3 T magnetic resonance diffusion tensor imaging and fibre trackingPg465 73[PMIDS20451014]

Original Paper3T magnetic resonance diffusion tensor imaging and fibre tracking in cervical myelopathyM.Xiangshui,C.Xiangjun,Z.Xiaoming,Z.Qingshi,C.Yi,Q.Chuanqiang,M.Xiangxing,L.Chuanfu,H.Jinwen *Department of Radiology,Qi Lu Hospital of Shandong University,Chinaarticle information Article history:Received 4July 2009Received in revised form 20January 2010Accepted 27January 2010AIM:To analyse the characterization of diffusion tensor imaging (DTI)with 3T magnetic resonance imaging (MRI)in cervical myelopathy.METHODS:A total of 21healthy controls and 84patients with cervical myelopathy under-went T2-weighted imaging and DTI.The patients were divided into four groups based on the degree of cord compression and MRI signal intensity of the compressed cord as seen on T2-weighted images.The values of apparent diffusion coefficient (ADC),fractional anisotropy (FA),and eigenvalues (l i )were analysed,and fibre tracking (FT)was performed.RESULTS:For healthy controls,the mean values from the DTI of the cervical spinal cord were ADC ¼0.784Æ0.083Â10À3mm 2/s,FA ¼0.721Æ0.027,l 1,l 2,and l 3¼1.509Æ0.145Â10À3,0.416Æ0.094Â10À3,and 0.411Æ0.102Â10À3mm 2/s,respectively.Only values for l 2and l 3differed significantly between the control and A groups (p <0.05).The mean values of l 2and l 3of group A were 0.516Æ0.105Â10À3and 0.525Æ0.129Â10À3mm 2/s,respectively.ADC,FA,l 1,l 2and l 3differed significantly between the control and B,C,D groups (p <0.01).The FT map for group A showed a normal spinal cord,but that for groups B,C,and D showed a distorted spinal cord at the sites of compression.CONCLUSION:The values of ADC,FA,and l i obtained with DTI could assess subtle structural damage and changes of anisotropy in the cord of cervical myelopathy.Fibre tracking was useful in verifying changes in the compressed cord.Ó2010The Royal College of Radiologists.Published by Elsevier Ltd.All rights reserved.IntroductionSpinal cord compression can be caused by variety of diseases,such as metastases,abscess,degeneration,and myelopathy.Among them,cervical myelopathy is the most common and results in cord demyelination,necrosis,and cavitation.Symptoms of cervical myelopathy include motor,sensory deficits and major functional disability.Magnetic resonance imaging (MRI)is an effective imaging method fordiagnosing cervical myelopathy with abnormal T2-weighted images of dura mater spinalis and,occasionally,the cord.T2-weighted imaging alone,however,has low sensitivity for detecting the subtle structural damage of the cord in myelopathy,especially in patients with chronic onset of symptoms.1–3Optimal treatment of cervical myelopathy requires early detection of the abnormality of the spinal cord.In recent years,diffusion-weighted imaging (DWI)sequences have been obtained for spinal cord diseases,with improvement of the diagnostic sensitivity for chronic spinal abnormalities,4,5but few parameters are derived from DWI.Diffusion tensor imaging (DTI)is an MRI technique.The values of apparent diffusion coefficient (ADC),fractional anisotropy (FA)and eigenvalues (l i )obtained with DTI*Guarantor and correspondent:H.Jinwen,Department of Radiology,Qi Lu Hospital of Shandong University,Wenhuaxi Road 107Jinan,China.Tel.:þ86053182169511;fax:þ86053186927544.E-mail address:cxj20080803@ (H.Jinwen).Contents lists available at ScienceDirectClinical Radiologyjournal homepage:www.elsevierhea/journals/crad0009-9260/$–see front matter Ó2010The Royal College of Radiologists.Published by Elsevier Ltd.All rights reserved.doi:10.1016/j.crad.2010.01.019Clinical Radiology 65(2010)465–473evaluate the scalar properties of the diffuse translation of extracellular water molecules within white matterfibres.6 ADC values represent diffusive strength,with high ADC values indicating robust diffusive