read periods with lated by using exemplary time values; the images is essentially the same as that in larger signal bandwidths are used. Conju- selective 180#{176}RF nefocusings and sun- spin-echo images obtained with similar gate synthesis spin-echo imaging, not hay- rounding free-induction-decay spoiler TRs and TEs. The abdominal tissues (Fig ing the time constraint of many signal read gradients (T51) equal 5 msec; the net time 4), including liver, spleen, kidney, muscle, periods in each excitation, permitted twice of the initial phase-encoding pulse and and fat have the expected spin-echo T2- as long a read period as the GRASE and final rephasing pulse (TPE) equals 4 msec; a weighted contrast. While breath holding RARE techniques in these experiments, 4-msec echo readout period plus gradient effectively eliminates much of the motion- compensating for its intrinsic 35%-41 % rise times (TRw) equals 5.2 msec. Average related artifact, some residual artifact re- S/N loss. The demonstrated image quality time per echo = [TRI + TPE + (NCR X T5))]/ mains because of pulsations in the liver in head and body studies obtained with N,5. and adjacent structures. Pulse sequence the GRASE technique with NSA = I is For RARE imaging with NCR = L the modifications to reduce cardiac pulsation suitable for routine clinical imaging. average time per signal is (5 + 4 + 5.2)/I, effects and other related artifacts are cur- It is significant that GRASE imaging in- or 14.2 msec. For GRASE imaging with nently being investigated (10). volves use of about a third the number of NGR _3, the average time per echo is [5 + The primary determinant of S/N in 180#{176}RF pulses per multisection acquisition 4 + (3 x 5.2)]/3, on 8.2 msec. For GRASE GRASE, RARE, and spin-echo imaging is that are used in a similar RARE acquisi- imaging with N(;5 = 7, the average time signal bandwidth (S/N is proportional to tion, which has a higher SAR. The speed per echo is 6.5 msec. Although the effi- the square root of signal read time). While advantage of the GRASE over the RARE ciency of GRASE imaging is currently in-

600 #{149}Radiology November 1991 a. b.

Figure 4. Abdominal MR images. (a, b) Coronal studies of the upper abdomen obtained during single breath holds. The total image time was 18 seconds for 11 multisection images, pulse sequence was 2,000/80, section thickness was 10 mm, and field of view was 38 cm. (c, d) Axial images obtained with parameters identical to those of the coronal images except for a TE of 90 msec.

C. d.

a. b. C. Figure 5. (a-c) GRASE MR images obtained in the head of a healthy volunteer with NSA = 2. Imaging time was 72 seconds, pulse sequence was 4,000/104, field of view was 24 cm, and section thickness was 5 mm.

termediate between echo-planar and of gradient refocusing, its large time over- than in these echo-planar techniques, RARE imaging, application of the GRASE head in 180#{176}RF pulses, and SAR limita- which necessarily have a much longer technique with an imaging system with tions. 90#{176}-180#{176}time interval when no signals can high performance gradients (fast rise At first glance, in comparing GRASE be acquired. This advantage of GRASE times and strong maximal gradient) will with MBEST and Instascan, the additional imaging may largely offset the efficiency further increase the efficiency of GRASE time cost of the multiple 180#{176}RE refocus- differences between these techniques for imaging, making it closer to that of echo- ings certainty appears to directly reduce T2-weighted imaging. planar imaging. The RARE technique can- the efficiency of GRASE imaging. These It is important to realize that GRASE not benefit nearly as much from high per- additional nefocusings, however, permit imaging is a variant of spin-echo imaging formance gradients because of the absence signal acquisition to occur much earlier that involves use of the CPMG sequence

