| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Technical Innovation |
1 Department of Radiology, National Institutes of Health, Bldg. 10, Rm. 1C-640,
10 Center Dr., MSC 1182, Bethesda, MD 20892-1182.
2 Siemens Corporation, Iselin, NJ 08830.
Received October 17, 2001;
accepted after revision March 21, 2002.
Presented in part at the annual meeting of the American Roentgen Ray
Society, Washington, DC, May 2000.
Introduction
 Top Introduction Subjects and MethodsResultsReferences |
|---|
We describe a scheme in which radiofrequency prepulses are placed outside the slice phase-encoding loop for 3D gradient-echo MR imaging. We illustrate the manner in which this segmented prepulsing strategy can affect fat saturation or magnetization transfer contrast [1] without compromising scanning speed.
|
TopIntroductionSubjects and MethodsResultsReferences |
|---|
A magnetization transfer pulse was placed outside the slice phase-encoding loop. This pulse consisted of a 7.7-msec Gaussian pulse centered 1.5 kHz downfield from the water resonance. The total preparatory time, including spoiler gradients, was 12 msec (duty cycle, <3%). The magnitude of the magnetization transfer pulse (1500-2500°) was chosen to adhere to a local specific absorption rate limit of 8 W/kg.
Spectral fat saturation was also implemented using segmented prepulses. The fat-saturation pulse was 9.2 msec in duration and 120° in magnitude and was centered 220 Hz upfield from the water peak. The preparatory time for spectral fat saturation, including spoiler gradients, was 21 msec.
Sequence testing was performed on agar, peanut oil, and doped water phantoms. After obtaining approval from our institution's internal review board, we performed brain imaging with and without segmented magnetization transfer prepulses on four male volunteers (age range, 29-44; mean age, 37 years). A quadrature head coil was used for radiofrequency transmission and reception. In the oblique axial plane, MR imaging was performed with the following parameters: field of view, 18 x 24 cm; imaging matrix, 144 x 256; slice section, 5 mm interpolated to 2.5 mm; reconstructed slices, 44; magnitude of magnetization transfer pulse, 2000°; and imaging flip angle, 14°.
The 3D segmented prepulse technique was implemented for MR arthrography of the shoulder. Five subjects (age range, 24-45; mean age, 33 years) were imaged after the intraarticular injection of dilute (2.5 mmol) gadopentetate dimeglumine (Magnevist; Schering, Berlin, Germany) in saline. A quadrature-detection flexible surface coil was used for signal reception. Imaging was performed in the axial plane with the following parameters: field of view, 16 cm; section thickness, 2.2 mm interpolated to 1.1 mm; reconstructed slices, 64; imaging matrix, 256 x 256; number of signals averaged, 2; magnitude of magnetization transfer pulse, 1800°; and imaging flip angle, 12°. Acquisitions in these subjects were performed with segmented fat-saturation prepulses alone and with both magnetization transfer and fat-saturation prepulses.
Regions of interest were measured to calculate the signal-to-noise ratio, contrast-to-noise ratio, and magnetization transfer ratio for specific tissues. Each magnetization transfer ratio was calculated using the following formula where MT is magnetization transfer and S is signal intensity: [S (MT) / S (no MT)]. Image noise was estimated using the standard deviation of the background signal. The magnetization transfer ratio for each pixel was calculated to generate magnetization transfer ratio images using IDL software (Research Systems, Boulder, CO). In these calculations, pixels with a magnetization transfer ratio of greater than 1 or with raw signal values less than the mean background signal level were set to 1.
|
TopIntroductionSubjects and MethodsResultsReferences |
|---|
Figure 1A,1B shows representative 3D gradient-echo MR images of the brain that were obtained with and without prepulsing. The mean magnetization transfer ratios in the white and gray matter of the volunteer subjects were 82% (SD, ±3%) and 89% (SD, ±4%), respectively.
|
|
Figure 2A,2B shows a representative MR arthrogram of the shoulder. The scanning duration for the 3D acquisition with both segmented magnetization transfer and fat-saturation prepulses was 2 min 24 sec. For the images obtained without magnetization transfer (not shown), the scanning duration was 2 min 18 sec. The scanning duration without segmented prepulses would be 2 min 7 sec.
|
|
Figure 2B is a magnetization transfer ratio image that shows the shoulder anatomy by revealing the varying magnetization transfer contrast behavior of different tissues. The mean magnetization transfer ratio in muscle for the subjects was 78% (SD, ±3%), whereas the injected fluid exhibited no significant magnetization transfer contrast. The magnetization transfer ratios were more variable in the glenoid labrum (mean ± SD, 68% ± 16%).
In conclusion, although performing 3D MR imaging with magnetization transfer pulses can effectively produce magnetization transfer contrast [2], magnetization transfer pulses considerably prolong scanning duration. Reductions in scanning time can be achieved using multishot two-dimensional gradient-echo MR imaging by placing the magnetization transfer pulses outside a segmented phase-encoding loop [3]. Although the intermittent application of the magnetization transfer pulses reduces the specific absorption rate, a reduction in the magnetization transfer contrast is expected [4].
The images that we obtained using this technique illustrate how prepulses placed outside the slice partitions loop in 3D gradient-echo acquisitions can also affect magnetization transfer contrast. Any artifact caused by modulation of magnetization between magnetization transfer pulses is likely to be less problematic when propagated across slices rather than in the imaging plane. Interpolating the phase encoding of slice partitions in 3D imaging with segmented prepulses further minimizes the duration between prepulses. The effect of interpolation on slice profile varies depending on any asymmetry in interpolation and on the spatial and contrast characteristics of the tissue [5]. The optimal interpolation scheme for 3D imaging with segmented prepulses warrants further study.
Magnetization transfer not only reduces the visible magnetization of particular tissues, but also shortens the T1 of the visible or free proton pool, which warrants special consideration in rapid 3D gradient-echo imaging for which the TR is very short. The flip angle influences magnetization transfer contrast to a greater degree in 3D imaging than in two-dimensional multislice imaging. How this phenomenon influences the utility of this technique for contrast-enhanced MR studies warrants further study.
Our experience with 3D gradient-echo MR imaging has not revealed artifacts associated with segmented prepulses. Experience with fast spin-echo imaging has highlighted the compromises in spatial resolution or image quality that may result from nonlinear sampling of k-space [6]. Segmented, pulsed magnetization transfer modulates magnetization as a function of the T1 of the restricted proton species and the rate of exchange between the free and restricted proton pools; these influences are typically smaller than those related to T2 decay in fast spin-echo imaging.
Magnetization transfer contrast generated by segmented prepulses might be improved with different magnetization transfer pulse profiles or schemes. Our preliminary experience suggests that for given specific absorption rate limitations, multiple sequential magnetization transfer pulses of correspondingly lower magnitude do not improve magnetization transfer contrast. On-resonance pulsed magnetization transfer might generate magnetization transfer contrast more effectively for a given specific absorption rate level [7], but would place greater demands on field homogeneity and may be less robust for body and extremity applications.
Prepulses placed outside the slice phase-encoding loop can alter tissue contrast in rapid 3D gradient-echo MR imaging at a negligible cost in acquisition time. Adaptation of this segmented prepulse strategy to steady-state gradient-echo sequences may broaden the utility of this strategy. As it stands, the segmentation of prepulses as described here may benefit many contrast-enhanced gradient-echo MR imaging applications.
Acknowledgments
We thank Jeffrey Stanczak for assisting in data collection, Shirley Bada
for her patience in sequence testing, and Shin Shou Lin for his computer and
network support.
|
TopIntroductionSubjects and MethodsResultsReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
