magnetic resonance imaging of diffuse axonal injury: quantitative

JOURNAL OF NEUROTRAUMAVolume 24, Number 4, 2007© Mary Ann Liebert, Inc.Pp. 591–598DOI: 10.1089/neu.2006.0214

Magnetic Resonance Imaging of Diffuse Axonal Injury:Quantitative Assessment of White Matter Lesion Volume

CARLOS MARQUEZ DE LA PLATA,1 ANDREEA ARDELEAN,4 DELLA KOOVAKKATTU,4PRIYA SRINIVASAN,4 ANNA MILLER,1 VIET PHUONG,1 CARYN HARPER,1CAROL MOORE,1 ANTHONY WHITTEMORE,3 CHRISTOPHER MADDEN,2

RAMON DIAZ-ARRASTIA,1 and MICHAEL DEVOUS, Sr.3,4

ABSTRACT

Diffuse axonal injury (DAI) is a common mechanism of traumatic brain injury (TBI) for whichthere is no well-accepted anatomic measures of injury severity. The present study aims to quanti-tatively assess DAI by measuring white matter lesion volume visible in fluid-attenuated inversionrecovery (FLAIR) weighted images and to determine whether higher lesion volumes are associatedwith unfavorable functional outcome 6 months after injury. Twenty-four patients who experiencedmoderate to severe TBI without extra-axial or major cortical contusions were included in this study.Lesion volume was assessed by quantifying areas of hyperintensities in the white matter utilizingdigitized FLAIR images. Two independent raters processed the magnetic resonance (MR) imagesand determined the total DAI volume. Functional outcome was assessed at 6 months after injuryusing the Glasgow Outcome Scale–Extended (GOSE). Interclass correlation analyses showed veryhigh interrater reliability for each measure between the two raters (Interclass Correlation Coeffi-cient � 0.95, p � 0.001). Total DAI volume was significantly, although modestly, correlated to GOSE(r � �0.453, p � 0.034). White matter lesion volume resulting from DAI can be quantitatively andreliably assessed from standard FLAIR-weighted MRIs. Patients with greater DAI volume havepoorer functional outcomes. These methods may be useful in stratifying injury severity and for theassessment of DAI-directed therapies.

Key words: diffuse axonal injury (DAI), FLAIR MRI, functional outcome, TBI

591

Departments of 1Neurology, 2Neurological Surgery, and 3Radiology, University of Texas Southwestern Medical Center, and4Department of Brain and Behavioral Sciences, University of Texas at Dallas, Dallas, Texas.

INTRODUCTION

DIFFUSE AXONAL INJURY (DAI) is characterized by mi-croscopic axonal lesions that commonly appear in

subcortical white matter of patients with acceleration-de-celeration type traumatic brain injuries (TBI) (Adams et

al., 1989). DAI lesions, or shear-injuries, occur whenbrain tissue of relatively lighter consistency is torn frommore rigid tissue when exposed to linear or rotational de-celeration forces (Gentry, 2001; Hammoud and Wasser-man, 2002). The most common locations of these injuries,as determined by pathologic studies, are the centroaxial

white matter regions, including the corpus callosum, in-ternal capsule, and fornix due to their rigid attachment torelatively more mobile cerebral hemispheres (Gentry etal., 1988a; Cotran et al., 1999).

Moderate to severe TBI patients with DAI often havecomputed tomography (CT) images absent of significanthematomas or hemorrhages, but their neurologic status is far more serious than expected. CT scans are useful mea-sures of injury severity for patients with extra-axialhematomas and parenchymal contusions (Marshall et al.,1991), but in DAI CT is usually normal or reveals onlysmall deep (shear) hemorrhages. Magnetic resonance(MR) imaging is recommended in these situations, as MR is significantly more sensitive to axonal shear injuriesthan CT (Gentry et al., 1988b; Ogawa et al., 1992). How-ever, the usefulness of MRI as a measure of injury sever-ity has not been quantitatively established. Fluid-attenu-ated inversion recovery (FLAIR) imaging has significantlyimproved the ability of imaging to identify DAI lesions,as it greatly reduces the signal from cerebral spinal fluidthereby increasing specificity for injured tissue. The le-sions appear as white matter hyperintensities (WMH) andare subject to either qualitative or quantitative analysis.