strength of extracellular water molecules.FA is an anisotropic parameter:values closer to1represent a more anisotropic structure.7The eigenvectors of the symmetric diffusion tensor,a set of orthogonal vectors,defines the orientation of the principal axes of the diffusion tensor ellipsoid in space.Their corre-sponding eigenvalues represent the lengths of the axes. Therefore these quantities define the size and shape of the ellipsoids and are independent of its orientation in its frame of measurement.The degree of anisotropy could be measured by these eigenvalues.The use offibre track(FT) three-dimensional(3D)reconstructions can track the white matter pathways in the brain and spinal cord.8,9 The application of DTI for brain lesions has enabled the characterization of micro-structural changes in multiple sclerosis,schizophrenia,trauma,stroke,and aging,10–12and the lesions andfibre connectivity of the white matter pathways can be visualized using FT.13,14Although DTI is not routinely used in the clinic,a few studies have demon-strated the technique’s potential for the depiction of path-ological conditions of the spinal cord.The aim of the present study was to investigate the characterization of DTI for detecting and quantitatively analysing myelopathological changes in the cervical spinal cord.Materials and methodsFrom September2006to May2007,936patients with cervical radiculopathy or myelopathy were referred to our institution for MRI of the cervical spinal cord.Eighty-four patients(mean age45years,range16–63years;48men) who met the following criteria were enrolled.The criteria were as follows:disk protrusion seen on MRI,clinical symptoms of neck pain,motor dysfunction,and sensory deficits lasting3months,imagingfindings matched the clinical symptoms(compression level).Patients with previous spine surgery or spine radiation therapy or those who had contraindications to MRI were excluded.Twenty-one healthy volunteers(mean age43years,range18–60 years;12men)with no history of neurological disorders were also enrolled as controls.The MRI protocol was approved by the review board of our institution,and all patients and volunteers gave their informed consent for the study.MRI and DTIImaging involved the use of a3T MRI system(GE Medical Systems,EXCITE,Wisconsin,USA)with a maximum gradient strength of40mT/m.A neck coil was applied. The protocol began with the acquisition of a coronal T2-weighted scout view,then a sagittal T2-weighted fast spin echo(FSE)sequence[field of view320Â320mm; image matrix640Â384;section thickness3mm;repeti-tion time/echo time(TR/TE)3000ms/110ms]and a sagittal T1-weighted spin echo sequence(field of view 320Â320mm;image matrix640Â384;section thickness 3mm;TR/TE550ms/9ms).A sequence of FSE,axial,T2-weighted images(field of view180Â180mm;image matrix256Â256;section thickness4mm;TR/TE3000/ 110ms)was obtained at C1–C2,C2–C3,C3–C4,C4–C5,C5–C6, and C6–C7levels of the cervical spinal cord.Subsequently, a sequence of axial sensitivity-encoding echo-planar DTI (field of view270Â270mm;image matrix96Â96;28 sections,4m thick;TR/TE6000/83ms)was obtained at C1–C7.The acquisition time of DTI per participant was5min. Image analysis involved use of the GE Functool software. Before performing tensor estimation,a GE correction procedure was applied to the DTI dataset to correct distor-tions related to eddy currents induced by the large diffusion-sensitizing gradients.Based on a preliminary study,15non-collinear gradient directions with b value of 1000s/mm2were applied.MeasurementsFor healthy volunteers,the values of ADC,FA,eigenvalues l1,l2,and l3were measured at levels C2–C3,C4–C5,and C6–C7of the cervical spinal cord.The regions of interest (ROI)were drawn manually to set the spinal cord at the most accurate axial b0image,which matched the size of transverse spinal cord.Special attention was paid to elimi-nate