| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Neuroradiology |
1 From the Departments of Radiology (P.W.S., T.A.G.M.H., A.G.S., R.G.G.) and Neurology (L.H.S.), Massachusetts General Hospital, 55 Fruit St, Gray B285, Boston, MA 02114. Received July 25, 2003; revision requested October 6; final revision received January 13, 2004; accepted February 13. Address correspondence to P.W.S. (e-mail: pschaefer@partners.org).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
MATERIALS AND METHODS: Twenty-six patients (18 male and eight female patients; mean age, 25.2 years; age range, 4–72 years) with diffuse axonal injury were examined with diffusion-weighted MR imaging and with fluid-attenuated inversion recovery, T2-weighted fast spin-echo, and T2*-weighted gradient-echo sequences. All images were evaluated by two neuroradiologists in consensus. Tissue volume with trauma-related signal-intensity abnormality on images from each sequence, number of lesions for each sequence, number of lesions for all sequences, and number of lesions with reduced apparent diffusion coefficient were correlated with scores on Glasgow Coma Scale and modified Rankin scale. Involvement of brainstem, deep gray matter, and corpus callosum were also correlated with clinical scores. Spearman rank correlation coefficients (r) were calculated.
RESULTS: The strongest correlation was between signal-intensity abnormality volume on diffusion-weighted images and modified Rankin score (r = 0.772, P < .001). The strength of this correlation did not improve when only volume of lesions with decreased apparent diffusion coefficient was considered. For lesion number, the strongest correlation was between lesion number on images acquired with all sequences and modified Rankin score (r = 0.662, P < .001). For lesion location, the strongest correlation was between lesion location in the corpus callosum and modified Rankin score (r = 0.513, P = .007).
CONCLUSION: Volume of lesions on diffusion-weighted MR images provides the strongest correlation with a score of subacute on modified Rankin scale at discharge. Total lesion number also correlates well with modified Rankin score. In future, diffusion-weighted images may be useful in determining treatment strategies for acute head injury.
© RSNA, 2004
Index terms: Brain, injuries, 10.40 • Brain, MR, 10.121412, 10.121413, 10.121416, 10.121419 • Magnetic resonance (MR), comparative studies, 10.121412, 10.121413, 10.121416, 10.121419 • Magnetic resonance (MR), diffusion tensor, 10.121419, 10.12144 • Trauma, 10.40, 10.436
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
Magnetic resonance (MR) imaging is valuable in detecting traumatic lesions (10–17). T2-weighted MR imaging can demonstrate hyperintense lesions at gray matter–white matter junctions, in the white matter, and in the brainstem. T2*-weighted MR imagingimproves detection of hemorrhagic lesions, which occur in 10%–30% of patients with diffuse axonal injury (17,18). However, both sequences are thought to lead to underestimation of injury extent (11,12), a hypothesis supported by the results of studies in which patients showed progressive global cerebral atrophy at follow-up after initial imaging that did not show widespread abnormalities (19,20). In addition, a discrepancy has been reported between findings with conventional MR imaging sequences and final outcomes. Studies have shown that although extensive white matter injury was consistently associated with a poor prognosis, the presence of focal signal-intensity abnormalities indicating shearing injury was seen in patients with both poor and good clinical outcomes (10).
Recently, diffusion-weighted MR imaging has shown promise in the evaluation of diffuse axonal injury. Diffusion-weighted imaging can demonstrate lesions that are not visualized with conventional MR sequences (21). In addition, the results of previous animal and human studies show that lesions with decreased or increased diffusion occur in head trauma (22–24), a finding that suggests that diffusion-weighted imaging may enable differentiation of cytotoxic from vasogenic edema in diffuse axonal injury. Thus, the purpose of our study was to determine whether diffusion-weighted imaging findings and conventional MR imaging findings correlate with the initial Glasgow Coma Scale score and the score on the modified Rankin scale at discharge.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
Patients were selected with an electronically assisted search of all radiologic reports for the following key phrases: diffuse axonal injury, shearing injury, closed head injury, and petechiae. Ninety-eight patients with closed head injuries were identified. Patients were subsequently excluded if (a) they had traumatic lesions that required emergency surgical intervention prior to MR imaging, (b) the time delay between trauma and imaging exceeded 48 hours, (c) images from one or several of the studied sequences were missing or were severely degraded by motion artifacts, or (d) there was a prior history of clinically important hypertension, diabetes mellitus, cerebrovascular disease, or another chronic medical problem that might lead to imaging abnormalities similar to those described for diffuse axonal injury. Ultimately, 26 patients (mean age, 25.2 years; age range, 4–72 years) were included in the study. There were 18 male patients (mean age, 25.4 years; age range, 4–72 years) and eight female patients (mean age, 24.5 years; age range, 5–42 years). There was no statistically significant difference between the age distributions in male versus female patients (P > .05). A review of the medical records by one of the authors (T.A.G.M.H.) showed that the mechanism of trauma (acceleration, deceleration, or rotatory force) was consistent with diffuse axonal injury of the brain in all patients. All patients were healthy prior to trauma, and none had experienced cardiopulmonary arrest. We recently reported our initial experience with diffusion-weighted MR imaging in 25 patients with diffuse axonal injury (24 of whom were included in the study reported in the present article) and the correlation of fractional anisotropy with clinical outcome in 20 patients (all of whom were included in the study reported here) (21,26). In an initial article (21), we delineated the number of diffuse axonal injury lesions identified on diffusion-weighted, gradient-echo (GRE), and T2-weighted images, and apparent diffusion coefficient (ADC) maps. In a second article (26), we measured fractional anisotropy and ADC values in deep gray matter nuclei and deep white matter locations and correlated each with scores on the Glasgow Coma Scale and modified Rankin scale. In this study, we counted the number of lesions depicted on images acquired with a fluid-attenuated inversion recovery (FLAIR) sequence, in addition to those depicted on ADC maps and images acquired with T2-weighted, diffusion-weighted, and GRE sequences, and measured lesion volumes on images obtained with each sequence. We then correlated these findings with scores on the Glasgow Coma Scale and modified Rankin scale.
