- Email: [email protected]
Alteration of diffusion tensor parameters in postmortem brain
Available online at www.sciencedirect.com
Magnetic Resonance Imaging 27 (2009) 865 – 870
Alteration of diffusion tensor parameters in postmortem brain Elysa Widjaja a,⁎, Xingchang Wei b , Logi Vidarsson a , Rahim Moineddin c , Christopher K. Macgowan a,d , Daniel Nilsson a a Diagnostic Imaging, Hospital for Sick Children, Toronto, M5G 1X8, Canada Diagnostic Imaging, Alberta Children’s Hospital, Calgary, Alberta, T2W 3N2, Canada c Department of Public Health, Family and Community Medicine, Faculty of Medicine, University of Toronto, Toronto, M5T 3M7, Canada d Department of Medical Biophysics and Medical Imaging, University of Toronto, Toronto, M5G 2M9, Canada Received 14 July 2008; revised 2 October 2008; accepted 17 November 2008 b
Abstract In autopsy of humans, there is usually an interval of hours to days between death and tissue fixation, during which the cadaver is stored below room temperature to retard tissue autolysis. We have attempted to model this process and evaluate the alteration in diffusion indices of the postmortem brain in pigs, which were kept at 4°C. The pigs were scanned prior to death and at 3, 6, 9, 12, 18, 24, 30, 36, 42, 48 and 72 h postmortem. Regions of interest were placed in the corpus callosum, internal capsule, periventricular and subcortical white matter anteriorly and posteriorly. There was a slight increase in fractional anisotropy (FA) in the first 3 h postmortem. The FA remained stable up to 72 h postmortem. There was a marked decrease in trace, eigenmajor (λmajor), eigenmedium (λmedium) and eigenminor (λminor), particularly in the first 3 h following death. This study supports the utility of measuring diffusion anisotropy if the time elapsed between death and tissue fixation is within 3 days. However, trace and eigenvalues decreased markedly within the first few hours postmortem. Therefore trace and eigenvalues obtained from ex vivo studies cannot be extrapolated to in vivo studies. © 2009 Elsevier Inc. All rights reserved. Keywords: Postmortem; Diffusion tensor imaging; Brain
1. Introduction Diffusion tensor imaging (DTI) is increasingly used in postmortem specimens to facilitate correlation of DTI assessment of brain white matter to histological findings of animal models of diseases [1,2]. One of the concerns in postmortem DTI study is the autolysis and bacterial decomposition of brain tissue that may result in significant change in DTI characteristics. This confounding factor can be inhibited in experimental animal models by scanning brain tissue that is chemically fixed either by premortem perfusion fixation or directly postmortem fixation. Formalin fixation retards autolysis and stabilizes the cellular and tissue constituents [3]. Sun et al. [4] have found no difference between the relative anisotropy of live and formalin-fixed mice brain.
⁎ Corresponding author. E-mail address: [email protected] (E. Widjaja). 0730-725X/$ – see front matter © 2009 Elsevier Inc. All rights reserved. doi:10.1016/j.mri.2008.11.009
In autopsy of humans, there is usually an interval of several hours to days between death and tissue fixation, during which the cadaver is stored below room temperature to retard tissue autolysis. During this interval, DTI parameters may be altered by tissue autolysis, even though autolysis occurs at a slower rate due to the lower temperature. It is anticipated that the longer the time interval between death and tissue fixation, the greater the severity of tissue decomposition. D'Arceuil and de Crespigny [5] have attempted to model the process of tissue retrieval from human cadaver prior to tissue fixation by examining the brain of mice. These authors have performed DTI at 0, 1, 4 and 14 days postmortem, with interval placement of the brain at 4°C. They have found progressive reduction in fractional anisotropy (FA), trace and eigenvalues of the mice brain in the first 2 weeks postmortem. In most pathology departments, the cadavers are kept at temperature of 4°C for only a few days prior to tissue fixation, rather than over a period of 2 weeks. Therefore it is important to determine whether diffusion characteristics are altered in the first 2 to 3 days
866
E. Widjaja et al. / Magnetic Resonance Imaging 27 (2009) 865–870
postmortem. If the diffusion characteristics were found changed, it would be useful to determine how they were changed, in order to direct future studies in terms of the ideal postmortem delay. We hypothesized that in the first 3 days postmortem, there is no significant alteration in the FA of the white matter but there is a reduction in trace and eigenvalues. The purpose of this study was to evaluate the alteration in measures of FA, trace and eigenvalues of the postmortem brain in piglets at close intervals in the first 3 days postmortem.
2. Methods All procedures were conducted according to criteria established by the Canadian Council on Animal Care and were approved by the Hospital for Sick Children Research Institute Animal Care Review Committee. Six male Yorkshire piglets (12±3 kg) were scanned under general anesthesia whilst they were still alive. Following intraperitoneal injection of pentobarbital, the animals were decapitated and the heads kept in the skull at 4°C. The pigs' head were removed from the refrigerator and scanned at 3, 6, 9, 12, 18, 24, 30, 36, 42, 48 and 72 h following death. The duration of time that the pigs' head were out of the refrigerator, including the time to scan, was 9 min. The DTI was acquired using a 1.5-T GE Signa EXCITE III MR scanner (General Electric, Milwaukee, WI, USA) with an eight-channel knee coil. Single-shot diffusion-
weighted echo planar imaging was used for the DTI sequence. The diffusion weighting b-factor was set to 1000 s/mm2, and the number of directions used was 7. The other parameters were repetition time (TR)=8300 ms; time to echo (TE)=113 ms; slice thickness=2.2 mm with no gap; field of view (FOV)=16 cm; matrix of 128×128 reconstructed to 256×256; scan time was 8 min. To enhance the signal-to-noise ratio and reduce temporal phase fluctuations, eight consecutive measurements were acquired and averaged. Echo planar image distortion was corrected automatically on the scanner. The raw data that have been eddy current corrected were then post processed on DTIStudio V 2.4 (Johns Hopkins Medical Institute, Laboratory of Brain Anatomical MRI, http://lbam.med.jhmi.edu/). The DTI raw datasets were fitted to the diffusion tensor equations to yield six independent tensor elements using a nonlinear leastsquares fitting routine [6]. Diagonalization of the tensor yielded three eigenvalues (λmajor, λmedium and λminor) and three eigenvectors [7,8]. From these elements, maps of FA, trace and tensor eigenvalues were calculated. Regions of interest (ROI) were placed in the corpus callosum, right and left internal capsules, right and left periventricular white matter anteriorly and posteriorly, and right and left subcortical white matter anteriorly and posteriorly (Fig. 1). The ROIs measured 12 voxels in size each. The ROIs were placed on the b=0 images and then transposed onto anatomically coregistered positions on the FA, trace, major, medium and minor eigenvalue maps. The
Fig. 1. Axial color-coded FA maps demonstrating the placement of ROIs in the corpus callosum (small arrowhead), internal capsules (medium arrowhead), periventricular white matter anteriorly (small arrow) and posteriorly (large arrow), and in the subcortical white matter anteriorly (large arrowhead) and posteriorly (large arrowhead).