Assessment of injury severity in patients with DAI le-sions is important for guiding clinical management, assess-ing prognosis and counseling families, as well as for strat-ifying injury severity when selecting patients for clinicaltrials. Additionally, quantitative assessments of DAI maybe useful as surrogate endpoints for clinical trials directedat white matter injury. Quantitative MRI techniques can beused to compare the extent of tissue damage in differentbrain regions, as location of such injuries may be correlatedwith resultant cognitive impairment and functional out-come. Quantitative MRI assessments have identified atro-phy of various neuroantomic structures chronically afterTBI (Gale et al., 1995; Bigler et al., 1996, 2002; Levin etal., 2000; Yount et al., 2002). However, the correlation be-tween outcome and a volume decrease has been equivocal(Levin et al., 1992; Johnson et al., 1996; Bigler et al., 1997,1999; Bigler, 2001; Yount et al., 2002; Himanen et al., 2005;Gale et al., 1995; Pierallini et al., 2000; Scheid et al., 2003;King et al., 2005; Tomaiuolo et al., 2005; Scheid et al.,2006). Further, few studies have analyzed MRI-visible le-sions (WMH) in the acute setting when prognostic infor-mation is most desired and when DAI-directed neuropro-tective therapies are more likely to be useful.

Most studies assess DAI by counting the number oflesions. Novel computerized image processing tech-niques allow quantitative measures of lesion volume bysumming pixels with signal above a certain threshold.Babikian et al. (2005) utilized a semiautomatic imageprocessing tool to quantify lesion volumes noted on sus-ceptibility-weighted/MR images (SWI-MRI) in pediatric

patients with acute TBI and found lesion volume to benegatively correlated with neuropsychologic functioning.This study found that lesion volume accounts for nearlya third of the variance in cognitive performance on neu-ropsychologic tests. SWI-MRI detects microhemorrhagesand is, at best, an indirect measure of white matter pathol-ogy. Lesions apparent on FLAIR-weighted images rep-resent areas of edema and/or gliosis, and in the setting ofcerebral infarction (Fu et al., 2005; Ouhlous et al., 2005),demyelinating disease (Randolph et al., 2005; Sanfilipoet al., 2006), and neurodegenerative dementia (Gootjeset al., 2004) are good indicators of disease severity. Fewstudies have analyzed FLAIR lesions in TBI (Pieralliniet al., 2000; Scheid, 2006), and we are unaware of anyattempts to examine the interrelationship between DAIlesion volumes visible on acute FLAIR-weighted MRIand functional outcome.

The purpose of the present study is to quantitate DAIlesion volumes using a novel image processing tool todetermine whether it is useful for predicting functionaloutcomes in adult patients with moderate to severe TBI.Additionally, we wanted to determine whether lesions inspecific white matter regions were better correlated withoutcome than those in other regions.

METHODS

Participants

The study sample was selected from patients who suf-fered moderate and severe TBI between January 2004and December 2005. Each subject had been enrolled ina prospective, observational study and evaluated usingMR imaging in the first 2 weeks after injury. Participantswere 14 years of age or older, suffered a brain injury con-sistent with the mechanism of DAI that required hospi-talization, and consented to participate. Patients with sub-dural hematomas and large cortical contusions wereexcluded from the study. Additionally, patients with con-ditions that may result in abnormal MRI and compromisecognitive functions (i.e., prior TBI intracranial tumor,stroke, epilepsy, multiple sclerosis, arteriovenous mal-formation, human immunodeficiency virus [HIV] en-cephalopathy, brain abscesses, Alzheimer’s disease,Parkinson’s disease, and meningitis) were also excluded.Prisoners and homeless patients were also excluded.

Approval for this study was obtained from the Institu-tional Review Board (IRB) of UT Southwestern MedicalCenter prior to its initiation.

