| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Neuroradiology |
1 From the Mallinckrodt Institute of Radiology, Washington University School of Medicine, 510 S Kingshighway Blvd, St Louis, MO 63110. Received January 19, 1999; revision requested March 9; final revision received July 15; accepted July 20. This work was supported by the Major Grants Program of the McDonnell Center for Higher Brain Function and the Charles A. Dana Foundation Consortium on Neuroimaging Leadership Training and National Institutes of Health grant NS06833. Address reprint requests to P.M. (e-mail: MukherjeeP@mir.wustl.edu).
 |
Abstract |
|---|
 TOP Abstract IntroductionMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIXReferences |
|---|
MATERIALS AND METHODS: Twelve patients with unilateral middle cerebral arterial infarcts were examined with diffusion tensor–encoded echo-planar MR imaging 17 hours to 5 days after stroke onset. Isotropic diffusion coefficient (
) and diffusion anisotropy (A
) images were computed. values were measured in ischemic and contralateral gray matter and white matter by using A images to differentiate white matter from gray matter. images were compared with unidirectional and directionally averaged diffusion-weighted images.
RESULTS: In all patients, images showed two distinct levels of diffusion reduction in the infarct; more severe reduction occurred exclusively in white matter. values were significantly less in infarcted white matter than in infarcted gray matter, whereas values in the contralateral white matter and gray matter were not significantly different. Relative to the contralateral side, values in the infarct were reduced by 46% in white matter and by 31% in gray matter (P < .001). Diffusion-weighted imaging caused underestimation of the magnitude and, in some cases, the spatial extent of the white matter diffusion abnormality.
CONCLUSION: Isotropic diffusion is more reduced in white matter than in gray matter in acute to early subacute middle cerebral arterial stroke. Diffusion-tensor imaging may be more sensitive than diffusion-weighted imaging to white matter ischemia.
Index terms: Anisotropy • Brain, diffusion, 10.12144, 10.92 • Brain, gray matter, 10.92 • Brain, infarction, 10.78 • Brain, MR, 10.121411, 10.121413, 10.121416, 10.12144 • Brain, white matter, 10.92 • Diffusion tensor • Magnetic resonance (MR), diffusion study, 10.12144, 10.92 • Magnetic resonance (MR), rapid imaging, 10.121413
|
Introduction |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIXReferences |
|---|
Most of the quantitative diffusion data available in the clinical stroke literature are based on determinations of the effective or apparent diffusion coefficient (ADC) along the direction of a single diffusion-sensitizing gradient. Unidirectional ADC images of this type contain uncontrolled intensity variations due to anisotropic diffusion along white matter fibers. The ADC can also be measured along three orthogonal gradients and averaged to compensate for the effects of anisotropy. Studies based on this averaged orthogonal approach have not revealed appreciable differences between gray matter and white matter diffusion in acute stroke (7–9).
However, the detection of these diffusion differences may be limited by factors such as a low signal-to-noise ratio, eddy current–induced image misregistration, anatomic distortions in the infarct, and small infarct size. It is important to further examine the issue of diffusion in ischemic gray versus white matter because the results may affect interpretation of the radiologic image, clinical treatment of patients with stroke, clinical studies of neuroprotective agents, and basic scientific studies of infarct heterogeneity and the pathophysiologic mechanisms of stroke.
We conducted this investigation to determine if differences in water diffusion exist between white matter and gray matter in acute to early subacute human stroke with use of optimized tetrahedral-orthogonal diffusion-tensor imaging with image realignment (10,11). This method is used to sample the diffusion tensor along four tetrahedral and three orthogonal directions and may offer advantages for the detection of diffusion differences in gray matter and white matter. Tetrahedral or combined tetrahedral-orthogonal encoding provides a higher signal-to-noise ratio than does orthogonal encoding alone (12).
In addition, we used the anisotropy information available from diffusion-tensor imaging to distinguish white matter from gray matter, even in cases of distorted anatomy, facilitating comparisons between the two tissue types. We also compared diffusion-tensor imaging with diffusion-weighted imaging to assess the diffusion abnormality within gray matter and white matter in stroke. Finally, we gauged the extent to which T2-weighted signal effects contribute to the observed disparities between diffusion-weighted and diffusion-tensor imaging.
|
MATERIALS AND METHODS |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIXReferences |
|---|
|
Six patients were imaged within 72 hours of symptom onset (range, 17–48 hours), and the other six patients were examined 3–5 days after symptom onset. The time of symptom onset was determined from a review of the medical records. Permission to use these clinical neuroimaging data for scientific research was granted by the institutional review board at our medical center.