E. Widjaja et al. / Magnetic Resonance Imaging 27 (2009) 865–870
867
Fig. 2. Plots of diffusion parameters (from the corpus callosum, internal capsule, subcortical and periventricular white matter) vs. time for (A) FA (mean±2 S.D.), (B) trace (×10−3 mm2 s−1) (mean±2 S.D.), (C) eigenmajor (×10−3 mm2 s−1) (mean±2 S.D.), (D) eigenmedium (×10−3 mm2 s−1) (mean±2 S.D.), (E) eigenminor (×10−3 mm2 s−1) (mean±2 S.D.). There was a small but significant increase in FA and a significant reduction in trace, eigenmajor, eigenmedium and eigenminor at 3 h postmortem compared to when the pigs were alive, at 0 h.
mean values of FA, trace, λmajor, λmedium, λminor values were measured. Statistical analysis was performed using SAS 9.1 (SAS Institute, Cary, NC, USA). The FA, trace and eigenvalues of the right and left internal capsules, the right and left periventricular white matter anteriorly and posteriorly, and the right and left subcortical white matter anteriorly and posteriorly were compared using paired t test. One-way analysis of variance (ANOVA) was used to assess whether there was a difference between different time points in mean FA, trace and eigenvalues of all the different ROIs.
Subsequently, paired t test was performed between DTI measurements at the time when the pigs were alive (time=0 h) and 3 h postmortem and between 3 h and subsequent times postmortem. A P value of b.05 was considered statistically significant. Linear and nonlinear regression was performed between DTI measurements from different ROIs and postmortem time. In linear regression, DTI indices were modeled as a linear function of time, (α+β time), and in nonlinear regression DTI indices were modeled as a rational function of time . 1 a + b time
868
E. Widjaja et al. / Magnetic Resonance Imaging 27 (2009) 865–870
3. Results There was no significant difference in all the DTI measurements between right and left internal capsules (PN.05 for all DTI measurements), the right and left periventricular white matter anteriorly or posteriorly (PN.05 for all DTI measurements), or the right and left subcortical white matter anteriorly or posteriorly (PN.05 for all DTI measurements). The mean of the right and left measurements was then used for subsequent analysis. One-way ANOVA of the FA, trace, major, medium and minor eigenvalues of all the ROIs demonstrated a significant difference at different time points (Pb.005 in analyses of all five DTI measurements in all ROIs). There was an increase in mean FA in the first 3 h postmortem compared to the FA when the piglets were alive (P=.01). A subsequent one-way ANOVA of the FA of all the time points after 3 h postmortem was per-
formed, which demonstrated no difference in FA between 3 h postmortem and subsequent time points postmortem (PN.05) in any of the ROIs. The mean FA of all the ROIs and the two standard deviations at each time point are shown in Fig. 2. The same statistical approach was applied to trace and eigenvalues. There was a marked decrease in trace in the first 3 h postmortem compared to the brain when the piglets were alive (P=.006). However, there was no difference in mean trace between 3 h postmortem and subsequent time point postmortem (PN.05). The superimposed nonparametric line fitted to the data showed a nonlinear relationship between postmortem interval and trace (Table 1). The trace of the corpus callosum, internal capsule, periventricular and subcortical white matter of the postmortem brain followed the same pattern of alteration. The mean trace of all the ROIs and the two standard deviations at each time point are demonstrated in Fig. 2.
Table 1 Linear and nonlinear regression of FA, trace, λmajor, λmedium and λminor from different ROIs and time points postmortem Linear regression (h)
R2
Nonlinear regression (h)
R2
FA Corpus callosum Internal capsule Periventricular anterior Periventricular posterior Subcortical anterior Subcortical posterior
0.56+0.001 0.61+0.001 0.55+0.001 0.61−0.001 0.50+0.004 0.50+0.003
0.03 0.03 0.05 0.02 0.30 0.21
1/(1.77−0.003) 1/(1.63−0.002) 1/(1.81−0.003) 1/(1.65−0.002) 1/(1.93−0.009) 1/(1.92−0.007)
0.02 0.03 0.04 0.02 0.27 0.17
Trace (×10−3 mm2 s−1) Corpus callosum Internal capsule Periventricular anterior Periventricular posterior Subcortical anterior Subcortical posterior
1.15−0.014 1.14−0.012 1.15−0.014 1.16−0.014 1.20−0.012 1.20−0.011
0.32 0.33 0.30 0.36 0.23 0.22
1/(0.62+0.049) 1/(0.74+0.026) 1/(0.59+0.053) 1/(0.64+0.044) 1/(0.65+0.030) 1/(0.64+0.030)
0.60 0.47 0.62 0.63 0.40 0.39
λmajor (×10−3 mm2 s−1) Corpus callosum Internal capsule Periventricular anterior Periventricular posterior Subcortical anterior Subcortical posterior
0.62−0.007 0.67−0.007 0.62−0.007 0.67−0.008 0.63−0.004 0.62−0.004
0.38 0.37 0.33 0.42 0.13 0.11
1/(1.30+0.060) 1/(1.32+0.036) 1/(1.21+0.076) 1/(1.18+0.067) 1/(1.42+0.027) 1/(1.49+0.023)
0.59 0.48 0.59 0.66 0.19 0.15
λmedium (×10−3 mm2 s−1) Corpus callosum Internal capsule Periventricular anterior Periventricular posterior Subcortical anterior Subcortical posterior
0.34−0.004 0.31−0.003 0.36−0.003 0.34−0.003 0.37−0.002 0.38−0.003
0.23 0.26 0.20 0.23 0.09 0.09
1/(2.14+0.134) 1/(2.82+0.075) 1/(2.17+0.098) 1/(2.39+0.097) 1/(2.43+0.040) 1/(2.22+0.050)
0.43 0.36 0.36 0.38 0.14 0.15
λminor (×10−3 mm2 s−1) Corpus callosum Internal capsule Periventricular anterior Periventricular posterior Subcortical anterior Subcortical posterior
0.20−0.004 0.16−0.002 0.18−0.003 0.15−0.002 0.20−0.004 0.20−0.004
0.25 0.19 0.27 0.21 0.32 0.31
1/(2.60+0.811) 1/(3.83+0.587) 1/(2.64+1.079) 1/(3.51+0.832) 1/(2.57+0.917) 1/(2.49+0.948)
0.70 0.48 0.83 0.63 0.75 0.79
E. Widjaja et al. / Magnetic Resonance Imaging 27 (2009) 865–870
There was also a marked decrease in λmajor, λmedium and λminor in the first 3 h postmortem as compared to the time when the pigs were alive (P=.006, P=.01 and P=.004, respectively). However, there was no significant difference in mean λmajor, λmedium and λminor 3 h postmortem and at subsequent time point postmortem (PN.05). The decrease was also more pronounced for λminor compared to λmedium and λmajor. There was also a nonlinear relationship between postmortem time interval and the three eigenvalues (Table 1).
