| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Original Report |
1 Both authors: Department of Radiology and Radiological Sciences, The Johns Hopkins Hospital, The Johns Hopkins Medical Institutions, 600 N. Wolfe St., Baltimore, MD 21287-2182.
Received June 8, 2000;
accepted after revision July 14, 2000.
Address correspondence to E. R. Melhem.
Abstract
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CONCLUSION. Thresholds for T1 relaxation times increased as TE increased during FLAIR MR imaging sequences. Such thresholds defined the transition from hyperintense to hypointense lesions.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
We studied interactions between the T1 relaxation time of the lesion and the TE of the FLAIR MR imaging sequence. Using computer-generated brain maps, we aim to define T1 relaxation time thresholds above which a T2 hyperintense lesion becomes hypointense on FLAIR MR images. Thresholds are identified for FLAIR MR imaging sequences with different TEs.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
A mixed multiecho spin-echo and inversion recovery MR sequence was used to image the brain of a 40-year-old male volunteer at the level of the lateral ventricles. The mixed sequence simultaneously provided two image data sets: eight spin-echo (TR/averages, 1500/1) images with different TEs (20, 40, 60, 80, 100, 120, 140, and 160 msec); and eight inversion recovery images (TR/TI/averages, 2000/400/1) with different TEs (20, 40, 60, 80, 100, 120, 140, and 160 msec). The image slice thickness was 5 mm, the in-plane resolution was 0.86 x 0.86 mm2 (field of view, 220 mm; matrix, 256 x 256), and the acquisition time was 9 min 30 sec.
Generation of Brain Maps
The generation of brain maps involved two steps: First, pixel-by-pixel
T1-relaxation, T2-relaxation, and proton density brain maps at the level of
the lateral ventricles were generated online, using ratios and least-squares
algorithm (software release 6.2, Philips Medical Systems) applied to the image
data sets from the mixed MR sequence
[8]. Second, calculated maps
simulating T2-weighted and FLAIR MR imaging sequences were generated off-line
(SUN Enterprise 5500; Sun Microsystems, Mountain View, CA) using proton
density, T1-relaxation, and T2-relaxation pixel values from the corresponding
maps.
The following numeric calculations and image displays were performed using Interactive Data Language software (Research Systems, Boulder, CO).
Each pixel value S(x,y) of the calculated map was calculated using
the following equation:
 |
T1(x,y), T2(x,y), and 
(x,y) were derived
from corresponding coordinate pixel values of the T1-relaxation,
T2-relaxation, and proton density maps, respectively.
TR, TE, and TI for the simulated sequences were selected based on the parameters used in clinical brain imaging.
For T2-weighted MR imaging sequences, TR, TE, and TI were 3000, 90, and 0 msec, respectively. For FLAIR MR imaging sequences, TR was 11,000 msec, TE was variable, and TI was 2600 msec.
For the FLAIR MR imaging sequence six different maps were generated using TEs of 20, 40, 60, 100, 140, and 180 msec.
Creation of Simulated Lesions
Using an in-house program written in Interactive Data Language, we placed
12 simulated lesions in the T2-weighted and all the FLAIR maps except the one
with a TE of 180 msec (Fig. 1).
Seventeen simulated lesions were placed in the FLAIR map with a TE of 180
msec.
|
The simulated lesions were round and approximately 6 mm (7 pixels) in diameter. They were placed in the white matter of the cerebral hemisphere parallel to the lateral ventricle.
The T2 relaxation time and proton density (relative to adjacent white matter) for every pixel within the lesions were fixed at 300 msec and 1.07, respectively. The long T2 relaxation time of the simulated lesions (compared with white matter) was chosen to reflect common brain abnormalities such as demyelination and gliosis [4, 9]. For the T2-weighted map and all the FLAIR maps except the one with a TE of 180 msec, the T1 relaxation time of the lesions was varied from 800 msec (posteriormost lesion) to 3000 msec (anterior-most lesion) by increments of 200 msec. For the FLAIR map with a TE of 180 msec, the T1 relaxation time of the lesions was varied from 800 to 4000 msec by increments of 200 msec.
The writing and refinement of the program used for the generation of the different maps and simulated lesions took a moderately experienced programmer approximately 1 month. Once the program was operational, it took approximately 30 sec to generate each map and simulated lesion.