All examinations were performed on a 1.5-T system (Magnetom Vision; Siemens, Erlangen, Germany) by using circularly polarized clinical head coils and single-shot spin-echo echo-planar pulse sequences with Stejskal-Tanner diffusion gradients (14). Both diffusion-weighted and diffusion-tensor protocols used a 24 x 24-cm field of view, 5-mm section thickness, and 1-mm gap between sections. The diffusion-weighted protocol used a commercial pulse sequence (
/123 [repetition time msec/echo time msec]; b = 1,006 sec/mm2) with a diffusion gradient in only the section-select direction and a 128 x 128-voxel matrix (1.88 x 1.88 x 5.00-mm voxels) that was interpolated to a 256 x 256-pixel matrix. Twenty images were obtained in the entire brain in both transverse and coronal orientations; imaging time for each orientation was 4 seconds.
Unless otherwise stated, the tetrahedral-orthogonal diffusion-tensor protocol and subsequent image processing was performed as described in the Data Acquisition (sequence B) and Data Analysis sections in the article by Shimony et al (11). Briefly, diffusion-tensor data were generated with a single-shot multisection spin-echo echo-planar pulse sequence (3,000/97.4) that used four tetrahedrally oriented diffusion gradients (b = 1012.4 sec/mm2 with a 22.1 mT/m input gradient strength), three orthogonally oriented diffusion gradients (b = 337.5 sec/mm2 with the same input strength), and a reference intensity image (b = 0 sec/mm2). Fourteen transverse sections were imaged in 35 seconds with a 96 x 128-voxel matrix (2.50 x 1.88 x 5.00-mm voxels with 1-mm gaps) that was interpolated to a 192 x 256-pixel matrix. All images were realigned in two dimensions by using a combination of intra- and cross-modality affine realignment procedures to correct for image displacements and linear stretch and/or shear due to eddy currents (11). Images were postprocessed offline in approximately 45 minutes by using an UltraSPARC 2 Unix workstation (Sun Microsystems, Palo Alto, Calif). A similar tetrahedral-orthogonal diffusion-tensor procedure has been used to investigate water diffusion in the human neonatal brain (15).
Fat-suppressed turbo fluid-attenuated inversion recovery (FLAIR) images (9,999/119/2,309 [repetition time msec/echo time msec/inversion time msec]; echo train length of seven) were obtained in 16 transverse sections of 6-mm thickness with a 1-mm gap between sections (16). Images were acquired in 4 minutes 29 seconds with a 182 x 256-voxel matrix and a 20 x 23-cm field of view (1.10 x 0.90 x 6.00-mm voxels). The inversion time was optimized to suppress the cerebrospinal fluid signal, and fat was suppressed with radio-frequency pulses. The transverse FLAIR and diffusion-weighted images were oriented along the anterior commissure–to–posterior commissure line. They were, therefore, not necessarily acquired in register with the diffusion-tensor images, which were oriented in the plane transverse to the magnet bore.
Image Computation
The derivation of the elements of the diffusion tensor D from the tetrahedral and perpendicular diffusion measurements is given elsewhere (11). Further details of image computation are also provided in the Appendix. Two useful parameters that can be derived from diffusion-tensor imaging are the directionally averaged (isotropic) diffusion coefficient () (7) and the diffusion anisotropy measure (A) (10,17). represents the component of diffusion that is the same in all directions; that is, is a general measure of the magnitude of water diffusion averaged over all directions. It is related to the trace of the diffusion tensor and is equivalent to the orthogonally averaged ADC (trace ADC) used previously (7–9). A represents the component of diffusion that varies with spatial direction; that is, A measures the degree to which water diffusion varies with direction. The relatively anisotropic white matter is highlighted on images of A.
Two other types of images were derived from the diffusion-tensor data, namely, isotropic diffusion-weighted images and diffusion-intensity images. Isotropic diffusion-weighted images were computed as the geometric mean of the four tetrahedral diffusion-weighted images. This directional averaging operation removes the effects of anisotropy while retaining the spin-density and T2 weighting present in unidirectional diffusion-weighted imaging. This procedure is similar to the technique described by Chong et al (18) that is based on the geometric mean of three orthogonal diffusion-weighted images.
The final type of image was the diffusion-intensity image, which represents the component of diffusion-weighted signal intensity that is purely due to isotropic diffusion (19). The diffusion-intensity image is effectively an isotropic diffusion-weighted image on which the spin-density and T2 effects have been removed. As on the images, the signal intensity on the diffusion-intensity image is solely dependent on isotropic diffusion. The contrast polarity of these diffusion-intensity images is the same as that of the unidirectional and isotropic diffusion-weighted images and opposite to that of the images. Areas of decreased diffusion appear bright on these diffusion-intensity images, and areas of increased diffusion appear dark.