Magnetic Resonance Imaging

MRI data were collected with a GE Genesis Signa 1.5-Tesla scanner (General Electric Medical, Milwaukee,

MARQUEZ DE LA PLATA ET AL.

592

WI). FLAIR images (8800/2200/130 [repetition time/in-version time/echo time]) were acquired in the axial planewith 5-mm slice thickness and interslice gap of 0.5 mm.The field of view (FOV) was 220 mm, and matrix sizewas 256*256. FLAIR sequence allows better differenti-ation of brain lesions by suppressing the effect of cere-brospinal fluid (CSF), which normally appears as high-intensity signal in the conventional T2-weighted images.

Image Processing and Diffuse Axonal ImagingLesion Measurement

Images were converted from DICOM to ANALYZEformat to facilitate analysis (Fig. 1). They were then ana-lyzed via a semi-automatic interactive software tool that

was developed in our lab on MATLAB (MathWorks, Inc.,Natick, MA). This tool uses signal thresholding, morpho-logic erosion, connectivity, and dilation methods in com-bination with human knowledge to segment and quantifyWMH and whole brain volume (WBV) from FLAIR im-ages. Analyses are divided into the following operations:

1. Preprocessing: Removal of skull and connective tis-sues using the Brain Extraction Tool (BET) functionin MRIcro software (Rorden and Brett, 2000)

2. Segmentation: Brain tissue and/or WMH were ex-tracted by determining a threshold that included eitherall brain tissue (WBV) or WMH. The tool first high-lights voxels on the FLAIR images that are above aninitial threshold intensity. The initial threshold value

MRI OF DIFFUSE AXONAL INJURY

593

FIG. 1. Unprocessed and segmented FLAIR magnetic resonance image. (A) Unprocessed axial fluid-attenuated inversion-recovery (FLAIR) magnetic resonance (MR) image of a 39-year-old male taken 3 days post–traumatic brain injury (TBI). (B) Seg-mented image of aforementioned patient with areas of hyperintense signal in the corpus collosum, basal ganglia region, and dif-fuse white matter identified by a semiautomated threshold procedure in MATLAB. (C) Unprocessed FLAIR MR image of a 19-year-old female taken 2 days after head injury. (D) Segmented image indicating areas of hyperintense signal in the corpus collosum.

is established for each subject using statistical para-meters (image mean intensity, standard deviation, andmaximum intensity) derived from each subject’sFLAIR images. The tool then provides an option toadjust the threshold intensity to include all hyperin-tense regions or brain tissues in the FLAIR imagesbased on visual inspection.

3. Editing: Signal intensity in some normal brain tissuesoverlaps with the intensity range of hyperintense tissue.Thus, non-WMH tissues as well as WMH might behighlighted by the tool for a chosen threshold. There-fore, WMH from the highlighted areas are identified byvisually comparing to the original FLAIR image. Usersmay then trace a region of interest encompassing non-WMH tissues by clicking points in the image on eachimage slice. This action is repeated as many times asneeded to include all non-WMH tissues. Once selectionis completed, a mask is created that filters out the se-lected non-WMH regions leaving behind only WMH.

4. Classification and Quantification: Voxel intensitieshigher than the threshold are classified as WMHor/and brain tissue as needed. The DAI volumes(DAIV; the volume of the WMH) and WBV were ob-tained by automatically counting the number of vox-els in the extracted regions and multiplying that num-ber by voxel size. Furthermore, the software toolautomatically stores the DAIV, WBV, and the DAIindex (DAIV expressed as percent of WBV) in an Ex-cel file for further analysis.

After a brief orientation to the segmentation and quan-titation procedures, two independent raters followed theprotocol to obtain DAIV and WBV, as well as the ratioof DAI volume to whole brain volume (DAI index). Re-sulting values were analyzed to determine inter-rater re-liability for the manualized procedure.