the partial volume effect of the cerebrospinalfluid (CSF)surrounding the cord,magnetic susceptibility effects, and motion artefacts in the ROI.A reference database was created by pooling the values of ADC,FA,l1,l2,and l3at different levels of the spinal cord for controls.The84patients were divided into four groups according to severity of disk herniation and MRI signal intensity of the compressed cord seen on T2-weighed images,which were reviewed by two experienced neuroradiologists.Differ-ences in diagnoses were resolved by discussion.Group A patients showed only mild compression of the dura mater spinalis,which referred to the disk protruding beyond the posterior vertebral line with obliteration of the subarach-noid space,without cord compression in the spinal cord adjacent to the site of disk herniation.Group B patients showed slight cord compression,which referred to the observation of a concave defect less than30%of the ante-roposterior diameter of the spinal cord at the site of compression,without any change in signal intensity.Group C patients showed moderate cord compression,which referred to the observation of a concave defect in the spinal cord associated with subtle hyperintense signal changes, the concave defects were30–70%of the anteroposterior diameter of spinal cord.Group D patients showed marked cord compression with obvious hyperintense signal changes,the concave defects of the cord at the sites of compression were more than70%of the anteroposterior diameter of spinal cord.DTI measurements were performed at the sites of compression of the spinal cord in all of the groups(groups A,B,C,and D).Values of ADC,FA,l1,l2,and l3were obtained by using small ROIs of10mm2that were matched in size. The ROIs were at the sites of spinal cord compression forM.Xiangshui et al./Clinical Radiology65(2010)465–473 466Figure 1Representative T2-weighted images,FT,and ADC maps,FA,l 1,l 2,and l 3from DTI of the cervical spine cord of a healthy man.(a)Sagittal and (b)axial T2-weighted images show normal cervical cord and dura matter spinalis.(c)FT of the cervical spinal cord and nervi spinales is clearly demonstrated.The nervi spinales appear disconnected from spinal cord due to the small size of radix anterior,radix posterior,and physiological motion artefact.(d)The b0image is similar to T2-weighted image.(e)FA map:the red region of the cervical cord indicates high anisotropy (long arrow),the blue region of cerebrospinal fluid (CSF)surrounding the red region indicates high isotropy (short arrow).(f)ADC map:the central part of the cervical cord indicates low diffusion strength (long arrow)and the yellow region of CSF that surrounds the spinal cord indicates high diffusion strength (short arrow).(g)l 1map:the red region of the cervical cord symbolizes high diffusion strength (arrow).Maps of l 2(h)and l 3(i):the yellow region of the cervical cord symbolizes similar low diffusion strength(arrow).Table 1M.Xiangshui et al./Clinical Radiology 65(2010)465–473467groups A and B,but were at the abnormal hyperintense areas close to the sites of spinal cord compression at the level of disk herniation according to the b0image for groups C and D.Three-dimensional FT maps were reconstructed with the use of b0images,ADC maps,and FA maps at a GE AW 2.0workstation.For controls,ROIs including the cervical spinal cord and nervi spinales were drawn at each level of the cervical spinal cord (C1–C2,C2–C3,C3–C4,C4–C5,C5–C6,C6–C7)to reconstruct the FTs step by step.To visu-alize the distortion of the cervical spinal cord,ROIs were set at the cervical spinal cord only to reconstruct FTs for patients.A FA threshold value of 0.18and an angulation threshold of 45 were adopted to prevent fibres from sudden transition and to keep tracking based on the connectivity of the neighbourhood.The eigen system (with eigenvalues l 1,l 2,and l 3)were