Three-Dimensional Spin-Echo-Train Proton-Density-Weighted Imaging
`Using Shaped Signal Evolutions
`
`JP Mugler III1, S Bao2, RV Mulkern2,3, CRG Guttmann2, FA Jolesz2, JR Brookeman1
`Departments of Radiology,1University of Virginia School of Medicine, Charlottesville, VA, USA,
`2Brigham and Women’s Hospital and3Children’s Hospital, Harvard Medical School, Boston, MA, USA
`
`Figure 2 presents sagittal, coronal, and axial proton-
`density-weighted images reconstructed at 1.4 mm thickness
`from a 10-minute 3D acquisition of the whole head. The flip-
`angle series of Fig. 1 was used for the refocusing RF pulses.
`The direction corresponding to the shaped-signal evolution
`is left-right in the axial image. The image contrast is compa-
`rable to that for conventional-SE proton-density-weighted
`images, and despite the long echo train length no substantial
`blurring is evident. Pulse sequence parameters included:
`TR/TE, 2400/20 ms; echo train length, 48; echo spacing,
`4 ms; matrix, 148 x 144 x 256, FOV, 20 x 16.5 x 22 cm.
`
`150
`
`120
`
`90
`
`60
`
`30
`
`0
`
`FLIP ANGLE (degrees)
`
`0
`
`40
`30
`20
`10
`REFOCUSING PULSE NUMBER
`FIG 1: Refocusing-RF-pulse flip angle as a function of echo
`position for the shaped signal evolution discussed in the text.
`
`50
`
`FIG 2: Proton-density-weighted sagittal, coronal, and axial
`images, 1.4 mm thick, from a 10-minute 3D acquisition of the
`whole head.
`
`CONCLUSIONS
`Shaped signal evolutions have been successfully
`incorporated into a single-slab 3D proton-density-weighted
`pulse sequence and image sets of the whole brain, with a
`volume resolution of 1.3 mm3, have been obtained in an
`acquisition time of 10 minutes. These results suggest the
`possibility of incorporating preparatory RF and/or gradient
`pulses into the 3D sequence to yield high-resolution 3D
`spin-echo data sets with additional contrast behaviors such
`as strong T1-weighting (inversion preparation) or diffusion.
`REFERENCES
`1. Hennig J, Nauerth A, Friedburg H. Magn Reson Med 1986; 3:823-833.
`2. Melki PS, Jolesz FA, Mulkern RV. Magn Reson Med 1992; 26:328-341.
`3. Mugler III JP, Epstein FH, Brookeman JR. Magn Reson Med 1992;
`28:165-185.
`4. Le Roux P, Hinks RS. Magn Reson Med 1993; 30:183-190.
`5. Mugler III JP, Brookeman JR, Mulkern RV, et al. 6th ISMRM; 1998, 1959.
`
`This work was supported by NIH grant NS-35142.
`
`INTRODUCTION
`Turbo / fast spin-echo (SE) pulse sequences [1,2] are
`widely used for T2-weighted imaging. The long effective
`echo time (TEeff) required for T2 weighting permits concom-
`itantly long echo trains in conjunction with a k-space weight-
`ing that yields minimal
`image artifacts. For short TEeff
`values, however, as required for T1- or proton-density-
`weighted imaging, similarly long echo trains result in sub-
`stantial image blurring because the center of k space must
`be sampled early in the echo train. To avoid this blurring
`short echo trains can, of course, be used, but this then
`lengthens the acquisition time, approaching that for a con-
`ventional SE pulse sequence.
`An alternative that may permit long echo trains to be
`used with short TEeff values is to prospectively shape the
`signal evolution [3,4] using refocusing-RF-pulse flip angles
`that vary as a function of the position in the echo train. Uni-
`form signal evolutions can, at least in theory, be achieved,
`thus eliminating the blurring problem. We have incorporated
`this variable-flip-angle strategy into a single-slab three-
`dimensional (3D) pulse-sequence structure [5] and devel-
`oped a technique for 3D proton-density-weighted imaging.
`MATERIALS AND METHODS
`The flip-angle series required to yield the desired sig-
`nal evolutions during the spin-echo train were derived using
`a computer-based theoretical model
`implemented on an
`Ultra-60 workstation (Sun Microsystems Inc., Mountain
`View, CA). The signal-shaping algorithm was based a previ-
`ously published method [3]. The theoretical model was used
`to explore the maximum attainable signal levels as a func-
`tion of the signal evolution shape, pulse-sequence charac-
`teristics, and tissue relaxation times.
`The 3D proton-density-weighted pulse sequence incor-
`porating the variable-flip-angle refocusing-RF pulses was
`implemented on a 1.5 T commercial whole-body imager
`(Vision, Siemens Medical Systems, Iselin NJ). Images were
`acquired of doped-water phantoms and of the heads of vol-
`unteers, after obtaining informed consent.
`RESULTS
`The maximum signal level attainable from a given tis-
`sue decreased as the echo-train length or echo spacing
`increased. In contrast, the maximum signal level increased
`roughly in proportion to the degree to which the signal evo-
`lution was allowed to decrease during the echo train. For a
`given flip-angle series, the signal evolution varied somewhat
`among tissues, but all signal evolutions corresponding to the
`range of biological relaxation times were similar.
`These results led us to select a signal evolution that
`decreased rapidly for the few echoes, was approximately
`uniform for 70% of the total duration, and then decreased
`slowly over the remaining echoes. The rapid signal
`level
`decrease at the beginning of the signal evolution occurred
`during the first few (2-4) echoes which were discarded to
`allow the signal from flowing blood to decay to a negligible
`level. The flip-angle series corresponding to this signal evo-
`lution is shown in Fig. 1.