| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Original Report |
1 All authors: Department of Radiology, Duke University Medical Center, Box 3808, Durham, NC 27710.
Received April 29, 2004;
accepted after revision June 30, 2004.
Address correspondence to J. M. Provenzale.
Abstract
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CONCLUSION. In this small preliminary study, dynamic susceptibility perfusion MRI allowed distinction of cerebral abscesses from rim-enhancing high-grade gliomas. Validation of these results using a prospective large study is warranted.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Dynamic susceptibility contrast perfusion MRI is a technique that allows derivation of relative cerebral blood volume (rCBV) maps on the basis of the susceptibility effect related to passage of MRI contrast material through tissue. This technique has been previously used in evaluation of brain tumors with success. High-grade tumors typically have higher rCBV than low-grade tumors, a finding that is thought to be due to the higher capillary density within high-grade tumors [4]. Few reports attempting to distinguish cerebral abscesses from neoplasms using dynamic susceptibility contrast perfusion MRI have been published, and those reports either have focused on abscesses caused by a single specific organism (thereby not establishing applicability to abscesses caused by other organisms) or have not clearly specified methodology for their study to be reproduced by other investigators [5-7]. We set out to determine whether detailed analysis of abscesses using many small regions of interest (ROIs) within the enhancing walls of lesions would allow distinction of abscesses and tumors. Our hypothesis was that rCBV within the rim-enhancing portion of cerebral abscesses would be decreased relative to the rCBV seen within the rim-enhancing portion of centrally necrotic high-grade neoplasms.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
All imaging was performed on two 1.5-T scanners (Signa, GE Healthcare). MRI sequences consisted of unenhanced axial T1-weighted images; axial T2-weighted images; contrast-enhanced T1-weighted axial and coronal images; isotropic (3D) spin-echo echo-planar DWI with a b value of 1,000 mm2/sec; and dynamic susceptibility contrast perfusion MR images. The dynamic susceptibility images were obtained with spin-echo echo-planar imaging (TR/TE, 1,500/80) using a rapid bolus (5 mL/sec) of 0.2 mmol/kg of MRI contrast material through a 20-gauge IV line. This technique allowed 40 brain volumes consisting of eight contiguous 7-mm slices to be obtained over a period of 1 min. An additional set of contrast-enhanced T1-weighted images was obtained using the same slice locations and slice thickness as the dynamic susceptibility images.
All four patients with cerebral abscesses and all four patients with high-grade tumors had rim-enhancing masses with a central unenhancing region on contrast-enhanced axial T1-weighted images. All patients with abscesses had two to six lesions. The cerebral abscesses were located in both frontal lobes in two cases; in the frontal and temporal lobes in one case; and in both frontal lobes, the parietal lobe, and the cerebellum in one patient. Two patients had satellite lesions, and two did not have this finding. All abscesses had a smooth inner margin of the enhancing border. Abscesses in three patients showed mildly bright signal on DWI, and one patient had an abscess that was isointense with white matter. Three patients had abscesses with capsules that showed mildly decreased signal on T2-weighted images compared with white matter, and one had abscesses in which the capsule was isointense with white matter.
Patients with neoplasms were chosen for inclusion in the study by two CAQ-certified neuroradiologists on the basis of findings of rim enhancement and central hypointensity on contrast-enhanced axial T1-weighted images, which produced an appearance that is typical for abscesses. All patients with tumors each had a solitary lesion. The primary brain neoplasms were located in the frontal lobe (n = 2), the occipital lobe (n = 1), and the thalamus (n = 1). No patients with tumors had satellite lesions. Two tumors had a smooth inner margin of the enhancing border, and two tumors did not show this finding. No tumors showed bright signal on DWI. One tumor had mildly decreased signal on T2-weighted images compared with white matter.
The rCBV maps were generated on an Advantage Windows workstation (GE
Healthcare) by a single observer using the negative enhancement integral
program available in the Functool software program (GE Healthcare). The same
observer analyzed all rCBV maps. Briefly, the rCBV maps constructed from
dynamic susceptibility perfusion MR images were based on the change in signal
intensity (S) from baseline (S0) observed during
the passage of contrast material. Subsequently, this measurement was converted
to a change in T2* relaxation rate using the following formula:
 |
Change in T2* relaxation rate is linearly proportional to the concentration of contrast material in tissue, which allows the time-signal-intensity curve to be converted to a time-concentration curve. This calculation allowed rCBV to be generated on a voxel-by-voxel basis using a previously described technique that drew the limits of integration between the end of the baseline and the trough signal intensity (i.e., peak contrast concentration).