DAIs in the subcortex were separated into four neu-roanatomical regions: corpus collosum, fornix, internalcapsule/external capsule, and other subcortical whitematter. The corpus collosum was partitioned axially asthe structures anterior to and posterior to the lateral ven-tricles, and the medial structures superior to the lateralventricles. The fornix was identified as the structure me-dial and inferior to the lateral ventricles. The basal gan-glia region included lesions believed to encompass bilat-eral medial white matter structures in and around thebasal ganglia such as the internal and external capsules,superior and anterior thalamic radiations, and uncinatefasciculus. The fourth category of DAI lesions was de-fined as any lesion in subcortical white matter that wasnot categorized in any of the three aforementionedgroups, which include superior longitudinal fasciculus,superior region of corona radiate, and cingulum.

Injury Severity Measures

The Glasgow Coma Scale (GCS) score was obtainedfrom the Emergency Department admission forms, andrepresents the first GCS obtained after stabilization ofcardiopulmonary function. Scores on the GCS range from3 to 15, with higher scores representing a less severe in-jury. GCS scores can be subdivided into three categorieswith respect to injury severity with scores of 13–15 sig-nifying a “mild” injury, 9–12 signifying a “moderate” in-jury, and 3–8 signifying a “severe” injury.

Functional Outcome Measure

Functional outcome was determined 6 months post-in-jury using the Glasgow Outcome Scale–Extended (GOSE;Wilson et al., 1998), a commonly used questionnaire thatassesses functional abilities in multiple domains followinga head injury. The eight-item test includes questions re-garding the patients ability to follow commands, performactivities of daily living, work, travel, participate in recre-ational activities, and inquires about the presence of emo-tional disruptions and of seizures that have begun since theinjury. Total GOSE scores range from one to eight, withhigher scores associated with better outcome. Outcomescores are divided into disability categories such that in-dividuals who score at or below four are considered “se-verely disabled,” scores of five or six are “moderately dis-abled,” and scores of seven or eight are considered “goodrecovery.” A score of one is given to individuals who aredead at the time the questionnaire would have been ad-ministered, and a score of two is given to patients in a “per-sistent vegetative state.”

Statistical Analysis

Intraclass Correlation Coefficients (ICC) were utilizedto determine the inter-rater reliability among the derivedDAI measures (DAIV, WBV, and DAI index) as obtainedby the two independent quantitation operators. The meanvalues for the derived DAI measures were utilized in sub-sequent analyses. Spearman’s rank-order correlation co-efficient was used to assess relationships among the de-rived DAI measures and functional outcome scores andother ordered categorical variables.

RESULTS

Sample Characteristics

Twenty-four patients were identified for this analysis.Patients who were enrolled in this study had a mean ageof 28 (�13) years and were predominantly male (66%).These patients had an average GCS score of 5 (�4) andan average hospital stay of 20 (�8) days. CT scans for


594

all patients were rated as “Diffuse Injury II” using theMarshall TCBD classification (Marshall et al., 1991).Sample characteristics are summarized on Table 1.

Inter-Rater Reliability

Inter-rater reliability analysis for the three mean DAIvalues showed significant consistency between raters(DAIV—ICC � 0.968, p � 0.001; WBV—ICC �0.914, p � 0.001; and DAI index—ICC � 0.956, p �0.001).

Correlations

Volumetric values were analyzed to determine whetherDAI measurements are sensitive to threshold effects withrespect to outcome. Analysis of variance showed only

DAI index was significantly greater for patients who fellin the severely impaired or persistent vegetative staterange than patients who had better outcomes (F � 4.65,1; p � 0.043). DAI volume showed a similar trend, butit was not statistically significant (F � 3.76; 1; p �0.067; Table 2).

Spearman correlations demonstrated that both totalDAI volume and a DAI index expressed as the ratio ofDAI volume to whole brain volume were significantlyinversely correlated to GOSE (r � 0.453, p � 0.034; r �0.485, p � 0.022, respectively), such that greateramounts of DAI coincide with lower functional outcomescores (Table 3).

Additionally, GOSE scores were significantly worsefor patients with a greater DAI index in the internal cap-sule/external capsule region (r � �0.484, p � 0.022).