Volume 181 #{149}Number 2 Radiology #{149}601 Figure 7. Sagittal GRASE images of the spine obtained with a TR of 3 seconds, NSA =2, and section thickness of 5 mm. Image time was 56 seconds. Note the myelographic effect of cerebrospinal fluid. Degenerative disk disease is demonstrated at multiple disk levels, with more severe central disk bulging at the L2-3 level and loss of signal intensity a. b. in the L4-5 disk. Figure 6. GRASE MR images obtained in the head with NSA = 1, image time of 36 seconds, and pulse sequence of 4,000/104. (a) Image obtained in healthy volunteer and (b) image ob- tamed in a patient with multiple sclerosis. The image quality and S/N can be compared with those of brain images obtained with NSA = 2 in Figure 5, which were otherwise acquired with 2. denBoefJH, van Uijen CMJ, Holzcheres identical image parameters. CD. Multiple-slice NMR imaging by three-dimensional Fourier zeugmatogra- phy. Phys Med Biol 1984; 29:857-867. 3. Feinberg DA, HaleJD, Watts JC, Kauffman L, Mark A. Halving MR imaging time by to produce spin-echo contrast. GRASE im- and spatial resolution with a 24-fold re- conjugation: demonstration at 3.5 kG. Ra- aging lies on a continuum between RARE duction in image acquisition time. The diology 1986; 161:527-531. imaging, which involves use of multiple time advantage of GRASE over spin-echo 4. Hennig J, Nauerth A, Friedburg H. RARE spin echoes, and single-shot echo-planar imaging is accomplished by more effi- imaging: a fast imaging method for clinical imaging, which involves use of multiple ciently using the T2 decay period. A pen- MR. Magn Reson Med 1986; 3:823-833. gradient echoes. In GRASE imaging, a alty of fewer image sections in the current 5. Mansfield P, Maudsley AA. Planar spin imaging by NMR. J Magn Reson 1977; 27: small amount of chemical shift is ac- implementation of the GRASE technique 101-107. cepted with the decrease in imaging time can be optimized with tradeoffs in T2 6. Ordidge RJ, Howseman A, Coxon R, et al. and SAR afforded by the use of gradient weighting and imaging time. Combination Snapshot imaging at 0.5 T using echo-pla- echoes. of the GRASE technique with high-penfor- nan techniques. Magn Reson Med 1989; Several obvious modifications of GRASE mance gradient systems should produce 10:227-240. imaging are possible: division of the echo further large reductions in imaging time 7. Rzedzian RR, Pykett IL. Instant images of train into two identically phase-encoded while maintaining image quality and spa- the body by magnetic resonance. Magn data sets for simultaneous proton-density tial resolution. The ability to implement Reson Med 1987; 5:563-571. 8. Hahn EL. Spin echoes. Phys Rev 1950; and T2-weighted (double-echo) imaging, the GRASE technique without the need to 50:580-594. use of different TEs and TRs for desired change gradient hardware, however, is a 9. Oshio K, Feinberg DA. GRASE (Gradient image contrast; extension to multislab significant advantage of this new high- and Spin Echo): a novel fast MR imaging three-dimensional volume imaging; and speed imaging technique. U technique. Magn Reson Med 1991; 20:344- use of 512 x 512 high-resolution imaging 349. in clinically acceptable imaging times. Acknowledgment: We are grateful to Ferenc 10. Feinberg DA, Oshio K. Gradient echo Methods of performing diffusion and flow Jolesz, MD, for his assistance in comparing neu- time shifting in fast imaging (abstr). In: imaging with the CPMG sequence have rologic imaging studies and for his comments in Book of abstracts: Society of Magnetic Res- been described (II). Another possible ap- preparation of this manuscript. We are very onance in Medicine 1991. Berkeley, Calif: plication of GRASE imaging in systems grateful to Lawrence E. Crooks, PhD, for his edi- Society of Magnetic Resonance in Medi- tonal comments. cine, 1991; 1239. with low magnetic field strength could be 1 1. Feinberg DA, Jakab PD. Tissue perfusion to perform imaging with NSA = 4 or 8 to in humans studied by Fourier velocity dis- increase image S/N in clinically acceptable References tnbution, line scan, and echo-planar imag- imaging times. ing. Magn Reson Med 1990; 16:280-293. 1. Crooks L, Arakawa M, HoenningerJ, et al. In summary, multisection GRASE imag- Nuclear magnetic resonance whole-body ing is similar to long-TR spin-echo imag- imager operating at 3.5 KGauss. Radiology ing in terms of image fidelity, contrast, 1982; 143:169-174.

602 #{149}Radiology November 1991