Region-of-Interest Analysis
For the measurement of values, regions of interest (ROIs) were manually traced on a representative transverse image in each patient's data set by using ANALYZE (Mayo Foundation, Rochester, Minn). The representative transverse section was chosen as the one on which the imaging plane passed near the center of the infarct as determined from the isotropic diffusion-weighted image (Fig 1, a). A tracing that encompassed the total region of infarction (total infarct ROI) was drawn on the image (Fig 1, b). This tracing was confirmed to contain all regions of abnormal signal intensity on the corresponding isotropic diffusion-weighted image. A similarly sized region encompassing gray matter and white matter was manually drawn in the contralateral hemisphere (total contralateral ROI). Regions of increased diffusion in periventricular locations (Fig 1, b), which were indicative of age-related leukoaraiosis that also appeared as areas of high signal intensity on T2-weighted images, were not excluded from the total contralateral ROI. This was done to assess the baseline values in noninfarcted brain that were specific to each patient.
|
images were used to define white matter regions to generate unbiased measurements of gray matter and white matter values. A tracing that circumscribed the regions of anisotropic diffusion was drawn on the A image in both hemispheres (Fig 1, c). This A tracing was used to differentiate white matter from gray matter in both the infarct and contralateral ROIs. As illustrated in Figure 1, d, the intersection of the areas enclosed by the and A tracings was designated as white matter. Areas within the tracing but outside the A tracing were designated as gray matter. All tracings were drawn by one author (P.M.) and were confirmed to be anatomically accurate by a board-certified radiologist with a certificate of added qualification in neuroradiology (M.M.B.) and a second board-certified radiologist (T.E.C.). This procedure typically generated one to four closed regions of gray matter in each hemisphere in each patient. Multiple gray matter regions occurred more commonly on the contralateral side (Fig 1, d). In patients with multiple regions, the gray matter region on each side with the largest area and the least-visible partial-volume averaging with adjacent cerebrospinal fluid was chosen as the infarct or contralateral gray matter ROI for further analysis. In white matter, this procedure typically yielded only one closed region on each side. In the few cases with more than one closed region of white matter, the region with the largest area was selected as the infarct or contralateral white matter ROI for further analysis. No partial-volume effects with cerebrospinal fluid were noted in the white matter regions.
The mean and SD of the pixel values were computed in each ROI in each of the 12 patients. The mean values in the ROIs were compared across patients with a paired two-population two-tailed Student t test by using Origin 4.1 (Microcal Software, Northampton, Mass). Figures were prepared by using Photoshop 4.0 for Windows (Adobe Systems, San Jose, Calif) to create montages; to adjust size, brightness, and contrast; and to remove background noise outside of the brain. Brain images were not edited except for the placement of markers to indicate findings of special interest.
|
RESULTS |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIXReferences |
|---|
and diffusion-intensity images (Fig 2). However, the signal intensity in gray matter was greater than that of white matter in the infarct on diffusion- and T2-weighted images. (The preponderance of gray matter in the areas of abnormal signal intensity on T2-weighted images can help explain this discrepancy between the findings on diffusion-weighted and diffusion-intensity images in acute stroke [see Discussion]).
|
and diffusion-intensity images but appeared relatively spared on diffusion- and T2-weighted images (Fig 3). On diffusion-weighted images, the areas of apparent white matter sparing were located adjacent to areas of gray matter with abnormally high signal intensity.
|
and diffusion-intensity images showed lesion heterogeneity that was characterized by two distinct levels of diffusion reduction termed moderate and marked (Figs 1b, 2d, 2e, 3d, 3e, 4d, 4e). In all patients, markedly reduced diffusion occurred exclusively in white matter, while moderately reduced diffusion occurred exclusively in gray matter. This assignment was based on two observations. First, in all 12 patients, the regions of infarct that demonstrated anisotropic diffusion also had markedly reduced values (Figs 1c, 1d, 2e, 2f, 3e, 3f, 4e, 4f). This means that all regions of infarcted white matter with detectable anisotropy had markedly reduced diffusion. Second, and in converse, the entire region of markedly reduced values also demonstrated anisotropic diffusion in 10 of the 12 patients (Figs 1c, 1d, 2e, 2f, 3e, 3f).
|
values did not have visible anisotropy. This is compatible with a loss of white matter anisotropy in the later stages of stroke evolution (20–22). However, in these two patients, regions with markedly reduced values had a morphology that was characteristic of white matter (Fig 4e, 4f). Thus, all regions with markedly reduced diffusion were white matter.