595

TABLE 1. SAMPLE CHARACTERISTICS (N � 24)

Mean SD Median Range

Age, years 28.17 13.32 22 15–63�20 (n, %) 8 33%.021–40 12 50%.041–60 3 13%.060� 1 4%.0

Male gender 16 66%.0GCS 5.13 4.15 3 3–15

Mild, 13–15 3 13%.0Moderate, 9–12 1 4%.0Severe, �8 20 83%.0

GOSE 4.86 2.19 4.5 1–8GOSE 1–2 3 13.6%GOSE 3–4 8 36.3%GOSE 5–6 5 22.7%GOSE 7–8 6 27.2%

Descriptive statistics for the sample.GCS, Glasgow Coma Scale; GOSE, Glasgow Outcome Scale–Extended; FSE, Functional Status Examination.

TABLE 2. DAI VOLUMES BY GOSE GROUPS

DAIV WBV DAII

Mean SD Mean SD Mean SD

GOSE (all subjects) 10.46 8.35 1246.92 118.40 0.84 0.68PVT or death (1–2) 16.96 12.75 1269.28 49.23 1.36 1.02Severe (3–4) 12.95 8.94 1211.04 126.03 1.08 0.77Moderate (5–6) 7.73 6.52 1298.54 62.52 0.58 0.48Good (7–8) 7.02 6.67 1217.56 166.71 0.55 0.44

No significant differences in volume of hyperintense signal on fluid-attenuated inversion recovery (FLAIR) magnetic resonanceimaging (MRI) across utcome groups. Volume measured in cubic centimeters.

DAIV, diffuse axonal injury volume; WBV, whole brain volume; DAII, diffuse axonal injury index; GOSE, Glasgow OutcomeScale–Extended; PVT, persistent vegetative state.

Though the correlations between many of the other brainregions were not significantly correlated with functionaloutcomes, DAI in the fornix and the unspecified whitematter structures trended toward significance (Table 4).

DISCUSSION

DAIs can be visualized indirectly through shear hem-orrhages (traumatic microbleeds) caused by inertial in-juries on blood vessels (Tong et al., 2004; Scheid et al.,2003, 2006) or more directly by analyzing white matterhyperintensities on FLAIR MRI (Takaoka et al., 2002;Pierallini et al., 2000). To our knowledge, the presentstudy is the first attempt to quantify DAI volume inFLAIR-weighted images with moderately to severely in-jured adults in the acute phase after TBI. Functional out-come is inconsistently correlated to number of shear he-morrhages, with many studies suggesting that whilepatients with traumatic microbleeds have neurocognitivedeficits, there is no correlation between number of trau-matic microbleeds and outcome (Scheid et al., 2003,2006). In the pediatric TBI literature, volume of micro-hemorrhages detected using susceptibility-weighted MRIcorrelates highly with injury severity and poor outcomeat 6 months post-head injury (Babikian et al., 2005; Tonget al., 2004). In our study, white matter DAI volumes inadult patients with moderate to severe TBI are correlatedwith functional outcomes at 6 months after head injury.We found that the greater the proportion of brain volumeaffected by DAI, the poorer their functioning at 6 monthspost-injury as measured by the GOSE.

The quantitation of DAI lesions in specific brainregions during the acute phase of TBI may have prog-nostic value. Moderately to severely head injured indi-viduals with a greater proportion of the internal cap-sule/external capsule region impacted by DAI have aworse outcome than those with lower lesion volume inthat region. However, DAI in the fornices and undiffer-entiated subcortical white matter trended toward signifi-

cance when correlated to outcome and may show signif-icance with a greater sample size. Therefore, while theinternal capsule/external capsule region is the only sub-cortical brain area that showed a statistically significantcorrelation with outcome, it is unlikely that this is theonly subcortical region correlated with functional out-come. Our results are consistent with the notion that le-sions in any white matter region are potentially impor-tant in producing a poor outcome. The modestcorrelations between white matter regions and outcomemay be a reflection of the diffuse nature of DAI(Tomaiuolo et al., 2005; Pierallini et al., 2000).

In our present sample of moderate to severe patientswith TBI, the predominant location of DAI lesions andgreatest DAI volume was the subcortical gray-white mat-ter region. This is consistent with previous studies thathave found less DAI lesion volume in deeper subcorticalbrain regions (Schaefer et al., 2004; Tong et al., 2004),and speaks to the vulnerability of the more superior braintissue to shear injury.