4. Discussion There is no consensus in the literature as to whether FA remains constant or decreased following death at different postmortem time intervals and with different techniques of tissue fixation. Kim et al. [9] have reported no significant difference in relative anisotropy of the ventrolateral white matter of five mice spinal cord in vivo and up to 10 h postmortem. However, they have observed a statistically significant reduction in the relative anisotropy of the white matter of the spinal cord 2 weeks after immersion fixation and 8% reduction in relative anisotropy at 15 weeks after fixation. In contrast, D'Arceuil and de Crespigny [5] have found a progressive reduction in FA at 1, 4 and 14 days postmortem, with reduction in FA of approximately 28% at 14 days postmortem. Sun et al. [4,10] have evaluated seven male mice and have found no difference in FA between live and formalin-fixed mice brain that were subjected to perfusion fixation. They have also found the FA was relatively preserved in both acutely injured brain and normal brain in formalin-fixed mouse brain [4,10], suggesting that acute brain injury did not alter the tissue decomposition. Madi et al. [11] have found a reduction in FA of a rat spinal cord white matter ex vivo compared to that observed in vivo. We have found a slight elevation in FA in the first 3 h postmortem compared to the brains when the piglets were alive. The FA remained relatively stable afterwards for 72 h postmortem. Elevated FA in the first 3 h postmortem has not been described in the literature. However, an increase in FA has been reported following an acute infarction less than 24 h after the onset of symptoms [12–14]. The elevated FA has been postulated to be related to increased viscosity in the intracellular space due to macromolecular structural protein breakdown, increased tortuosity in the shrunken extracellular space and restriction of water movement in the shrunken extracellular space [12]. The lack of major reduction in FA between 3 h and 72 h postmortem may be related to retardation of tissue decomposition below room temperature. In the study by D'Arceuil and de Crespigny [5], regional FA of gray and white matter decreased significantly with time. The reason for the discrepancy between their results and ours is unknown. It is noted that in their study the mice were scanned 18 days after fixation. This long postmortem time interval and the tissue fixation technique may play a role.
869
We have used diffusion weighting with seven diffusion encoding directions (DEDs) and eight numbers of signal averaging (NSA). This is in contrast to the study by D'Arceuil and de Crespigny [5], where 20 DEDs and one signal averaging were used. Jones [15] has found that at least 20 DEDs are needed for robust estimation of anisotropy using Monte Carlo simulations. However, Ni et al. [16] have evaluated three combinations of DEDs and NSA (6DED10NSA, 21DED3NSA and 31DED2NSA) in 15 volunteers, which allowed for similar scan times. They have found no difference in FA amongst the three different protocols. The difference in DTI technique alone was unlikely to have contributed to significant differences in the findings between our study and the one by D'Arceuil and de Crespigny [5]. A 50–70% reduction in trace has been reported following perfusion fixation [4,10], when there was no time interval between death and tissue fixation. Kim et al. [9] have found significant reduction in trace within 5 h of death, before tissue fixation. In our study, the reduction in trace postmortem was nonlinear over time as demonstrated by higher R2 in nonlinear regression analysis as compared to linear regression analysis and was most marked in the first 3 h postmortem. Our finding of reduced trace in the first few h postmortem was in keeping with the findings reported in the literature. Due to the rapid reduction in trace following death, trace values should not be extrapolated from ex vivo to in vivo studies. The reduction in trace may be secondary to changes in energy-requiring processes such as cytoplasmic streaming or water compartment shifts associated with cell swelling [10]. The marked reduction in trace and relative preservation of FA in our study suggest that there was a proportional reduction in the apparent diffusion coefficient values in all directions. In this study, we have found a reduction in λmajor, λmedium and λminor of all the white matter structures evaluated, which demonstrated a nonlinear relationship with time, and this was more pronounced in the first 3 h postmortem. D'Arceuil and de Crespigny [5] have also found a reduction in all the three eigenvalues of both the white matter and the gray matter in the first 14 days postmortem, and the reduction was most pronounced with λmajor. Kim et al. [9] have also reported a reduction in parallel and perpendicular diffusivity in the ventrolateral white matter of the mouse spinal cord at 5 h postmortem and this remained persistently reduced 2 and 15 weeks following tissue fixation. The three eigenvalues fully determine the size and shape of the diffusion ellipsoid corresponding to the diffusion tensor. λmajor represents water diffusivity parallel to the axonal fibers, also referred to as axial diffusivity, whilst λmedium and λminor represent water diffusion perpendicular to the axonal fibers, that is, radial or transverse diffusivities. Animal models have shown the tensor eigenvalues were more specific markers of myelination and axonal morphology [17–19]. Changes in λmajor reflect axonal integrity and have been observed in cases of Wallerian degeneration [20]. In contrast, transverse or radial diffusivities reflect alteration in myelination [19,21].