|
values in all 12 patients. The mean value in the total infarct ROI was (0.54 ± 0.09) x 10-3 mm2/sec and included both white matter and gray matter. This value was significantly less than the mean value in the total contralateral ROI, which was (0.89 ± 0.07) x 10-3 mm2/sec (P < .001). The percent reduction of values in the total infarct ROI relative to that in the total contralateral ROI was 39% ± 10. Both gray matter and white matter exhibited decreased values in the infarct compared with the values in the contralateral hemisphere. The percent reduction of values in infarcted gray matter averaged across all 12 patients was 31% ± 7, relative to that of contralateral hemisphere. In infarcted white matter, the percent reduction of values was 46% ± 12. This difference between white matter and gray matter was significant (P < .001).
In the infarcted regions, values in white matter were less than those in gray matter in every patient. This is illustrated in Figure 5, which shows that the values in all infarct ROIs are above the line of identity, that is, the line that indicates equal gray matter and white matter diffusion. The mean value in affected white matter across all patients was (0.46 ± 0.10) x 10-3 mm2/sec and was significantly less (P < .001) than the mean value in affected gray matter, which was (0.61 ± 0.06) x 10-3 mm2/sec.
|
values in gray matter and white matter tended to be greater at lower overall values of . (Note the divergence of the infarct regression line from the line of identity in Figure 5.) The slope of the regression line that best fit the infarct data was 0.51 (Fig 5), which was significantly different from the unit slope (P < .01). In contrast, contralateral hemisphere data in the 12 patients was not significantly different; the mean value of was (0.86 ± 0.08) x 10-3 mm2/sec in white matter and (0.88 ± 0.08) x 10-3 mm2/sec in gray matter (P > .05).
Subgroup analysis of the six patients who underwent imaging within 72 hours of stroke onset (<3-days group, Table 1) revealed mean values of (0.42 ± 0.08) x 10-3 mm2/sec in affected white matter and (0.59 ± 0.07) x 10-3 mm2/sec in affected gray matter (P < .001). In the six patients who underwent imaging 3–5 days after infarction, the difference in white matter and gray matter diffusion remained significant (P < .01); values were (0.51 ± 0.11) x 10-3 mm2/sec and (0.62 ± 0.06) x 10-3 mm2/sec, respectively. The difference between white matter and gray matter in the percentage reduction in values was significant in both the <3-days group (P < .001) and the 3–5-days group (P < .05), although the percentage decrease in was less in the latter case.
The reduction in the values in white matter tended to be greater in the <3-days group (53% ± 9 decrease) than in the 3–5-days group (39% ± 12 decrease), but this difference was not statistically significant (P > .05). The percent reduction in gray matter values was similar in the <3-days group (33% ± 7 decrease) and 3–5-days group (28% ± 7 decrease, P > .05). The mean percentage reduction in values in the total infarct ROI relative to that of the total contralateral ROI was greater in the <3-days group (45% ± 8) than in the 3–5-days group (34% ± 10), but this difference was not statistically significant (P > .05) with these sample sizes.
|
DISCUSSION |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIXReferences |
|---|
In one diffusion-tensor study of hyperacute and early acute stroke, Sorensen et al (23) reported no statistically significant difference in isotropic diffusion in ischemic gray matter and white matter. Their result raises the possibility that the difference between gray matter and white matter diffusion reported herein might not be present in the initial hyperacute phase of stroke. However, their study was not specifically designed to investigate diffusion differences in gray matter and white matter in individual patients. Many of their patients had infarcts that were isolated to only gray or white matter.
In the present study, we examined patients with infarcts that encompassed confluent regions of both gray matter and white matter. In this design, each patient served as his or her own control subject in terms of the relative magnitude of changes in gray matter and white matter diffusion in the infarcted territory compared with the changes in the uninvolved contralateral hemisphere. Therefore, further research is needed to resolve whether the differences in gray matter and white matter diffusion established herein in acute to early subacute stroke are also detectable in the clinically important hyperacute stage. Follow-up serial imaging of these patients during the acute and subacute stages would also be helpful in the definition of the time course of stroke-related diffusion alterations in white matter versus gray matter.
The 39% decrease in values in the first 5 days after infarction that was found in this investigation agrees with the 41.7% decrease in ADC values in the first 4 days after stroke onset that was determined in a large study of 157 patients (13). The accuracy of the values in the present study is further supported by measurements obtained in regions contralateral to the infarct; these measurements are well within the range reported both in healthy volunteers (12,24) and in uninvolved brain regions in stroke patients (9,25). We also found no statistically significant difference in values in contralateral gray matter and white matter. These findings agree with those of quantitative diffusion MR imaging studies in healthy adults (12,24,26).