calculated voxel by voxel using a previously described method.15Thus,ADC mean ¼(l 1þl 2þl 3)/3¼l andFA ¼ffiffiffi32r $ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiðl 1Àl Þ2þðl 2Àl Þ2þðl 3Àl Þ2q ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffil 21þl 22þl 23q were calculated based on formulas that incorporate thetensor elements to generate quantitative parametric ADC and FA maps in healthy volunteers and patients.Statistical analysisResults are expressed as mean ÆSD.Data analysis involved the use of SPSS 13.0(SPSS,Chicago,IL,USA).The statistical significance of the computed mean values of ADC,FA,l 1,l 2,and l 3of groups A,B,C,and D and the controlswere assessed using analysis of variance (ANOVA)and Newman–Keuls test.P <0.05was considered statistically significant.ResultsConventional T2-weighted imaging showed a normal cervical spinal cord in healthy controls.T2-weighted images and maps of ADC,FA,l 1,l 2,and l 3are shown in Fig.1.The mean values of ADC,FA,l 1,l 2,and l 3of C2–C3,C4–C5,and C6–C7levels are listed in Table 1.ANOVA was used to compare the mean values of ADC,FA,l 1,l 2,and l 3at C2–C3,C4–C5and C6–C7levels in cervical spinal cords of 21healthy volunteers.Values of ADC,FA,l 1,l 2,and l 3did not significantly differ between the three different levels of the spinal cord (Fig.2).The mean values for ADC,l 1,l 2,and l 3at the sites of spinal cord compression for patients are listed in Table 2.ANOVA was used to compare the mean values of ADC,FA,l 1,l 2,and l 3at the sites of compression and healthy spinal cords among groups A,B,C,D,and controls.There were significant differences in the ADC values (F ¼32.84,p <0.05),FA (F ¼115.32,p <0.05),l 1(F ¼7.66,p <0.05),l 2(F ¼63.93,p <0.05),and l 3(F ¼43.57,p <0.05)of groups A,B,C,D,and controls.Multiple comparisons with Newman–Keuls test were used to compare the mean values of ADC,FA,l 1,l 2,and l 3among groups A,B,C,D,and controls (21participants per group).Comparison of the values of ADC,FA,l 1,l 2,l 3between the control group and groups A,B,C,D are listed in Fig.3and Table 3.Group A showed only mild compression of the dura mater spinalis with obliteration of the subarachnoid space,without cord compression on T2-weighted images;l 2and l 3but not ADC.FA and l 1differed significantly between the control group and group A (p <0.05;Table 3).Group B showed concave defects in the spinal cord at the compression sites without high signal intensity on T2-weighted images (Fig.4).Group C showed moderate cord compression with subtle hyperintense signal changes on T2-weighted images (Fig.5).Group D showed marked cord compression with conspicuous hyperintense signal changes on T2-weighted images (Fig.6).Groups B,C,and D signifi-cantly differed from controls in ADC,FA,l 1,l 2,and l 3(p <0.01;Fig.3,Table 3).Healthy controls showed continuous and intact FT of the cervical spinal cord.The main white matter tracts could be seen,and the nervi spinales appeared disconnected from the spinal cord because of the small size of radix anterior,radix posterior,and physiological motion artefacts (Fig.1c).The three-dimensional FT map for group A patientsshowedFigure 2Values of ADC,FA,l 1,l 2,and l 3at levels C2–C3,C4–C5,C6–C7of the spine by DTI in healthy controls.Mean values of ADC,FA,l 1,l 2,and l 3did not significantly differ at C2–C3,C4–C5and C6–C7levels.Table 2M.Xiangshui et al./Clinical Radiology 65(2010)465–473468a normal spinal cord,but that for groups B,C,and D showed a distorted spinal cord at sites of compression,which concurred with findings from the T2-weighted images (Figs.4c,5c,and 6c).DiscussionHolder et al.16and Summers et al.17reported successful in vivo diffusion imaging of the human spinal cord using 1.5and 3T MRI.Ford et al.18demonstrated decreased longitu-dinal ADC and increased transverse ADC in a rat spinal cord weight-drop injury model.DTI of spinal