870
E. Widjaja et al. / Magnetic Resonance Imaging 27 (2009) 865–870
D'Arceuil and de Crespigny et al.'s findings of reduced parallel diffusivity suggested a more prominent effect of axonal disintegration relative to myelin degradation. Electron microscopic and electrophoretic studies of white matter that has been stored at 4°C have demonstrated marked autolysis of glial cells cytoplasm and axoplasm 24 h postbiopsy but relative preservation of myelin sheaths [22]. Due to the difference in temperature of the brain when the piglets were alive and at postmortem, the lower temperature at postmortem may have a confounding effect on the difference in DTI indices on the postmortem brain and in the live brain. Le Bihan [23] reported a 2.4% change in water diffusion with a 1°C change in temperature. The observed reduction in trace and eigenvalues in our study could in part be attributed to the reduction in temperature of the postmortem brain. Sun et al. [4] have observed a 50–75% drop in water diffusion of the fixed mice brain even though the temperature of their fixed mice brain decreased by 16°C, which would have resulted in an expected reduction of 38% of water diffusion. Based on the findings by Le Bihan [23], we expect a 33°C change in temperature to produce 79% change in diffusivity. In this study, we have found a difference of 43–70% in diffusivity when the pigs were alive and at postmortem. The extent of contribution of the difference in temperature and the effects of postmortem tissue ischemia to the difference in diffusivity was not entirely clear. We have not scanned the postmortem tissue at 37°C as we attempted to model the effects of DTI changes in postmortem brain in humans prior to tissue fixation, and the human brains are usually kept at 4°C in pathology departments prior to tissue fixation. Kim et al. [9] have found no difference in FA of the white matter of the mouse spinal cord when the mouse was scanned at a temperature of 37°C compared to when the mouse was scanned postmortem in situ at room temperature. We have found a slight but significant increase in FA at 3 h postmortem when the pigs were scanned at 4°C as compared to the time point when the pigs were alive at 37°C. The difference in FA may have in part been contributed by the difference in tissue temperature. In summary, we have found a slight increase in FA in the first 3 h postmortem, but the FA remained stable for the first 3 days postmortem when the brain was kept below room temperature at 4°C. This study supports the utility of measuring diffusion anisotropy if the time elapsed between death and tissue fixation is within 3 days. This time scale reflects the typical time interval prior to tissue fixation in most pathology departments. However, trace and eigenvalues decreased markedly within the first few hours postmortem and therefore trace and eigenvalues obtained from ex vivo studies cannot be extrapolated to in vivo studies. References [1] Song SK, Yoshino J, Le TQ, Lin SJ, Sun SW, Cross AH, et al. Demyelination increases radial diffusivity in corpus callosum of mouse brain. Neuroimage 2005;26:132–40.
[2] Zhang J, Miller MI, Plachez C, Richards LJ, Yarowsky P, van Zijl P, et al. Mapping postnatal mouse brain development with diffusion tensor microimaging. Neuroimage 2005;26:1042–51. [3] Srinivasan M, Sedmak D, Jewell S. Effect of fixatives and tissue processing on the content and integrity of nucleic acids. Am J Pathol 2002;161:1961–71. [4] Sun SW, Neil JJ, Song SK. Relative indices of water diffusion anisotropy are equivalent in live and formalin-fixed mouse brains. Magn Reson Med 2003;50:743–8. [5] D'Arceuil H, de Crespigny A. The effects of brain tissue decomposition on diffusion tensor imaging and tractography. Neuroimage 2007; 36:64–8. [6] Bruder H, Fischer H, Reinfelder HE, Schmitt F. Image reconstruction for echo planar imaging with nonequidistant k-space sampling. Magn Reson Med 1992;23:311–23. [7] Basser PJ, Mattiello J, LeBihan D. MR Diffusion tensor spectroscopy and imaging. Biophys J 1994;66:259–67. [8] Basser PJ, Mattiello J, LeBihan D. Estimation of the effective selfdiffusion tensor from the NMR spin echo. J Magn Reson B 1994;103: 247–54. [9] Kim JH, Trinkaus K, Ozcan A, Budde MD, Song SK. Postmortem delay does not change regional diffusion anisotropy characteristics in mouse spinal cord white matter. NMR Biomed 2007;20:352–9. [10] Sun SW, Neil JJ, Liang HF, He YY, Schmidt RE, Hsu CY, et al. Formalin fixation alters water diffusion coefficient magnitude but not anisotropy in infarcted brain. Magn Reson Med 2005;53:1447–51. [11] Madi S, Hasan KM, Narayana PA. Diffusion tensor imaging of in vivo and excised rat spinal cord at 7 T with an icosahedral encoding scheme. Magn Reson Med 2005;53:118–25. [12] Ozsunar Y, Grant PE, Huisman TA, Schaefer PW, Wu O, Sorensen AG, et al. Evolution of water diffusion and anisotropy in hyperacute stroke: significant correlation between fractional anisotropy and T2. AJNR Am J Neuroradiol 2004;25:699–705. [13] Yang Q, Tress BM, Barber PA, Desmond PM, Darby DG, Gerraty RP, et al. Serial study of apparent diffusion coefficient and anisotropy in patients with acute stroke. Stroke 1999;30:2382–90. [14] Armitage PA, Bastin ME, Marshall I, Wardlaw JM, Cannon J. Diffusion anisotropy measurements in ischaemic stroke of the human brain. Magma 1998;6:28–36. [15] Jones DK. The effect of gradient sampling schemes on measures derived from diffusion tensor MRI: a Monte Carlo study. Magn Reson Med 2004;51:807–15. [16] Ni H, Kavcic V, Zhu T, Ekholm S, Zhong J. Effects of number of diffusion gradient directions on derived diffusion tensor imaging indices in human brain. AJNR Am J Neuroradiol 2006;27:1776–81. [17] Kinoshita Y, Ohnishi A, Kohshi K, Yokota A. Apparent diffusion coefficient on rat brain and nerves intoxicated with methylmercury. Environ Res 1999;80:348–54. [18] Ono J, Harada K, Takahashi M, Maeda M, Ikenaka K, Sakurai K, et al. Differentiation between dysmyelination and demyelination using magnetic resonance diffusional anisotropy. Brain Res 1995;671: 141–8. [19] Song SK, Sun SW, Ramsbottom MJ, Chang C, Russell J, Cross AH. Dysmyelination revealed through MRI as increased radial (but unchanged axial) diffusion of water. Neuroimage 2002;17:1429–36. [20] Thomalla G, Glauche V, Koch MA, Beaulieu C, Weiller C, Rother J. Diffusion tensor imaging detects early Wallerian degeneration of the pyramidal tract after ischemic stroke. Neuroimage 2004;22:1767–74. [21] Song SK, Sun SW, Ju WK, Lin SJ, Cross AH, Neufeld AH. Diffusion tensor imaging detects and differentiates axon and myelin degeneration in mouse optic nerve after retinal ischemia. Neuroimage 2003;20: 1714–22. [22] Hukkanen V, Roytta M. Autolytic changes of human white matter: an electron microscopic and electrophoretic study. Exp Mol Pathol 1987; 46:31–9. [23] Le Bihan D. Diffusion and perfusion magnetic resonance imaging applications to functional MRI. New York: Raven Press; 1995.