The larger values of in infarcted gray matter relative to those in infarcted white matter cannot be explained by the effect of partial-volume averaging with cerebrospinal fluid. These differences persisted even after normalization with the data obtained from the contralateral side, where partial-volume averaging of cortex with cerebrospinal fluid is more likely. Partial-volume averaging with cerebrospinal fluid was less consequential in the infarct ROIs than in the contralateral ROIs because of the sulcal effacement that is characteristic of acute and early subacute stroke. Furthermore, no statistically significant difference in values was found in white matter and gray matter ROIs in the contralateral hemisphere (Table 2). This finding suggests that cerebrospinal fluid averaging in gray matter was effectively minimized in the contralateral gray matter ROIs, as well.
Information about anisotropy, obtained by using the A measure, was the sole criterion that was used in the differentiation of white matter and gray matter in this study, since white matter has much greater anisotropy than does gray matter (11,26,27). However, white matter anisotropy decreases in areas in which infarction is evolving (20–22). A small decline in anisotropy has also been reported in acute stroke (23). In our study, enough residual anisotropy remained to allow adequate identification of infarcted white matter in 10 of 12 patients (Figs 1–3). In six of the 12 patients, the infarcted region was sufficiently distorted from mass effect so that areas of markedly decreased diffusion could not be confidently identified as white matter on images alone without correlation with A images (Figs 1, 3). Correlation of the contrast patterns on and A images may be important for the differentiation of diffusion contrast due to tissue type (ie, gray vs white matter) from diffusion contrast due to heterogeneity of the ischemic process.
Two of the 12 patients had severe reductions in anisotropy in the infarct, and both reductions involved early subacute stroke. Despite these findings, only regions of residual anisotropy were used to objectively measure values in white matter. Nevertheless, the images demonstrated regions of markedly reduced diffusion that extended beyond the regions of residual anisotropy; the regions had the morphology of white matter (temporal lobe in Fig 4). In these two patients, the use of A values to segment white matter likely resulted in the inclusion of areas of white matter within the zone that was deemed to be gray matter.
However, this averaging of white matter with gray matter reduces apparent differences between the two tissue types rather than causes artifactual differences where none exists. An exception is if there is preferential loss of anisotropy in regions where values are not substantially reduced (eg, selective loss of anisotropy in the periphery of the lesion with a more severe reduction in values in the center of the lesion where anisotropy is preserved). Such a scenario is unlikely given the pattern of anisotropy loss in Figure 4, f and the relative uniformity of the reduction in values in white matter in Figures 1–4. However, additional experience with cases of marked loss of anisotropy in subacute stroke is needed for a full evaluation of this possibility.
Diffusion-weighted imaging within the first 48 hours of stroke typically shows greater increases in signal intensity in gray matter than in white matter (Figs 2b, 3b); this finding is similar to the contrast pattern on T2-weighted images (Figs 2a, 3a). Traditionally, gray matter has been considered to be more vulnerable than white matter to the early effects of ischemia (28); this consideration would appear to be corroborated by the findings of diffusion-weighted imaging.
After the initial 48 hours, the area of abnormal signal intensity on diffusion-weighted images typically becomes more homogeneous. It extends throughout the region of infarction with little or no gray-white contrast (Fig 1, a and posterior temporal lobe in Fig 4, b). This contrast pattern can also be observed on T2-weighted images (posterior temporal lobe in Fig 4, a). The progression of the area of abnormal signal intensity on diffusion-weighted images over time—from a predominant distribution in gray matter in acute infarcts to a more uniform involvement of both gray matter and white matter in subacute infarcts—was noted in the earliest diffusion-weighted studies of stroke (eg, fig 2 in reference 2).
However, these diffusion-weighted imaging findings should not necessarily be interpreted as an indication that the underlying diffusion abnormality is greater in gray matter than in white matter. In fact, the data herein demonstrate that the diffusion abnormality is greater in infarcted white matter. The predominant involvement of gray matter in acute stroke on diffusion-weighted images with more uniform involvement of gray matter and white matter in early subacute stroke results from multiplicative signal effects from elevated T2-weighted signal with diffusion-weighted imaging (see Appendix, Eq [A5]).
The areas of high signal intensity on T2-weighted FLAIR images are predominantly distributed in the gray matter in acute infarcts (Fig 2, a) and progress to a more homogenous involvement of gray matter and white matter in the early subacute phase (Fig 4, a). This evolution of signal intensity on T2-weighted images is similar to that on diffusion-weighted images and has been also attributed to the greater vulnerability of gray matter to early ischemia (29,30).