For each patient, multiple ROIs were placed on the sites on the rCBV map that best corresponded to the rim-enhancing region on contrast-enhanced axial T1-weighted images (Figs. 1A, 1B, 1C, 1D, 1E). The quantitative measurements of rCBV within the rim-enhancing regions of each lesion were guided by superimposing the rCBV maps on axial contrast-enhanced T1-weighted images. In patients who had multiple cerebral abscesses, measurements were taken from the largest abscess. The adequacy of superimposition of the rCBV maps on the contrast-enhanced T1-weighted images was judged by visual inspection, and in all cases, the two images were found to be similar in terms of slice location and offset in the x- and y-planes. This process allowed placement of ROIs on rCBV maps that best corresponded to the enhancing rim on the contrast-enhanced T1-weighted images. The observer could move back and forth between the T1-weighted image and the rCBV map for guidance of ROI placement.
|
|
|
|
|
Measurements within the enhancing rim of all lesions were made in the following manner at a single slice location on which the lesion showed maximal cross-sectional dimension: The field of view was set to either 6 x 6 or 3 x 3 cm (depending on the size of the rim-enhancing lesion). Next, each rCBV map was unsmoothed; this step provided an image consisting of individual pixels (Figs. 1A, 1B, 1C, 1D, 1E). Finally, square ROIs, each measuring one pixel, were placed on the rCBV map in areas that best corresponded to the superimposed contrast-enhancing rim of the axial T1-weighted images. The number of ROIs needed to completely measure the rCBV of the enhancing rim varied depending on the diameter and thickness of the rim in each patient. A total of between 25 and 30 ROIs were used to measure rCBV for each lesion (Figs. 2A, 2B, 2C). An attempt was made not to place more than one ROI at any point along the periphery of the rim. In a separate procedure, square ROIs measuring one pixel were also used to measure the rCBV values within the central unenhancing necrotic portion of each lesion. The number of ROIs used for analysis of the central portion of the lesion ranged between 23 and 30 for each lesion. The time for rCBV analysis and tabulation was approximately 30 min per case.
|
|
|
The mean rCBV measured in the enhancing rim of each abscess or tumor was calculated. Mean rCBV of a region of contralateral normal-appearing white matter (as determined from T2-weighted images) was measured using an ROI measuring between 70 and 90 mm2, and the ratio of the rCBV within the wall of the lesion divided by the rCBV in contralateral normal-appearing white matter was derived for abscesses and tumors. Two objective characteristics of the abscesses and tumors were compared. The rCBV ratios within enhancing portions of each type of lesion were compared. Then, the rCBV ratios for the central unenhancing portions of lesions were compared between groups. In all comparisons, the Wilcoxon-Mann-Whitney test of rank sums was performed to detect any statistically significant differences between the two groups.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The mean (± SD) rCBV ratio for enhancing portions of abscesses was 0.79 ± 0.18 (range, 0.61-0.95). The mean rCBV ratio for enhancing portions of tumors was 1.40 ± 0.54 (range, 1.03-2.20). The difference between the two mean values was statistically significant (p = 0.03). A plot of mean rCBV values in rim-enhancing tumors and cerebral abscesses, depicted in Figure 3, shows mean rCBV values in tumors to be much higher than those in cerebral abscesses. The central regions of the abscess group had a mean rCBV ratio of 0.24 ± 0.08. The central portions of tumors had a mean ratio of 0.20 ± 0.51. The rCBV ratio of the central portions of abscesses and tumors was not statistically different (p = 0.86).
|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Dynamic susceptibility contrast (T2*-weighted) hemodynamic imaging is a technique that is based on the fact that the passage of MRI contrast material through tissue results in a decrease in signal intensity due to susceptibility effect. This decrease is primarily related to capillary density, and tissues with high capillary density (i.e., gray matter structures) experience a greater decrease in signal intensity than tissues with lower capillary density (i.e., white matter). For this reason, high-grade neoplasms, which are associated with a high degree of neovascularity, have been shown to have high capillary density and, on that basis, high rCBV [4]. We found that abscesses had significantly lower rCBV values compared with normal white matter than high-grade neoplasms. In the experimental setting, cerebral abscesses have been shown to have relatively high amounts of mature collagen and decreased neovascularity [12]. On a theoretic basis, such histologic features should be associated with low capillary density and low rCBV, which may serve as the explanation for the lower rCBV ratios seen in abscesses.