Magnetic Resonance Imaging 27 (2009) 865 – 870
Alteration of diffusion tensor parameters in postmortem brain Elysa Widjaja a,⁎, Xingchang Wei b , Logi Vidarsson a , Rahim Moineddin c , Christopher K. Macgowan a,d , Daniel Nilsson a a Diagnostic Imaging, Hospital for Sick Children, Toronto, M5G 1X8, Canada Diagnostic Imaging, Alberta Children’s Hospital, Calgary, Alberta, T2W 3N2, Canada c Department of Public Health, Family and Community Medicine, Faculty of Medicine, University of Toronto, Toronto, M5T 3M7, Canada d Department of Medical Biophysics and Medical Imaging, University of Toronto, Toronto, M5G 2M9, Canada Received 14 July 2008; revised 2 October 2008; accepted 17 November 2008 b
Abstract In autopsy of humans, there is usually an interval of hours to days between death and tissue fixation, during which the cadaver is stored below room temperature to retard tissue autolysis. We have attempted to model this process and evaluate the alteration in diffusion indices of the postmortem brain in pigs, which were kept at 4°C. The pigs were scanned prior to death and at 3, 6, 9, 12, 18, 24, 30, 36, 42, 48 and 72 h postmortem. Regions of interest were placed in the corpus callosum, internal capsule, periventricular and subcortical white matter anteriorly and posteriorly. There was a slight increase in fractional anisotropy (FA) in the first 3 h postmortem. The FA remained stable up to 72 h postmortem. There was a marked decrease in trace, eigenmajor (λmajor), eigenmedium (λmedium) and eigenminor (λminor), particularly in the first 3 h following death. This study supports the utility of measuring diffusion anisotropy if the time elapsed between death and tissue fixation is within 3 days. However, trace and eigenvalues decreased markedly within the first few hours postmortem. Therefore trace and eigenvalues obtained from ex vivo studies cannot be extrapolated to in vivo studies. © 2009 Elsevier Inc. All rights reserved. Keywords: Postmortem; Diffusion tensor imaging; Brain
1. Introduction Diffusion tensor imaging (DTI) is increasingly used in postmortem specimens to facilitate correlation of DTI assessment of brain white matter to histological findings of animal models of diseases [1,2]. One of the concerns in postmortem DTI study is the autolysis and bacterial decomposition of brain tissue that may result in significant change in DTI characteristics. This confounding factor can be inhibited in experimental animal models by scanning brain tissue that is chemically fixed either by premortem perfusion fixation or directly postmortem fixation. Formalin fixation retards autolysis and stabilizes the cellular and tissue constituents [3]. Sun et al. [4] have found no difference between the relative anisotropy of live and formalin-fixed mice brain.
⁎ Corresponding author. E-mail address: [email protected] (E. Widjaja). 0730-725X/$ – see front matter © 2009 Elsevier Inc. All rights reserved. doi:10.1016/j.mri.2008.11.009
In autopsy of humans, there is usually an interval of several hours to days between death and tissue fixation, during which the cadaver is stored below room temperature to retard tissue autolysis. During this interval, DTI parameters may be altered by tissue autolysis, even though autolysis occurs at a slower rate due to the lower temperature. It is anticipated that the longer the time interval between death and tissue fixation, the greater the severity of tissue decomposition. D'Arceuil and de Crespigny [5] have attempted to model the process of tissue retrieval from human cadaver prior to tissue fixation by examining the brain of mice. These authors have performed DTI at 0, 1, 4 and 14 days postmortem, with interval placement of the brain at 4°C. They have found progressive reduction in fractional anisotropy (FA), trace and eigenvalues of the mice brain in the first 2 weeks postmortem. In most pathology departments, the cadavers are kept at temperature of 4°C for only a few days prior to tissue fixation, rather than over a period of 2 weeks. Therefore it is important to determine whether diffusion characteristics are altered in the first 2 to 3 days
866
E. Widjaja et al. / Magnetic Resonance Imaging 27 (2009) 865–870
postmortem. If the diffusion characteristics were found changed, it would be useful to determine how they were changed, in order to direct future studies in terms of the ideal postmortem delay. We hypothesized that in the first 3 days postmortem, there is no significant alteration in the FA of the white matter but there is a reduction in trace and eigenvalues. The purpose of this study was to evaluate the alteration in measures of FA, trace and eigenvalues of the postmortem brain in piglets at close intervals in the first 3 days postmortem.
2. Methods All procedures were conducted according to criteria established by the Canadian Council on Animal Care and were approved by the Hospital for Sick Children Research Institute Animal Care Review Committee. Six male Yorkshire piglets (12±3 kg) were scanned under general anesthesia whilst they were still alive. Following intraperitoneal injection of pentobarbital, the animals were decapitated and the heads kept in the skull at 4°C. The pigs' head were removed from the refrigerator and scanned at 3, 6, 9, 12, 18, 24, 30, 36, 42, 48 and 72 h following death. The duration of time that the pigs' head were out of the refrigerator, including the time to scan, was 9 min. The DTI was acquired using a 1.5-T GE Signa EXCITE III MR scanner (General Electric, Milwaukee, WI, USA) with an eight-channel knee coil. Single-shot diffusion-
weighted echo planar imaging was used for the DTI sequence. The diffusion weighting b-factor was set to 1000 s/mm2, and the number of directions used was 7. The other parameters were repetition time (TR)=8300 ms; time to echo (TE)=113 ms; slice thickness=2.2 mm with no gap; field of view (FOV)=16 cm; matrix of 128×128 reconstructed to 256×256; scan time was 8 min. To enhance the signal-to-noise ratio and reduce temporal phase fluctuations, eight consecutive measurements were acquired and averaged. Echo planar image distortion was corrected automatically on the scanner. The raw data that have been eddy current corrected were then post processed on DTIStudio V 2.4 (Johns Hopkins Medical Institute, Laboratory of Brain Anatomical MRI, http://lbam.med.jhmi.edu/). The DTI raw datasets were fitted to the diffusion tensor equations to yield six independent tensor elements using a nonlinear leastsquares fitting routine [6]. Diagonalization of the tensor yielded three eigenvalues (λmajor, λmedium and λminor) and three eigenvectors [7,8]. From these elements, maps of FA, trace and tensor eigenvalues were calculated. Regions of interest (ROI) were placed in the corpus callosum, right and left internal capsules, right and left periventricular white matter anteriorly and posteriorly, and right and left subcortical white matter anteriorly and posteriorly (Fig. 1). The ROIs measured 12 voxels in size each. The ROIs were placed on the b=0 images and then transposed onto anatomically coregistered positions on the FA, trace, major, medium and minor eigenvalue maps. The
Fig. 1. Axial color-coded FA maps demonstrating the placement of ROIs in the corpus callosum (small arrowhead), internal capsules (medium arrowhead), periventricular white matter anteriorly (small arrow) and posteriorly (large arrow), and in the subcortical white matter anteriorly (large arrowhead) and posteriorly (large arrowhead).