FLAIR images are useful in the determination of the relative contribution of T2-weighted effects to diffusion-weighted imaging because FLAIR and diffusion-weighted imaging are both T2-weighted and cerebrospinal fluid–suppressed, yet FLAIR imaging is insensitive to diffusion (31). Hence, areas of increased signal intensity on FLAIR images contribute to contrast with diffusion-weighted imaging. These signal effects in gray matter with T2-weighted imaging (Fig 2, a) were especially pronounced on unidirectional diffusion-weighted images (Fig 2, b) compared with isotropic diffusion-weighted imaging (Fig 2, c). This is because the longer echo time of the unidirectional diffusion-weighted sequence (123.0 vs 97.4 msec) more closely matches the characteristic range of T2 values in the infarct.
Diffusion-weighted imaging has been shown to have better sensitivity, specificity, and accuracy in the detection of acute stroke than does trace ADC imaging (18). Those authors attribute this to T2-weighted effects with diffusion-weighted imaging, which trace ADC images lack. Their results agree with our findings, as the infarcts in Figures 1–4 were more conspicuous on diffusion-weighted images than on diffusion-intensity images, which also lack T2 weighting. We refer to this phenomenon as cooperative contrast between T2 and diffusion (32), which amplifies the contrast between ischemic and normal gray matter in diffusion-weighted imaging. Thus, diffusion-intensity imaging may be expected to be less sensitive than diffusion-weighted imaging for the detection of small focal cortical infarcts when changes in T2-weighted signal are present.
However, our comparison of diffusion-weighted and diffusion-intensity imaging indicate that diffusion-weighted imaging may cause underestimation of the magnitude and spatial extent of white matter involvement in stroke. In all 12 patients, diffusion-intensity images demonstrated more severe diffusion changes in white matter compared with those in gray matter, whereas areas of abnormal signal intensity on diffusion-weighted images were equal in the two tissues or were greater in gray matter. In four of the 12 patients, diffusion-intensity images showed regions of white matter involvement that appeared relatively spared at diffusion-weighted imaging (Fig 3). The lower sensitivity of diffusion-weighted imaging to white matter ischemia may be secondary to less contrast amplification from T2 effects, which are delayed in white matter compared with gray matter. Consequently, subtle changes in white matter on diffusion-weighted images may not be visible adjacent to the large changes in signal intensity in the surrounding gray matter.
Differences in gray matter and white matter diffusion in stroke could be due to variability between these two tissue types at any stage in the process that leads from ischemia to altered diffusion. Specifically, the observed diffusion contrast in gray matter and white matter could be caused by differences in (a) the mismatch between blood supply and metabolic demand (ischemia), (b) the type and/or severity of the histopathologic response to ischemic injury (vulnerability), or (c) the mechanisms by which histopathologic changes lead to altered diffusion. Findings from positron emission tomographic studies suggest that the flow-metabolism mismatch is similar in gray matter and white matter in middle cerebral arterial infarcts. (For example, see the uniformity of oxygen extraction fraction images in figure 1 in the article by Carpenter et al [33]).
Regarding the histopathologic response, gray matter has been traditionally considered to be more vulnerable than white matter to early ischemia (28). However, more recent findings in experimental models of stroke have demonstrated that ischemic damage to white matter occurs earlier and with greater severity than previously appreciated. In two studies, one conducted in rats within 24 hours of middle cerebral arterial occlusion (34) and the other conducted in cats within 3 hours of middle cerebral arterial occlusion (35), investigators documented histopathologic changes in affected white matter as early as 30 minutes after the onset of ischemia. Astrocytic swelling secondary to cytotoxic edema occurred in both white matter and gray matter, and white matter also displayed hydropic swelling of oligodendroglial cell bodies. Immunocytochemical experiments in which rat models of middle cerebral arterial stroke were used revealed pathophysiologic axonal changes in myelinated white matter within 4–6 hours after arterial occlusion (36,37) that are similar to those that resulted from traumatic diffuse axonal injury (37). Findings from these investigations highlight the vulnerability of white matter to even short periods of ischemia.
Regarding the histopathologic-biophysical diffusion mechanisms, intracellular accumulation of water and enlargement of the periaxonal space were found in myelinated fiber tracts in animal models (34,35), with differences in the dependence of ischemic gray matter and white matter diffusion on total water accumulation (35). Findings from these studies suggest that intra- and extracellular accumulation of fluid and resultant changes in diffusion might differ significantly in white matter and gray matter. Finally, additional histopathologic-biophysical mechanisms of diffusion reduction that do not exist in gray matter might occur in white matter; these mechanisms include reduced bulk water motion from cytoskeletal breakdown and disruption of fast axonal transport (36,37).
