Three-Dimensional T2-Weighted Imaging of the Brain Using Very Long Spin-Echo Trains
`
`JP Mugler III1, B Kiefer2, JR Brookeman1
`1Department of Radiology, University of Virginia School of Medicine, Charlottesville, VA, USA
`2Siemens Medical Engineering Group, Erlangen, Germany
`
`lesions seen in the 2D image appear, to varying degrees, in
`three adjacent 1-mm sections. Furthermore,
`the overall
`image quality for the very long SE-train and conventional-SE
`images is similar, despite the much thinner sections of the
`former. Figures 2e and f, depicting the largest lesion in sag-
`ittal and coronal orientations, respectively, demonstrate the
`capability for high-quality images in arbitrary orientations.
`
`160
`120
`80
`40
`0
`REFOCUSING RF−PULSE NUMBER
`
`c
`
`f
`
`100
`
`80
`
`60
`
`40
`
`20
`
`0
`
`FLIP ANGLE (degrees)
`
`b
`
`e
`
`FIG 1: Variable flip-
`angle series for the
`160-echo 3D images
`shown in Fig. 2.
`
`a
`
`d
`
`FIG 2: T2-weighted (a) 2D and (b-f) 3D SE images from a volun-
`teer with non-specific white-matter lesions (arrows). The three 1-
`mm thick 3D images in (b)-(d) correspond to the single 3-mm thick
`2D image in (a). In the 3D images, the phase-encoding direction
`corresponding to the 160-echo train is left-to-right in (b)-(d) and (f);
`no image artifacts secondary to this very long SE-train are appar-
`ent. For the 10 min. 3D acquisition, parameters were TR/effective
`TE, 2750/328 ms; matrix, 256 x 160 x 216; FOV, 25.6 x 16.0 x 21.6
`cm; echo spacing, 4.1 ms; ETL, 160. For the 14.8 min. 2D acqui-
`sition, parameters were TR/TE1/TE2, 2750/20/80 ms; matrix, 256
`x 160; FOV, 25.6 x 16.0 cm; thickness, 3.0 mm; sections, 54.
`
`CONCLUSIONS
`Very long SE trains with prescribed signal evolutions
`permit brain imaging with both adequate S/N and useful con-
`trast properties, and thus provide a vehicle for substantially
`reducing the imaging time. For example, a half-Fourier acqui-
`sition (as described in [3]) combined with the very long SE-
`train T2-weighted method (Fig. 2) could provide 1-mm isotro-
`pic resolution of the whole brain in about 5 minutes, and a 3D
`single-slab FLAIR version [4] could provide 3-mm contigu-
`ous sections of the whole-brain in less than 5 minutes.
`REFERENCES
`1. Hennig J. J Magn Reson 1988; 78:397.
`2. Alsop DC. Magn Reson Med 1997; 37:176.
`3. Mugler III JP, Brookeman JR, et al. 6th ISMRM; 1998, 1959.
`4. Mugler III JP, Brookeman JR, et al. 7th ISMRM; 1999, 8.
`This work was supported by NIH grant NS-35142.
`
`INTRODUCTION
`Spin-echo trains used in clinical fast-SE-based imaging
`generally employ high flip angles (>100˚) for the refocusing
`RF pulses. The echo-train durations are typically less than
`the T2s of interest for short effective-TEs or less than two to
`three times these T2s for long effective-TEs. For brain imag-
`ing at 1.5T, these limits yield echo-train durations of <100ms
`and <300ms, respectively. Longer durations can degrade
`image contrast and cause artifacts such as blurring.
`The use of low-flip-angle refocusing pulses has been
`proposed as a strategy for accelerating SE-train-based acqui-
`sitions by lengthening the usable duration of the echo train [1].
`Alsop extended this concept by deriving variable flip-angle
`series based on the pseudosteady-statecondition of a con-
`stant signal level when T1 and T2 relaxation are neglected [2].
`Two-dimensional, T2-weighted brain images were acquired
`using an 80-echo train with a duration of 400 ms [2].
`We have investigated the potential of very long SE trains
`based on prescribed signal evolutions which explicitly con-
`sider the T1s and T2s of interest. Using the resulting variable-
`flip-angle RF-pulse series, we achieved T2-weighted single-
`slab 3D imaging of the brain with effective-TEs and echo-train
`durations of greater than 300 and 600ms, respectively.
`MATERIALS AND METHODS
`Using a computer-based theoretical model, variable-
`flip-angle refocusing RF-pulse series were calculated for
`several prescribed signal evolutions, including the following
`evolution for gray matter at 1.5T: exponential decay for the
`first 20 echoes (decay constant 114 ms), constant for 66 ech-
`oes, and exponential decay for the remaining echoes (decay
`constant 189 ms); 160 echoes with 4.1ms echo spacing.
`This variable-flip-angle series was implemented in a 3D
`single-slab T2-weighted fast-SE-based pulse sequence,
`adapted from previously-described techniques [3]. Imaging
`was performed on a 1.5 T whole-body imager (Symphony,
`Siemens Medical Systems).
`Images of
`the head were
`acquired in volunteers after obtaining informed consent. The
`performance of the 3D T2-weighted technique was also com-
`pared to that for a 2D T2-weighted conventional-SE sequence.
`RESULTS
`Figure 1 shows the calculated variable-flip-angle series
`for the gray matter signal evolution described above. All of the
`flip angles are less than 100˚, introducing a strong T1 depen-
`dence which can thereby lengthen the usable duration of the
`echo train substantially beyond the T2 value (~100ms).
`Figure 2 compares T2-weighted 2D and 3D images
`from a 59 year-old subject with age-related non-specific
`white-matter lesions. The very long SE-train images (Figs.
`2b-f) display high contrast between the lesions and sur-
`rounding white matter, suggesting that this echo train may
`provide clinically useful contrast characteristics that appear
`very similar to those for conventional T2-weighed SE images
`(Fig. 2a). However, as evidenced by the exceptionally long
`effective TE of 328ms, the associated contrast behavior dif-
`fers from that of established echo-train techniques and
`requires further investigation. The thin 1-mm sections pro-
`vide an improved definition of lesion location and extent; the