E. Widjaja et al. / Magnetic Resonance Imaging 27 (2009) 865–870
867
Fig. 2. Plots of diffusion parameters (from the corpus callosum, internal capsule, subcortical and periventricular white matter) vs. time for (A) FA (mean±2 S.D.), (B) trace (×10−3 mm2 s−1) (mean±2 S.D.), (C) eigenmajor (×10−3 mm2 s−1) (mean±2 S.D.), (D) eigenmedium (×10−3 mm2 s−1) (mean±2 S.D.), (E) eigenminor (×10−3 mm2 s−1) (mean±2 S.D.). There was a small but significant increase in FA and a significant reduction in trace, eigenmajor, eigenmedium and eigenminor at 3 h postmortem compared to when the pigs were alive, at 0 h.
mean values of FA, trace, λmajor, λmedium, λminor values were measured. Statistical analysis was performed using SAS 9.1 (SAS Institute, Cary, NC, USA). The FA, trace and eigenvalues of the right and left internal capsules, the right and left periventricular white matter anteriorly and posteriorly, and the right and left subcortical white matter anteriorly and posteriorly were compared using paired t test. One-way analysis of variance (ANOVA) was used to assess whether there was a difference between different time points in mean FA, trace and eigenvalues of all the different ROIs.
Subsequently, paired t test was performed between DTI measurements at the time when the pigs were alive (time=0 h) and 3 h postmortem and between 3 h and subsequent times postmortem. A P value of b.05 was considered statistically significant. Linear and nonlinear regression was performed between DTI measurements from different ROIs and postmortem time. In linear regression, DTI indices were modeled as a linear function of time, (α+β time), and in nonlinear regression DTI indices were modeled as a rational function of time . 1 a + b time
868
E. Widjaja et al. / Magnetic Resonance Imaging 27 (2009) 865–870
3. Results There was no significant difference in all the DTI measurements between right and left internal capsules (PN.05 for all DTI measurements), the right and left periventricular white matter anteriorly or posteriorly (PN.05 for all DTI measurements), or the right and left subcortical white matter anteriorly or posteriorly (PN.05 for all DTI measurements). The mean of the right and left measurements was then used for subsequent analysis. One-way ANOVA of the FA, trace, major, medium and minor eigenvalues of all the ROIs demonstrated a significant difference at different time points (Pb.005 in analyses of all five DTI measurements in all ROIs). There was an increase in mean FA in the first 3 h postmortem compared to the FA when the piglets were alive (P=.01). A subsequent one-way ANOVA of the FA of all the time points after 3 h postmortem was per-
formed, which demonstrated no difference in FA between 3 h postmortem and subsequent time points postmortem (PN.05) in any of the ROIs. The mean FA of all the ROIs and the two standard deviations at each time point are shown in Fig. 2. The same statistical approach was applied to trace and eigenvalues. There was a marked decrease in trace in the first 3 h postmortem compared to the brain when the piglets were alive (P=.006). However, there was no difference in mean trace between 3 h postmortem and subsequent time point postmortem (PN.05). The superimposed nonparametric line fitted to the data showed a nonlinear relationship between postmortem interval and trace (Table 1). The trace of the corpus callosum, internal capsule, periventricular and subcortical white matter of the postmortem brain followed the same pattern of alteration. The mean trace of all the ROIs and the two standard deviations at each time point are demonstrated in Fig. 2.
Table 1 Linear and nonlinear regression of FA, trace, λmajor, λmedium and λminor from different ROIs and time points postmortem Linear regression (h)
R2
Nonlinear regression (h)
R2
FA Corpus callosum Internal capsule Periventricular anterior Periventricular posterior Subcortical anterior Subcortical posterior
0.56+0.001 0.61+0.001 0.55+0.001 0.61−0.001 0.50+0.004 0.50+0.003
0.03 0.03 0.05 0.02 0.30 0.21
1/(1.77−0.003) 1/(1.63−0.002) 1/(1.81−0.003) 1/(1.65−0.002) 1/(1.93−0.009) 1/(1.92−0.007)
0.02 0.03 0.04 0.02 0.27 0.17
Trace (×10−3 mm2 s−1) Corpus callosum Internal capsule Periventricular anterior Periventricular posterior Subcortical anterior Subcortical posterior
1.15−0.014 1.14−0.012 1.15−0.014 1.16−0.014 1.20−0.012 1.20−0.011
0.32 0.33 0.30 0.36 0.23 0.22
1/(0.62+0.049) 1/(0.74+0.026) 1/(0.59+0.053) 1/(0.64+0.044) 1/(0.65+0.030) 1/(0.64+0.030)
0.60 0.47 0.62 0.63 0.40 0.39
λmajor (×10−3 mm2 s−1) Corpus callosum Internal capsule Periventricular anterior Periventricular posterior Subcortical anterior Subcortical posterior
0.62−0.007 0.67−0.007 0.62−0.007 0.67−0.008 0.63−0.004 0.62−0.004
0.38 0.37 0.33 0.42 0.13 0.11
1/(1.30+0.060) 1/(1.32+0.036) 1/(1.21+0.076) 1/(1.18+0.067) 1/(1.42+0.027) 1/(1.49+0.023)
0.59 0.48 0.59 0.66 0.19 0.15
λmedium (×10−3 mm2 s−1) Corpus callosum Internal capsule Periventricular anterior Periventricular posterior Subcortical anterior Subcortical posterior
0.34−0.004 0.31−0.003 0.36−0.003 0.34−0.003 0.37−0.002 0.38−0.003
0.23 0.26 0.20 0.23 0.09 0.09
1/(2.14+0.134) 1/(2.82+0.075) 1/(2.17+0.098) 1/(2.39+0.097) 1/(2.43+0.040) 1/(2.22+0.050)
0.43 0.36 0.36 0.38 0.14 0.15
λminor (×10−3 mm2 s−1) Corpus callosum Internal capsule Periventricular anterior Periventricular posterior Subcortical anterior Subcortical posterior
0.20−0.004 0.16−0.002 0.18−0.003 0.15−0.002 0.20−0.004 0.20−0.004
0.25 0.19 0.27 0.21 0.32 0.31
1/(2.60+0.811) 1/(3.83+0.587) 1/(2.64+1.079) 1/(3.51+0.832) 1/(2.57+0.917) 1/(2.49+0.948)
0.70 0.48 0.83 0.63 0.75 0.79
E. Widjaja et al. / Magnetic Resonance Imaging 27 (2009) 865–870
There was also a marked decrease in λmajor, λmedium and λminor in the first 3 h postmortem as compared to the time when the pigs were alive (P=.006, P=.01 and P=.004, respectively). However, there was no significant difference in mean λmajor, λmedium and λminor 3 h postmortem and at subsequent time point postmortem (PN.05). The decrease was also more pronounced for λminor compared to λmedium and λmajor. There was also a nonlinear relationship between postmortem time interval and the three eigenvalues (Table 1).