If the results from previous animal experiments can be generalized to humans, they may provide a pathophysiologic basis for the gray-white diffusion contrast found in this investigation. The rapid onset of the changes in white matter in these animal studies is also consistent with our findings, as gray-white diffusion contrast was observed at all of the time intervals after stroke onset that were studied, including the earliest (17 hours). These results suggest that this gray-white diffusion contrast represents a fundamental difference in the diffusion response of gray matter and white matter to ischemia and that these two tissues should be separately evaluated in studies of the quantitative assessment of neuroprotective therapy (38–40) or infarct heterogeneity (25,41).
In summary, diffusion-tensor imaging of acute and early subacute middle cerebral arterial infarcts demonstrates more severe reductions in isotropic diffusion in white matter than in gray matter, whereas diffusion-weighted imaging appears to cause underestimation of the magnitude and spatial extent of white matter involvement. The observed disparity in diffusion in gray matter and white matter at diffusion-tensor imaging increased with increasing overall magnitude of diffusion reduction. The discrepancy between diffusion-weighted and diffusion-tensor images was due to areas of increased signal intensity in the infarct on T2-weighted images; this increase appeared earlier and more often in gray matter than in white matter.
Diffusion-tensor imaging provides a full complement of isotropic diffusion, anisotropy, and diffusion-weighted images for the assessment of acute stroke in both white matter and gray matter. The images and measurements of diffusion in gray matter and white matter presented herein may be relevant to the interpretation of diffusion-weighted images, clinical treatment and rehabilitation of stroke patients, evaluation of neuroprotective therapies in acute stroke, and basic scientific studies of the pathophysiologic mechanisms of stroke and infarct heterogeneity. Further investigation of diffusion-tensor imaging is needed in the hyperacute phase of stroke, when results may affect the early clinical treatment of patients.
|
APPENDIX |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIXReferences |
|---|
value (an invariant measure of isotropic diffusion with units of square millimeter per second) was calculated as the arithmetic mean of the three diagonal elements of D as follows (7):
value is one-third the trace of D, where the trace is the sum of the eigenvalues of D. has also been referred to as Dav (9) or trace ADC in studies in which the diffusion coefficient was averaged over different encoding directions to remove the effects of anisotropy. is invariant to the coordinate system used to measure D and, thus, is not affected by the orientation of the head or the direction of white matter fibers.
The dimensionless absolute measure of anisotropy A was computed on a pixel-by-pixel basis from the elements of D as follows (10):
value is equivalent to the coefficient of variation of the eigenvalues of D divided by the square root of 2 (10). It is directly proportional to the relative anisotropy defined by Basser and Pierpaoli (17), with a scaling factor of the square root of 
, which places the value of A on an absolute scale from 0 (no anisotropy) to 1 (complete anisotropy). Like , A is invariant to the measurement coordinate system used to define the elements of D.
With echo times that are much shorter than T1 relaxation times, the intensity IDWI of the isotropic diffusion-weighted image is related to diffusion, T1 weighting, spin density, and T2 weighting as follows:
is the scalar spin density in the voxel; b is the diffusion-weighting factor; TE is the echo time; TR is the repetition time; and T1 and T2 are the longitudinal and transverse relaxation times, respectively. In single-shot echo-planar imaging techniques in which only one encoding direction is acquired per pulse sequence acquisition, the effective repetition time is infinite; hence, there is no T1 weighting.
The diffusion-intensity image IDI was computed by the method introduced by Sorensen et al (19) as follows:
The intensity of the isotropic diffusion-weighted image IDWI can thus be expressed in terms of the intensities of the diffusion-intensity image IDI and the reference T2-weighted–intensity image IT2WI as follows:
|
Footnotes |
|---|
= diffusion anisotropy
ADC = apparent diffusion coefficient
= isotropic diffusion coefficient
FLAIR = fluid-attenuated inversion recovery
ROI = region of interest Author contributions: Guarantors of integrity of entire study, P.M., M.M.B., T.E.C.; study concepts, P.M., M.M.B., R.C.M., J.S.S., T.E.C.; study design, P.M., M.M.B., T.E.C.; definition of intellectual content, P.M., M.M.B., R.C.M., J.S.S., A.Z.S., T.E.C.; literature research, P.M., R.C.M.; clinical studies, P.M., M.M.B.; data acquisition, P.M., E.A.; data analysis, R.C.M., J.S.S., T.S.C., A.Z.S., E.A., T.E.C.; statistical analysis, P.M., M.M.B., A.Z.S., T.E.C.; manuscript preparation, P.M.; manuscript editing and review, P.M., M.M.B., R.C.M., J.S.S., A.Z.S., T.E.C.