A novel feature of this study is that all of the MRIscans were conducted within 2 weeks of injury, as otherauthors who have reported atrophy in white and gray mat-ter studied patients months or years after injury (Pieralliniet al., 2000; MacKenzie et al., 2002; Scheid et al., 2003,2006; Tomaiuolo et al., 2004, 2005). It is possible thatthe acute white matter lesions in our patients may resultin delayed cortical atrophy via Wallerian degenerationpost-DAI, as has been noted in prior studies (Bigler etal., 2002; MacKenzie et al., 2002; Tomaiuolo et al.,2004). The mean volume of brain parynchema in this in-vestigation is commensurate with that reported amongnormal control subjects in MacKenzie et al. (2002), sug-gesting cortical atrophy had not yet occurred within thefirst week after injury.

The present results indicate that semi-automated im-age processing tools can be useful in quantitating white


596

TABLE 3. CORRELATIONS

AMONG DAI VOLUMES AND GOSE

GOSE p

DAIV �0.453 0.34WBV �0.191 0.395DAI Index �0.485 0.022

Correlations among volume measures and outcomes.Significant correlations in bold.DAIV, diffuse axonal injury volume; WBV, whole brain

volume; GOSE, Glasgow Outcome Scale–Extended.

TABLE 4. LOCATION OF DAI CORRELATED

WITH FUNCTIONAL OUTCOMES

GOSE p

DAII IC/EC �0.484 0.022DAII CC �0.335 0.127DAII Fnx �0.371 0.089DAI other subcortical �0.413 0.056

Correlation between diffuse axonal injury (DAI) volume andoutcome by location of injury.

Significant correlations in bold.DAII, diffuse axonal injury index; IC/EC, internal

capsule/external capsule; CC, corpus callosum; Fnx, Fornix;GOSE, Glasgow Outcome Scale–Extended.

matter lesions in patients with TBI. As MRI makes DAIlesions easier to identify, volumetric analysis of this typeof injury may be a useful method for taking structuralimaging beyond diagnosis. Using axial FLAIR MR im-ages, our newly developed image processing tool reliablyquantified previously identified white matter lesions, asindependent raters attained similar results after a brief in-troduction to the procedure. The inter-rater reliability forall volumetric measurements in the present study are ashigh or higher than those of prior investigations quanti-tating the number of microbleeds (Scheid et al., 2003).

In conclusion, our results demonstrate that volumetricassessment of DAI lesions in adult patients with TBI dur-ing the acute phase of the illness is reliable, is repro-ducible, and is associated, albeit modestly, with func-tional outcome. Such quantitative measures may beuseful in stratifying injury severity in patients enrolled inclinical trials, as well as surrogate outcome measures forstudies of neuroprotective therapies that target DAI.

ACKNOWLEDGMENTS

This work was supported by the NIH-NICDH (grantsR01 HD48179, U01 HD42652) and the U.S. Departmentof Education (grant H133A020526).

REFERENCES

ADAMS, J., DOYLE, D., FORD, I., GENNARELLI, T., GRA-HAM, D., and MCLELLAN, D. (1989). Diffuse axonal in-jury in head injury: definition, diagnosis, and grading.Histopathology 15, 49–59.

BABIKIAN, T., FREIER, M.C., TONG, K.A., et al. (2005).Susceptibility weighted imaging: Neuropsychologic outcomeand pediatric head injury. Pediatr. Neurol. 33, 184–194.

BIGLER, E.D. (2001). Quantitative magnetic resonance imag-ing in traumatic brain injury. J. Head Trauma Rehabil. 16,117–134.

BIGLER, E.D., BLATTER, D.D., GALE, S.D., et al. (1996).Traumatic brain injury and memory: The role of hippocam-pal atrophy. Neuropsychology 3, 333–342.

BIGLER, E.D., BLATTER, D.D., ANDERSON, C.V., et al.(1997). Hippocampal volume in normal aging and traumaticbrain injury. Am. J. Neuroradiol. 18, 11–23.