We set out to determine whether hemodynamic MRI would prove to be a valuable method for the distinction of cerebral abscess from neoplasm for a number of reasons. It would be advantageous to make the definitive diagnosis on a single study (i.e., MRI) rather than add an additional imaging study (e.g., PET or SPECT). The time penalty incurred by the dynamic susceptibility contrast pulse sequence is small. The imaging sequence itself is 60 sec long. Postprocessing time using the software available to us is generally approximately 30 sec. Therefore, the time for performance of this pulse sequence is much shorter than that needed for an alternative MRI technique, such as MR spectroscopy. On the other hand, in our experience, optimal dynamic susceptibility contrast imaging requires the use of a power injector through an IV line that is 20 gauge or larger and a double dose (0.2 mmol/kg) of contrast material. Therefore, a small amount of additional time for patient preparation (rather than scanner time) and additional cost (for additional contrast material) are incurred.
Reports of perfusion MRI in abscesses are few. In one previous study of four cerebral abscesses, investigators also noted that abscess capsules showed lower rCBV values than the rims of enhancing tumors [7]. However, those investigators did not outline whether a single region of the abscess capsule was sampled or whether a large portion of the capsule was interrogated, which could influence results. Our analysis was designed to interrogate the entire enhancing rim of an abscess capsule to minimize the bias introduced by an arbitrary sampling of only a portion of the capsule.
Two recent studies of perfusion MRI of cerebral abscesses have focused on toxoplasmosis abscesses in immunocompromised patients compared with lymphoma masses, rather than in comparison to the more common glial tumors examined in our series [5, 6]. In both studies, toxoplasmosis lesions were found to have lower rCBV values than lymphomas. Our study differed from those studies in a number of ways. Our study included only rim-enhancing lesions in both study groups (rather than both solid and cystic tumors in the tumor study group) and did not focus solely on toxoplasmosis abscesses. Some of the previous studies used a single ROI drawn around the entire lesion (whether abscess or tumor) rather than a series of small ROIs within the abscess capsule as in our study. Drawing a single ROI has the advantage of ease of placement. However, this technique has the disadvantage of averaging rCBV values of central portions of the abscess with those of the abscess capsule. As a result, rCBV values in individual abscesses can vary simply by varying the degrees of the size of central purulent collection to the thickness of the abscess capsule. In particular, a lesion containing a small central purulent collection and a thick capsule could potentially have a substantially higher rCBV than a lesion with a large central cavity.
Our study has a number of limitations. Our study population was small. Further studies are needed to corroborate our results and to establish the sensitivity, specificity, and accuracy of this technique. We encountered a technical limitation in the slight offset between contrast-enhanced T1-weighted data sets and rCBV maps. This was often due to slight patient motion between acquisition of the dynamic susceptibility contrast sequence and the contrast-enhanced T1-weighted sequence. This slight offset in some cases necessitated drawing the ROIs directly on the rCBV maps rather than on contrast-enhanced T1-weighted images. However, placing the ROIs directly on the rCBV maps was not difficult because the rim of the lesion was usually easily discernible on these images. Therefore, placement of ROIs in this manner likely did not substantially decrease accuracy of rCBV measurements within rim-enhancing regions.
Although dynamic susceptibility MRI offers some advantages over other imaging techniques, the process of obtaining rCBV measurements of the lesions proved too tedious and time-consuming for two reasons. The morphology of the lesions examined in this study required placement of a large number of small ROIs (between 23 and 30 per case). Because the rim of the lesions was relatively thin, it was necessary to place ROIs that were no larger than a pixel. This process is impractical in a busy clinical environment without the aid of dedicated personnel to perform rCBV measurements in such a manner. We found it took a trained user approximately 30 min per case to place ROIs and tabulate the rCBV measurements.
In summary, although our results are preliminary, it appears that dynamic susceptibility MRI may provide an objective measurement for distinguishing high-grade cerebral neoplasms from abscesses. Validation of these results using a prospective large study is warranted. Dynamic susceptibility contrast perfusion MRI is fast, is easy to perform, and provides additional information that may help in diagnosis. Because our analysis technique was tedious and time-consuming, development of an automated program that could analyze the data quickly and accurately would be of benefit.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  R. N. Al-Okaili, J. Krejza, J. H. Woo, R. L. Wolf, D. M. O'Rourke, K. D. Judy, H. Poptani, and E. R. Melhem Intraaxial Brain Masses: MR Imaging-based Diagnostic Strategy--Initial Experience Radiology, May 1, 2007; 243(2): 539 - 550. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Wintermark, M. Sesay, E. Barbier, K. Borbely, W. P. Dillon, J. D. Eastwood, T. C. Glenn, C. B. Grandin, S. Pedraza, J.-F. Soustiel, et al. Comparative Overview of Brain Perfusion Imaging Techniques Stroke, September 1, 2005; 36(9): e83 - e99. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