T2-weighted and six different FLAIR maps were reviewed by an experienced observer. The T1 relaxation time (isointense T1) at which the simulated lesion became isointense to surrounding white matter (transition from hyperintense to hypointense) was documented for each of the FLAIR maps. A plot of isointense T1 versus TE was generated (Fig. 2).
|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
All simulated lesions were hyperintense to white matter on the T2-weighted maps. There was, however, a decrease in contrast between the simulated lesions and white matter with increasing T1 relaxation time (Fig. 1).
On all the FLAIR maps, the simulated lesions transformed from hyperintense to hypointense compared with white matter, once the T1 relaxation time increased beyond a certain threshold (Fig. 1). The T1 relaxation time threshold increased with the TE of the FLAIR MR imaging sequence (Fig. 2).
As a result of this upward shift in the T1 relaxation time threshold, lesions with the same T1 and T2 relaxation times could appear hypointense on short-TE FLAIR and hyperintense on long-TE FLAIR MR imaging (Fig. 1). Also, if a lesion had a T1 relaxation time that matched that defined by a particular FLAIR MR imaging sequence, the lesion was isointense to white matter (invisible) on the FLAIR MR imaging sequence despite being hyperintense on T2-weighted MR imaging sequences (Fig. 1).
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
On the T2-weighted maps, all lesions are hyperintense to white matter. The observed gradual decrease in lesion contrast with increasing T1 relaxation time is caused by saturation of longitudinal magnetization imposed by the selected TR (3000 msec) (T1 effect). Longer TRs (on the order of three to four times the selected T1 relaxation time of the simulated lesion) will eliminate the effect of saturation and render lesion contrast predominantly determined by the selected T2 relaxation time (T2 effect).
In FLAIR MR imaging, as in all inversion-recovery sequences, T1 relaxation time influences image contrast by affecting the degree of longitudinal magnetization recovery during the T1 and TR-TI intervals. As the T1 relaxation time of a lesion increases, the degree of longitudinal magnetization recovery decreases and the signal intensity from the lesion decreases. At short TEs, the T1 effect is emphasized and the lesion appears hypointense to surrounding white matter. At long TEs, on the other hand, the T2 effect is emphasized and the lesion appears hyperintense [7].
In our FLAIR MR imaging simulations, near-complete recovery of longitudinal magnetization during the TR-TI interval is achieved by choosing a long TR (11,000 msec) and disregarding the influence of the commonly implemented fast spin-echo readout [10]. Thus, the degree of recovery during the TI interval (2600 msec) in combination with the T2 effect determines lesion contrast.
We have shown using computer-simulated brain maps that several combinations of intrinsic lesion parameter (T1 relaxation times) and FLAIR MR imaging pulse parameter (TE) exist in which T2-hyperintense lesions are isointense to white matter on FLAIR MR imaging. This may partly explain reported differences in the performance of FLAIR compared with T2-weighted MR imaging sequences [11, 12] and underscores the importance of implementing multiecho FLAIR MR imaging in routine practice.
Another potentially important role for multi-echo FLAIR MR imaging is in the evaluation of multiple sclerosis lesions. The respective volumes of hypointense multiple sclerosis lesions, as shown by T1-weighted and long-TE FLAIR MR imaging, correlate strongly and moderately, respectively, with clinical disability [6]. The difference in the strength of the correlation is attributed to the inability of long-TE FLAIR MR imaging to show mildly damaged lesions, (i.e., multiple sclerosis lesions with mild prolongation of T1 relaxation times) [6]. The multiecho FLAIR MR imaging sequence may improve the visibility of mildly damaged lesions as hypointense lesions on short-TE FLAIR MR imaging, help differentiate mildly from severely damaged lesions, and strengthen the correlation between the volume of hypointense lesions and clinical disability. Initial experience with cerebral adrenoleukodystrophy, a rare disease affecting the white matter, shows the ability of short-TE FLAIR MR imaging to separate between mildly (peripheral) and severely damaged (central) zones (Fig. 3).
|
One limitation of this model is that the T1, T2, and proton density pixel values used to generate the FLAIR and T2-weighted MR images were obtained from a single patient. The restricted nature of the database may influence the definition of the T1 relaxation time thresholds.
Lastly, in this work, we present a computer-based tool that allows us to generate brain maps of any pulse sequence and that provides the freedom to choose the location, size, shape, and intrinsic characteristics of simulated lesions. We believe that this tool will be useful for testing the effect of pulse sequence optimization on the visibility of brain lesions.
In conclusion, T1 relaxation time thresholds increase with TE of the FLAIR MR imaging sequence and define the transition from hyperintense to hypointense lesions.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