cord diseases,such as syringomyelia,multiple sclerosis,trauma,and tumours,have been reported recently from studies of animals and humans.19–22To the authors’knowledge,use of a 3T MRI system to measure values of ADC,FA,l 1,l 2,and l 3to quantitatively analyse spondylosis-related changes of the cervical spinal cord have not been previously reported.In the present study,values of ADC,FA,l 1,l 2,and l 3at levels C2–C3,C4–C5,and C6–C7of the cervical spinal cord were measured in 21healthy volunteers,and the values of ADC,FA,l 1,l 2,and l 3for chronic spinal cord compression were investigated in 84patients with cervical myelopathy.Values of ADC,FA,l 1,l 2,and l 3did not differ at different levels of the cervical spinal cord in healthy volunteers.Previously reported ADC values (0.96–1.14Â10À3mm 2/s)of normal spinal cord 9,14,23,24are similar to those from the present study (0.781–0.788Â10À3mm 2/s).Most previous investigations of FA of the cervical spine involved 1.5T MRI systems in healthy volunteers,with FA values ranging from 0.52to 0.745,9,14,23,24which concurred with the presentresults (0.718–0.726).According to the present study,values of l 2and l 3were much lower than l 1,whereas l 2values were similar to l 3.Results of Holder et al.16concurred with the present results in that diffusion within the cervical spinal cord was highly anisotropic from these FA values and eigenvalues,which was a diffusion tensor cylinder in space.Thus,the pattern of diffusivity matched the anatomic structure of the spinal cord.In fact,the axons running along the spinal cord create an anisotropic structure,whereby random thermal motion along the axons and the surrounding extracellular space is much less restricted than motion in the perpendicular plane.Ito et al.25and Larocca 26reported that the mechanism of lesion formation in cervical myelopathy increased in pres-sure due to continuous or intermittent pinching of the cord,which induced chronic hypoperfusion.Sequential changes,such as hypoxia or ischaemia,may lead to lesion formation.Demir et al.1reported that DWI in patients with cervical myelopathy revealed an increase in water diffusion and changes in diffusion direction.They hypothesized that changes may be located primarily in the interstitial space,in which water molecules may flow rapidly due to the spatial expansion associated with increased internal pressure.An increase in internal pressure may be the result of an increase in pressure around the spinal cord because of cord pinching by bone formation and disk movement and because of cerebrospinal fluid flow disturbances resulting from the narrowing of the spinal canal.In the present study,the mean ADC values were increased at the sites of compression in groups B,C,and D.The mechanisms that produced these abnormalities could not be determined.Changes of water diffusion in spinal cord compression may be one important reason as explained by Demir et al..1,25,26However,mean ADC values were not increased in group A,which suggests that the structure of spinal cord was not damaged with subtle spinal cord compression.Experimental study of spinal cord compression in rats found decreased FA values due to various elements,such as mechanical disruption,tearing of fibres and myelin sheaths,extracellular oedema,Wallerian degeneration,demyelin-ation,loss of spatial organization,liquefaction,or cystic degeneration.27In this study,spinal cord compression was experimentally designed and imaged several months after the onset of compression.The authors’pathophysiological findings concurred with the present findings of lower FA with spinal compression due to cervical myelopathy than that in healthy controls.In the present study,mean FA values decreased at compression sites in groups B,C,and D.The pathological changes in rats reported by Nevo et al.27Figure 3Comparison of the values of ADC,FA,l 1,l 2,and l 3between the control group and patients of groups A,B,C,and D.l 2and l 3,but not ADC,FA,and l 1,significantly differed between group A and controls (p <0.05);ADC,FA,l 1,l 2,and l 3significantly differed between groups B,C,and D and the controls (p <0.05).Table 3M.Xiangshui et al./Clinical Radiology 65(2010)465–473469Figure 4Representative T2-weighted images,FTs and maps of ADC,FA,l i from the DTI of a patient with cervical myelopathy from group B.