BIGLER, E.D., JOHNSON, S.C., and BLATTER, D.D. (1999).Head trauma and intellectual status: relation to quantitativemagnetic imaging findings. Appl. Neuropsychol. 6, 217–225.

BIGLER, E.D., ANDERSON, C.V., and BLATTER, D.D.(2002). Temporal lobe morphology in normal aging and trau-matic brain injury. Am. J. Neuroradiol. 23, 255–266.

COTRAN, R., KUMAR, V., and COLLINS, T. (eds). (1999).Trauma to the brain, in: Robbins Pathologic Basis of Dis-ease, 6th ed. WB Saunders: Philadelphia, pps. 1301–1304.

FU, J., LU, C., HONG, Z., DONG, Q., LUO, Y., and WONG,

K. (2005). Extent of white matter lesions is related to acutesubcortical infarcts and predicts further stroke risk in patientswith first ever ischaemic stroke. J. Neurol. Neurosurg. Psy-chiatry 76, 793–796.

GALE, S.D., JOHNSON, S.C., BIGLER, E.D., and BLATTER,D.D. (1995). Nonspecific white matter degeneration follow-ing traumatic brain injury. J. Int. Neuropsychol. Soc. 1,17–28.

GENTRY, L.R. (2001). Head trauma, in: Magnetic ResonanceImaging of the Brain and Spine, 3rd ed. S. Atlas (ed), Lippin-cott, Williams and Wilkins: Philadelphia, pps. 1069–1057.

GENTRY, L.R., THOMPSON, B., and GODERSKY, J.C.(1988a). Trauma to the corpus callosum: MR features. Am.J. Neuroradiol. 9, 1129–1138.

GENTRY, L.R., GODERSKY, J.C., and THOMPSON, B.(1988b). MR imaging of head trauma: review of the distrib-ution and radiopathologic features of traumatic lesions. Am.J. Roentgenol. 150, 663–672.

GOOTJES, L., TEIPEL., S., ZEBUHR, Y., et al. (2004). Re-gional distribution of white matter hyperintensities in vascu-lar dementia, Alzheimer’s disease and healthy aging. De-ment. Geriatr. Cogn. Disord. 18, 180–188.

HAMMOUD, D.A., and WASSERMAN, B.A. (2002). Diffuseaxonal injuries: pathophysiology and imaging. Neuroimag.Clin. North Am. 12, 205–216.

HIMANEN, L., PORTIN, R., ISONIEMI, H., et al. (2005).Cognitive functions in relation to MRI findings 30 years af-ter traumatic brain injury. Brain Inj. 19, 93–100.

JOHNSON, S.C., PINKSTON, J.B., BIGLER, E.D., andBLATTER, D.D. (1996). Corpus callosum morphology innormal controls and TBI: sex difference, mechanisms of in-jury, and neuropsychological correlates. Neuropsychology10, 408–415.

KING, J.T., CARLIER, P.M., and MARION, D.W. (2005).Early Glasgow Outcome Scale scores predict long-term func-tional outcome in patients with severe traumatic brain injury.J. Neurotrauma 22, 947–954.

LEVIN, H.S., WILLIAMS, D.H., EISENBERG, H.M., et al.(1992). Serial MRI and neurobehavioural findings after mildto moderate closed head injury. J. Neurol. Neurosurg. Psy-chiatry 55, 255–262.

LEVIN, H.S., BENAVIDEZ, D.A., VERGER-MAESTRE, K.,et al. (2000). Reduction of corpus callosum growth after se-vere traumatic brain injury in children. Neurology 54,647–653.

MACKENZIE, J.D., SIDDIQI, F., BABB, J.S., et al. (2002).Brain atrophy in mild or moderate traumatic brain injury: alongitudinal quantitative analysis. Am. J. Neuroradiol. 23,1509–1515.

MARSHALL, L.F., BOWERS-MARSHALL, S., KLUBER,M.R., et al. (1991). A new classification of head injury basedon computerized tomography. J. Neurosurg. 75, S14–S20.