|
References |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIXReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  T. Taoka, M. Sakamoto, S. Iwasaki, H. Nakagawa, A. Fukusumi, S. Hirohashi, K. Taoka, K. Kichikawa, T. Hoshida, and T. Sakaki Diffusion Tensor Imaging in Cases with Visual Field Defect after Anterior Temporal Lobectomy AJNR Am. J. Neuroradiol., April 1, 2005; 26(4): 797 - 803. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
H. Nakamura, K. Yamada, O. Kizu, H. Ito, S. Yuen, T. Ito, K. Yoshikawa, K. Shiga, M. Nakagawa, and T. Nishimura Effect of Thin-Section Diffusion-Weighted MR Imaging on Stroke Diagnosis AJNR Am. J. Neuroradiol., March 1, 2005; 26(3): 560 - 565. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S M. Maniega, M E Bastin, P A Armitage, A J Farrall, T K Carpenter, P J Hand, V Cvoro, C S Rivers, and J M Wardlaw Temporal evolution of water diffusion parameters is different in grey and white matter in human ischaemic stroke J. Neurol. Neurosurg. Psychiatry, December 1, 2004; 75(12): 1714 - 1718. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 G. J. del Zoppo TIAs and the pathology of cerebral ischemia Neurology, April 27, 2004; 62(8_suppl_6): S15 - S19. [Full Text]
|
|
|
|
|
|
A. Mazumdar, P. Mukherjee, J. H. Miller, H. Malde, and R. C. McKinstry Diffusion-Weighted Imaging of Acute Corticospinal Tract Injury Preceding Wallerian Degeneration in the Maturing Human Brain AJNR Am. J. Neuroradiol., June 1, 2003; 24(6): 1057 - 1066. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. P. Goldberg and B. R. Ransom New Light on White Matter Stroke, February 1, 2003; 34(2): 330 - 332. [Full Text] [PDF]
|
|
|
|
|
|
R. C. McKinstry, J. H. Miller, A. Z. Snyder, A. Mathur, G. L. Schefft, C. R. Almli, J. S. Shimony, S. I. Shiran, and J. J. Neil A prospective, longitudinal diffusion tensor imaging study of brain injury in newborns Neurology, September 24, 2002; 59(6): 824 - 833. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J H M Chan, W C G Peh, E Y K Tsui, L F Chau, K K Cheung, K B Chan, M K Yuen, E T H Wong, and K P C Wong Acute vertebral body compression fractures: discrimination between benign and malignant causes using apparent diffusion coefficients Br. J. Radiol., March 1, 2002; 75(891): 207 - 214. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
P. M. Desmond, A. C. Lovell, A. A. Rawlinson, M. W. Parsons, P. A. Barber, Q. Yang, T. Li, D. G. Darby, R. P. Gerraty, S. M. Davis, et al. The Value of Apparent Diffusion Coefficient Maps in Early Cerebral Ischemia AJNR Am. J. Neuroradiol., August 1, 2001; 22(7): 1260 - 1267. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K. M. Gauvain, R. C. McKinstry, P. Mukherjee, A. Perry, J. J. Neil, B. A. Kaufman, and R. J. Hayashi Evaluating Pediatric Brain Tumor Cellularity with Diffusion-Tensor Imaging Am. J. Roentgenol., August 1, 2001; 177(2): 449 - 454. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J H Gillard, N G Papadakis, K Martin, C J S Price, E A Warburton, N M Antoun, C L-H Huang, T A Carpenter, and J D Pickard MR diffusion tensor imaging of white matter tract disruption in stroke at 3 T Br. J. Radiol., July 1, 2001; 74(883): 642 - 647. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
P. Mukherjee, R. C. McKinstry, J. S. Shimony, E. Akbudak, A. Z. Snyder, T. E. Conturo, M. M. Bahn, T. Back, and A. Gass Heterogeneity of Apparent Diffusion Coefficients Within Infarcts Response Stroke, July 1, 2001; 32(7): 1695 - 1696. [Full Text] [PDF]
|
|
|
|
 |
|
 P. Mukherjee and R. C. McKinstry Reversible Posterior Leukoencephalopathy Syndrome: Evaluation with Diffusion-Tensor MR Imaging Radiology, June 1, 2001; 219(3): 756 - 765. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
P. Mukherjee, J. H. Miller, J. S. Shimony, T. E. Conturo, B. C. P. Lee, C. R. Almli, and R. C. McKinstry Normal Brain Maturation during Childhood: Developmental Trends Characterized with Diffusion-Tensor MR Imaging Radiology, November 1, 2001; 221(2): 349 - 358. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