(a)Sagittal and (b)axial T2-weighted images show the concave defect which is less than 30%of the anteroposterior diameter of the spinal cord at levels C3–C4,C4–C5and C5–C6,without abnormal signal intensity in the compression sites (arrow).(c)FT shows a concave defect of the spinal cord at the compression site corresponding to the T2-weighted images (arrow).(d)b0image shows distortion at the compression site (arrow).(e)FA map:the fading part of the red region indicates decreased anisotropy at the compression site (arrow).(f)ADC map:the blue region of the cervical cord indicates increased diffusion strength at the compression site (arrow).(g)l 1map:the red region of the cervical cord indicates increased diffusion strength at the compression site.Maps of l 2(h)and l 3(i):the yellow regions of the cervical cord suggest increased diffusion strength at the compression site,respectively.M.Xiangshui et al./Clinical Radiology 65(2010)465–473470may be also have occurred in spinal cord compression in patients with cervical myelopathy,which leads to a decrease in FA at the site of compression of the spinal cord.However,mean FA values did not decrease in group A,which suggests that the microstructure of the spinal cord was not damaged in these patients.In the present study,the values of l 1,l 2,and l 3increased with increasing severity of spinal cord compressioninFigure 5Representative T2-weighted images,FT and the maps of ADC,FA,l i from DTI of a patient with cervical myelopathy from group C.(a)Sagittal and (b)axial T2-weighted images show spinal cord compression at the C5–C6level with subtle hyperintensity at the compression site,the degree of cord compression is nearly 50%of the anteroposterior diameter of the spinal cord (arrow).(c)FT shows distortion of the spinal cord at the compression site corresponding to the T2-weighted image (arrow).(d)b0image shows distortion of the compressed site (arrow).(e)FA map:the part of red region is imperceptible,which suggests decreased anisotrophy at the compression site (arrow).(f)ADC map:the green and slight red region of the cervical cord indicate increased diffusion strength at the compressed site (arrow).Maps of l 1(g),l 2(h),and l 3(i):the dark red signal of the cervical cord at the compression site indicates increased diffusion strength.M.Xiangshui et al./Clinical Radiology 65(2010)465–473471groups B,C,and D.These results indicate that the strength of diffusion increased in different orientations of the principal axes of the diffusion tensor ellipsoid in space,but a cylin-drical diffusion pattern remained at compression sites inthese patients.Values of l 1did not differ between group A and the control group,in which no variation of the corre-sponding eigenvectors in group A occurred.However,values of l 2and l 3were significantly higher in group A than intheFigure 6Representative T2-weighted images,FT and the maps of ADC,FA,l i from DTI of a patient with cervical myelopathy from group D.(a)Sagittal and (b)axial T2-weighted images show spinal cord compression at the C4–C5and C5–C6levels with marked hyperintensity at the compression site,the degree of cord compression is nearly 75%of the anteroposterior diameter of spinal cord (arrow).(c)FT shows distortion of the spinal cord at the compressed site corresponding to the T2-weighted image (arrow).(d)b0image shows distortion of the compressed site (arrow).