4. Discussion There is no consensus in the literature as to whether FA remains constant or decreased following death at different postmortem time intervals and with different techniques of tissue fixation. Kim et al. [9] have reported no significant difference in relative anisotropy of the ventrolateral white matter of five mice spinal cord in vivo and up to 10 h postmortem. However, they have observed a statistically significant reduction in the relative anisotropy of the white matter of the spinal cord 2 weeks after immersion fixation and 8% reduction in relative anisotropy at 15 weeks after fixation. In contrast, D'Arceuil and de Crespigny [5] have found a progressive reduction in FA at 1, 4 and 14 days postmortem, with reduction in FA of approximately 28% at 14 days postmortem. Sun et al. [4,10] have evaluated seven male mice and have found no difference in FA between live and formalin-fixed mice brain that were subjected to perfusion fixation. They have also found the FA was relatively preserved in both acutely injured brain and normal brain in formalin-fixed mouse brain [4,10], suggesting that acute brain injury did not alter the tissue decomposition. Madi et al. [11] have found a reduction in FA of a rat spinal cord white matter ex vivo compared to that observed in vivo. We have found a slight elevation in FA in the first 3 h postmortem compared to the brains when the piglets were alive. The FA remained relatively stable afterwards for 72 h postmortem. Elevated FA in the first 3 h postmortem has not been described in the literature. However, an increase in FA has been reported following an acute infarction less than 24 h after the onset of symptoms [12–14]. The elevated FA has been postulated to be related to increased viscosity in the intracellular space due to macromolecular structural protein breakdown, increased tortuosity in the shrunken extracellular space and restriction of water movement in the shrunken extracellular space [12]. The lack of major reduction in FA between 3 h and 72 h postmortem may be related to retardation of tissue decomposition below room temperature. In the study by D'Arceuil and de Crespigny [5], regional FA of gray and white matter decreased significantly with time. The reason for the discrepancy between their results and ours is unknown. It is noted that in their study the mice were scanned 18 days after fixation. This long postmortem time interval and the tissue fixation technique may play a role.
869
We have used diffusion weighting with seven diffusion encoding directions (DEDs) and eight numbers of signal averaging (NSA). This is in contrast to the study by D'Arceuil and de Crespigny [5], where 20 DEDs and one signal averaging were used. Jones [15] has found that at least 20 DEDs are needed for robust estimation of anisotropy using Monte Carlo simulations. However, Ni et al. [16] have evaluated three combinations of DEDs and NSA (6DED10NSA, 21DED3NSA and 31DED2NSA) in 15 volunteers, which allowed for similar scan times. They have found no difference in FA amongst the three different protocols. The difference in DTI technique alone was unlikely to have contributed to significant differences in the findings between our study and the one by D'Arceuil and de Crespigny [5]. A 50–70% reduction in trace has been reported following perfusion fixation [4,10], when there was no time interval between death and tissue fixation. Kim et al. [9] have found significant reduction in trace within 5 h of death, before tissue fixation. In our study, the reduction in trace postmortem was nonlinear over time as demonstrated by higher R2 in nonlinear regression analysis as compared to linear regression analysis and was most marked in the first 3 h postmortem. Our finding of reduced trace in the first few h postmortem was in keeping with the findings reported in the literature. Due to the rapid reduction in trace following death, trace values should not be extrapolated from ex vivo to in vivo studies. The reduction in trace may be secondary to changes in energy-requiring processes such as cytoplasmic streaming or water compartment shifts associated with cell swelling [10]. The marked reduction in trace and relative preservation of FA in our study suggest that there was a proportional reduction in the apparent diffusion coefficient values in all directions. In this study, we have found a reduction in λmajor, λmedium and λminor of all the white matter structures evaluated, which demonstrated a nonlinear relationship with time, and this was more pronounced in the first 3 h postmortem. D'Arceuil and de Crespigny [5] have also found a reduction in all the three eigenvalues of both the white matter and the gray matter in the first 14 days postmortem, and the reduction was most pronounced with λmajor. Kim et al. [9] have also reported a reduction in parallel and perpendicular diffusivity in the ventrolateral white matter of the mouse spinal cord at 5 h postmortem and this remained persistently reduced 2 and 15 weeks following tissue fixation. The three eigenvalues fully determine the size and shape of the diffusion ellipsoid corresponding to the diffusion tensor. λmajor represents water diffusivity parallel to the axonal fibers, also referred to as axial diffusivity, whilst λmedium and λminor represent water diffusion perpendicular to the axonal fibers, that is, radial or transverse diffusivities. Animal models have shown the tensor eigenvalues were more specific markers of myelination and axonal morphology [17–19]. Changes in λmajor reflect axonal integrity and have been observed in cases of Wallerian degeneration [20]. In contrast, transverse or radial diffusivities reflect alteration in myelination [19,21].