OGAWA, T., SEKINO, H., UZURA, M., et al. (1992). Com-parative study of mmagnetc resonance and CT scan in casesof severe head injury. Acta Neurochir. (Suppl.) 55, 8–10.

OUHLOUS, M., FLACH, H., DE WEERT, T., et al. (2005).Carotid plaque composition and cerebral infarction: MRimaging study. Am. J. Neuroradiol. 26, 1044–1049.


597

PIERALLINI, A., PANTANO, P., FANTOZZI, L.M., et al.(2000). Correlation between MRI findings and long-termoutcome in patients with severe brain trauma. Neuroradiol-ogy, 42, 860–867.

RANDOLPH, J., WISHART, H., SAYKIN., et al. (2005).FLAIR lesion volume in multiple sclerosis: relation to pro-cessing speed and verbal memory. J. Int. Neuropsychol. Soc.11, 205–209.

RORDEN, C., and BRETT, M. (2000). Stereotaxic display ofbrain lesions. Behav. Neurol. 12, 191–200.

SANFILIPO, M., BENEDICT, R., WEINSTOCK-GUTTMAN,B., and BAKSHI, R. (2006). Gray and white matter brain at-rophy and neuropsychological impairment in multiple scle-rosis. Neurology 3, 12–20.

SCHAEFER, P.W., HULSMAN, T.A., SORENSEN, A.G.,GONZALEZ, R.G., and SCHWAMM, L.H. (2004). Diffu-sion-weighted MR imaging in closed head injury: high cor-relation with initial Glasgow Coma Scale score and score onModified Rankin Scale at discharge. Radiology 233, 58–66.

SCHEID, R., PREUL, C., GRUBER, O., WIGGINS, C., andCRAMON, D.Y. (2003). Diffuse axonal injury associated withchronic traumatic brain injury: evidence from T2-weighted gra-dient-echo imaging at 3T. Am. J. Neuroradiol. 24, 1049–1056.

SCHEID, R., WALHER, K., GUTHKE, T., PREUL, C., andCRAMON, D.Y. (2006). Cognitive sequelae of diffuse ax-onal injury. Arch. Neurol. 63, 418–424.

TAKAOKA, M., TABUSE, H., KUMURA, E., et al. (2002).Semiquantitative analysis of corpus callosum injury usingmagnetic resonance imaging indicates clinical severity in pa-tients with diffuse axonal injury. J. Neurosurg. Psychiatry 73,289–293.

TOMAIUOLO, F., CARLESIMO, G.A., DI PAOLA, M., et al.(2004). Gross morphology and morphometric sequelae in thehippocampus, fornix, and corpus callosum of patients withsevere non-missile traumatic brain injury without macro-scopically detectable lesions: a T1-weighted MRI study. J.Neurol. Neurosurg. Psychiatry 75, 1314–1322.

TOMAIUOLO, F., WORSLEY, K.J., LERCH, J., et al. (2005).Changes in white matter in long-term survivors of severenon-missile traumatic brain injury: a computational analysisof magnetic resonance images. J. Neurotrauma 22, 76–82.

TONG, K., ASHWAL, S., HOLSHOUSER, B., et al. (2004).Diffuse axonal injury in children: clinical correlation withhemorrhagic lesions. Ann. Neurol. 56, 36–50.

YOUNT, R., RASCHKE, K.A., BIRU, M., et al. (2002). Trau-matic brain injury and atrophy of the cingulated gyrus. J.Neuropsychiatry Clin. Neurosci. 14, 416–423.

WILSON, J.T., PETTIGREW, L.E., and TEASDALE, G.M.(1998). Structured interviews for the Glasgow Outcome Scaleand the Extended Glasgow Outcome Scale. J. Neurotrauma15, 573–585.

Address reprint requests to:Carlos Marquez de la Plata, Ph.D.

Department of NeurologyUniversity of Texas Southwestern Medical Center

5323 Harry Hines BoulevardDallas, TX 75390-9036

E-mail:[email protected]


598

magnetic resonance imaging of diffuse axonal injury: quantitative

Documents