(e)FA map:the green region of cord suggests significant decreased anisotrophy at the compression site (arrow).(f)ADC map:the green and yellow region of the cord indicate distinguished increased diffusion strength at the compression site (arrow).Maps of l 1(g),l 2(h),and l 3(i):the identical dark red signal of the cervical cord at the compression site indicates marked increased diffusion strength,respectively.M.Xiangshui et al./Clinical Radiology 65(2010)465–473472control group,which suggests that corresponding eigen-vectors increased,although measurement error cannot be discounted.This hypothesis needs further investigation.With regard to FT,this tool appears to be more than just a visualization tool for the white matter tracts of the spinal cord.Anatomical correlation to corticospinal and spino-thalamic pathways is excellent.FT measurement can help confirm the positive diagnosis of spinal cord abnormalities by locating the specific compression area.14In the present study,the main white matter tracts could be clearly demonstrated on FT maps in all healthy controls.In patients,the FT map showed distortion and interruption of the compressed spinal cord,which matchedfindings seen on T2-weighted images.Interruptedfibres on FT maps corresponded to FA values below0.18or a rupture angle of 45 ,and this aspect did not mean thatfibres were actually interrupted.We believe that the other appearance with spreadingfibres corresponded to oedema and gliosis of the cord at the compression sites,but there is no objective proof.This hypothesis has to be explored by further studies.The quality of DTI of the cervical spinal cord may be decreased due to the small diameter of the spinal cord,CSF flow,respiratory,cardiac motion artefacts,and magnetic susceptibility artefacts.This technique still needs further improvement.The image and motion artefacts may be improved by use of a better co-registration algorithm and more gradient directions in DTI sequences.28In the prelimi-nary study(performed infive healthy volunteers,with5,10, 15,20,25non-collinear gradient directions selected,and b values set at600,1000,1500,2000,and2500s/mm2, respectively),15non-collinear directions and a b value of 1000s/mm2were helpful in improving the quality of DTI of the spinal cord.It was important to eliminate the partial volume effect of CSF surrounding the cord,magnetic suscep-tibility effects,and motion artefacts in the ROI selection.In conclusion,values of ADC and FA measured by DTI could assess spinal cord abnormalities in patients with cervical myelopathy.The subtle structural damage of the cord in cervical myelopathy could be quantitatively ana-lysed in the early stages of the disease by measuring the values of ADC,FA,l1,l2,and l3.The size and shape of ellipsoids in space of diffusion tensor at compressed spinal cord could be evaluated by the values of l1,l2,and l3.FT was useful in revealing the main white matter tracts of the spinal cord in assessing chronic cord compression. AcknowledgementsThe authors thank Zheng Jinyong,Cong Peixin,Kang Xiaoshui,and Feng Dechao with their assistance in imaging the subjects.References1.Demir A,Ries M,Moonen C,et al.Diffusion-weighted MR imaging withapparent diffusion coefficient and apparent diffusion tensor maps in cervical spondylotic myelopathy.Radiology2003;229:37–43.2.Valsasina P,Agosta F,Benedetti B,et al.Diffusion anisotropy of thecervical cord is strictly associated with disability in amyotrophic lateral sclerosis.J Neurol Neurosurg Psychiatr2007;78:480–4.3.Takahashi M,Yamashita Y,Sakamoto Y,et al.Chronic cervical cordcompression:clinical significance of increased signal intensity on MR images.Radiology1989;173:219–24.4.Tsichiya K,Katase S,Fujikawa A,et al.Diffusion-weighted MRI of thecervical spinal cord using a single-shot fast spin-echo technique:findings in normal subjects and in myelomalacia.Neuroradiology2003;45:90–4.5.Plank