870
E. Widjaja et al. / Magnetic Resonance Imaging 27 (2009) 865–870
D'Arceuil and de Crespigny et al.'s findings of reduced parallel diffusivity suggested a more prominent effect of axonal disintegration relative to myelin degradation. Electron microscopic and electrophoretic studies of white matter that has been stored at 4°C have demonstrated marked autolysis of glial cells cytoplasm and axoplasm 24 h postbiopsy but relative preservation of myelin sheaths [22]. Due to the difference in temperature of the brain when the piglets were alive and at postmortem, the lower temperature at postmortem may have a confounding effect on the difference in DTI indices on the postmortem brain and in the live brain. Le Bihan [23] reported a 2.4% change in water diffusion with a 1°C change in temperature. The observed reduction in trace and eigenvalues in our study could in part be attributed to the reduction in temperature of the postmortem brain. Sun et al. [4] have observed a 50–75% drop in water diffusion of the fixed mice brain even though the temperature of their fixed mice brain decreased by 16°C, which would have resulted in an expected reduction of 38% of water diffusion. Based on the findings by Le Bihan [23], we expect a 33°C change in temperature to produce 79% change in diffusivity. In this study, we have found a difference of 43–70% in diffusivity when the pigs were alive and at postmortem. The extent of contribution of the difference in temperature and the effects of postmortem tissue ischemia to the difference in diffusivity was not entirely clear. We have not scanned the postmortem tissue at 37°C as we attempted to model the effects of DTI changes in postmortem brain in humans prior to tissue fixation, and the human brains are usually kept at 4°C in pathology departments prior to tissue fixation. Kim et al. [9] have found no difference in FA of the white matter of the mouse spinal cord when the mouse was scanned at a temperature of 37°C compared to when the mouse was scanned postmortem in situ at room temperature. We have found a slight but significant increase in FA at 3 h postmortem when the pigs were scanned at 4°C as compared to the time point when the pigs were alive at 37°C. The difference in FA may have in part been contributed by the difference in tissue temperature. In summary, we have found a slight increase in FA in the first 3 h postmortem, but the FA remained stable for the first 3 days postmortem when the brain was kept below room temperature at 4°C. This study supports the utility of measuring diffusion anisotropy if the time elapsed between death and tissue fixation is within 3 days. This time scale reflects the typical time interval prior to tissue fixation in most pathology departments. However, trace and eigenvalues decreased markedly within the first few hours postmortem and therefore trace and eigenvalues obtained from ex vivo studies cannot be extrapolated to in vivo studies. References [1] Song SK, Yoshino J, Le TQ, Lin SJ, Sun SW, Cross AH, et al. Demyelination increases radial diffusivity in corpus callosum of mouse brain. Neuroimage 2005;26:132–40.
[2] Zhang J, Miller MI, Plachez C, Richards LJ, Yarowsky P, van Zijl P, et al. Mapping postnatal mouse brain development with diffusion tensor microimaging. Neuroimage 2005;26:1042–51. [3] Srinivasan M, Sedmak D, Jewell S. Effect of fixatives and tissue processing on the content and integrity of nucleic acids. Am J Pathol 2002;161:1961–71. [4] Sun SW, Neil JJ, Song SK. Relative indices of water diffusion anisotropy are equivalent in live and formalin-fixed mouse brains. Magn Reson Med 2003;50:743–8. [5] D'Arceuil H, de Crespigny A. The effects of brain tissue decomposition on diffusion tensor imaging and tractography. Neuroimage 2007; 36:64–8. [6] Bruder H, Fischer H, Reinfelder HE, Schmitt F. Image reconstruction for echo planar imaging with nonequidistant k-space sampling. Magn Reson Med 1992;23:311–23. [7] Basser PJ, Mattiello J, LeBihan D. MR Diffusion tensor spectroscopy and imaging. Biophys J 1994;66:259–67. [8] Basser PJ, Mattiello J, LeBihan D. Estimation of the effective selfdiffusion tensor from the NMR spin echo. J Magn Reson B 1994;103: 247–54. [9] Kim JH, Trinkaus K, Ozcan A, Budde MD, Song SK. Postmortem delay does not change regional diffusion anisotropy characteristics in mouse spinal cord white matter. NMR Biomed 2007;20:352–9. [10] Sun SW, Neil JJ, Liang HF, He YY, Schmidt RE, Hsu CY, et al. Formalin fixation alters water diffusion coefficient magnitude but not anisotropy in infarcted brain. Magn Reson Med 2005;53:1447–51. [11] Madi S, Hasan KM, Narayana PA. Diffusion tensor imaging of in vivo and excised rat spinal cord at 7 T with an icosahedral encoding scheme. Magn Reson Med 2005;53:118–25. [12] Ozsunar Y, Grant PE, Huisman TA, Schaefer PW, Wu O, Sorensen AG, et al. Evolution of water diffusion and anisotropy in hyperacute stroke: significant correlation between fractional anisotropy and T2. AJNR Am J Neuroradiol 2004;25:699–705. [13] Yang Q, Tress BM, Barber PA, Desmond PM, Darby DG, Gerraty RP, et al. Serial study of apparent diffusion coefficient and anisotropy in patients with acute stroke. Stroke 1999;30:2382–90. [14] Armitage PA, Bastin ME, Marshall I, Wardlaw JM, Cannon J. Diffusion anisotropy measurements in ischaemic stroke of the human brain. Magma 1998;6:28–36. [15] Jones DK. The effect of gradient sampling schemes on measures derived from diffusion tensor MRI: a Monte Carlo study. Magn Reson Med 2004;51:807–15. [16] Ni H, Kavcic V, Zhu T, Ekholm S, Zhong J. Effects of number of diffusion gradient directions on derived diffusion tensor imaging indices in human brain. AJNR Am J Neuroradiol 2006;27:1776–81. [17] Kinoshita Y, Ohnishi A, Kohshi K, Yokota A. Apparent diffusion coefficient on rat brain and nerves intoxicated with methylmercury. Environ Res 1999;80:348–54. [18] Ono J, Harada K, Takahashi M, Maeda M, Ikenaka K, Sakurai K, et al. Differentiation between dysmyelination and demyelination using magnetic resonance diffusional anisotropy. Brain Res 1995;671: 141–8. [19] Song SK, Sun SW, Ramsbottom MJ, Chang C, Russell J, Cross AH. Dysmyelination revealed through MRI as increased radial (but unchanged axial) diffusion of water. Neuroimage 2002;17:1429–36. [20] Thomalla G, Glauche V, Koch MA, Beaulieu C, Weiller C, Rother J. Diffusion tensor imaging detects early Wallerian degeneration of the pyramidal tract after ischemic stroke. Neuroimage 2004;22:1767–74. [21] Song SK, Sun SW, Ju WK, Lin SJ, Cross AH, Neufeld AH. Diffusion tensor imaging detects and differentiates axon and myelin degeneration in mouse optic nerve after retinal ischemia. Neuroimage 2003;20: 1714–22. [22] Hukkanen V, Roytta M. Autolytic changes of human white matter: an electron microscopic and electrophoretic study. Exp Mol Pathol 1987; 46:31–9. [23] Le Bihan D. Diffusion and perfusion magnetic resonance imaging applications to functional MRI. New York: Raven Press; 1995.