C,Koller A,Mueller-Mang C,et al.Diffusion-weighted MR imaging(DWI)in the evaluation of epidural spinal lesions.Neuroradiology 2007;49:977–85.6.Basser P,Mattiello J,LeBihan D.MR diffusion tensor spectroscopy andimaging.Biophys J1994;66:259–67.7.Toosy AT,Werring DJ,Orrell RW,et al.Diffusion tensor imaging detectscorticospinal tract involvement at multiple levels in amyotrophic lateral sclerosis.J Neurol Neurosurg Psychiatr2003;74:1250–7.8.Conturo TE,Lori NF,Cull TS,et al.Tracking neuronalfiber pathways inthe living human brain.Proc Natl Acad Sci USA1999;96:10422–7.9.Renoux J,Facon D,Fillard P,et al.MR diffusion tensor imaging andfibertracking in inflammatory diseases of the spinal cord.AJNR Am J Neuro-radiol2006;27:1947–51.10.Andrade RE,Gasparetto EL,Cruz Jr LC,et al.Evaluation of white matterin patients with multiple sclerosis through diffusion tensor magnetic resonance imaging.Arq Neuropsiquiatr2007;65:561–4.11.Tang CY,Friedman J,Shungu D,et al.Correlations between diffusion tensorimaging(DTI)and magnetic resonance spectroscopy(1H MRS)in schizo-phrenic patients and normal controls.BMC Psychiatry2007;19(7):25. 12.Guadano JV,Jones PS,Fryer TD,et al.Local relationships betweenrestricted water diffusion and oxygen consumption in the ischemic human brain.Stroke2006;37:1741–8.13.Jones DK,Simmons A,Williams SC,et al.Non-invasive assessment ofaxonalfiber connectivity in the human brain via diffusion tensor MRI.Magn Reson Med1999;42:37–41.14.Facon D,Ozanne A,Fillard P,et al.MR diffusion tensor imaging andfibertracking in spinal cord compression.AJNR Am J Neuroradiol 2005;26:1587–94.15.Asser P,Mattiello J,LeBihan D.Estimation of the effective self-diffusiontensor from the NMR spin echo.J Magn Reson B1994;103:247–54. 16.Holder CA,Muthupillai R,Mukundan Jr S,et al.Diffusion-weighted MRimaging of the normal human spinal cord in vivo.AJNR Am J Neuroradiol 2000;21:1799–806.17.Summers P,Staempfli P,Jaermann T,et al.A preliminary study of theeffects of trigger timing on diffusion tensor imaging of the human spinal cord.AJNR Am J Neuroradiol2006;27:1952–61.18.Ford JC,Hackney DB,Alsop DC,et al.MRI characterization of diffusion coef-ficients in a rat spinal cord injury model.Magn Reson Med1994;31:488–94.19.Fujiyoshi K,Yamada M,Nakamura M,et al.In vivo tracing of neural tractsin the intact and injured spinal cord of marmosets by diffusion tensor tractography.J Neurosci2007;27:11991–8.20.Ducreux D,Lepeintre JF,Fillard P,et al.MR diffusion tensor imaging andfibertracking in5spinal cord astrocytomas.AJNR Am J Neuroradiol2006;27:214–6.21.Inglis BA,Neubauer D,Yang L,et al.Diffusion tensor MR imaging andcomparative histology of glioma engrafted in the rat spinal cord.AJNR Am J Neuroradiol1999;20:713–6.22.Agosta F,Rovaris M,Benedetti B,et al.Diffusion tensor MRI of thecervical cord in a patient with syringomyelia and multiple sclerosis.J Neurol Neurosurg Psychiatr2004;75:1647.23.Wheeler-Kingshott C,Hickman S,Parker G,et al.Investigating cervicalspinal cord structure using axial diffusion tensor imaging.Neuroimage 2002;16:93–102.24.Cerciqnani M,Horsfield M,Agosta F,et al.Sensitivity-encoded diffusion tensorMR imaging of the cervical cord.AJNR Am J Neuroradiol2003;24:1254–6. 25.Ito T,Oyanagi K,Takahashi H,et al.Cervical spondylotic myelopathy.Clinicopathologic study on the progression pattern and thin myelinated fibers of the lesions of seven patients examined during complete autopsy.Spine1996;21:827–33.rocca H.Cervical spondylotic myelopathy:natural history.Spine1988;13:854–5.27.Nevo U,Hauben E,Yoles E,et al.Diffusion anisotropy MRI for quantitativeassessment of recovery in injured rat spinal cord.Magn Reson Med 2001;45:1–9.28.Jones DK.The effect of gradient sampling schemes on measures derived fromdiffusion tensor MRI:a Monte Carlo study.Magn Reson Med2004;51:807–15.M.Xiangshui et al./Clinical Radiology65(2010)465–473473。