MR Imaging
All patients were examined with a 1.5-T MR imaging unit (Signa; GE Medical Systems, Milwaukee, Wis) with echo-planar capabilities. MR examinations were performed according to standard departmental protocols.
The following sequences were applied: (a) transverse T2-weighted fast spin-echo (SE) (repetition time msec/echo time msec [effective], 5000–6000/108; field of view, 240 x 180 mm; matrix, 256 x 192; section thickness, 5 mm; echo train length, eight; signals acquired, two), (b) transverse FLAIR (repetition time msec/echo time msec/inversion time msec, 9000/120/2200; field of view, 240 x 180 mm; matrix, 256 x 192; section thickness, 5 mm; one signal acquired), (c) transverse T2*-weighted GRE (750/25; flip angle, 20°; field of view, 240 x 180 mm; matrix, 256 x 192; section thickness, 5 mm; number of signals acquired, two), and (d) transverse full diffusion-tensor imaging averaged over three data sets for a total acquisition time of 126 seconds. The entire diffusion tensor was sampled by applying a single-shot echo-planar SE pulse sequence in six noncollinear directions (6000/118; field of view, 40 x 20 cm; matrix, 256 x 128; section thickness, 6 mm; intersection gap, 1 mm; signals acquired, three). Diffusion-weighting gradients were applied at a finite low b value (3 sec/mm2) and a high b value (1221 sec/mm2).
Isotropic diffusion-weighted images were digitally reconstructed by calculating the geometric mean of the six images obtained with a b value of 1221 sec/mm2. In addition, ADC maps were calculated. The details of the diffusion-weighted imaging sequence and data processing were similar to those described elsewhere (27,28).
Image Analysis
All images were evaluated in consensus by two neuroradiologists (P.W.S., with 10 years of experience; T.A.G.M.H., with 8 years of experience). Images were evaluated for the number of lesions defined as circumscribed foci of signal-intensity abnormality, including hyperintense, hypointense, and mixed hyperintense and hypointense lesions. Diffusion-weighted images were evaluated in conjunction with the corresponding ADC maps to differentiate between lesions with decreased, increased, or unchanged diffusion rate. The total number of lesions was determined for each sequence separately, as well as for the combination of all sequences. The number of lesions with decreased ADC was also determined. Since susceptibility effects from blood products can interfere with reliable ADC measurements, hypointense (hemorrhagic) lesions depicted on diffusion-weighted images were not evaluated for their ADC change. We compared lesion location on images obtained with each sequence so that identical lesions depicted with different sequences were not counted multiple times.
In addition, the total volume of signal-intensity abnormalities was determined on images obtained with each sequence. Images were transferred to a personal computer (Macintosh; Apple, Cupertino, Calif). To calculate lesion size, one radiologist (T.A.G.M.H.) outlined the margins of each lesion manually by using image display software (Alice; Hayden Image Processing Solutions, Boulder, Colo). The volume of the lesions was calculated by multiplying the number of voxels that contained signal-intensity abnormalities by the respective voxel volume.
The presence of deep traumatic lesions was determined for the following locations: brainstem, basal ganglia and/or thalamus, and corpus callosum. Results were rated as either positive (signal-intensity abnormality present) or negative (signal-intensity abnormality absent).
Clinical Scoring
The Glasgow Coma Scale score recorded at the scene or in the emergency department, during the acute phase of injury, and the modified Rankin score at discharge were obtained from the medical records. The Glasgow Coma Scale score ranged from 3 (worst score) to 15 (best score) and was based on the combined evaluation of three categories of neurologic function: eye opening, best verbal response, and best motor response. The modified Rankin score was assigned retrospectively by a board-certified neurologist (L.H.S.) using all information available. The modified Rankin scale is a commonly used outcome classification system for indicating the level of disability after cerebral stroke (25). The scale includes the following seven grades: 0, no symptoms; 1, no substantial disability, despite symptoms (able to carry out all usual duties and activities); 2, slight disability (unable to carry out all previous activities but able to look after own affairs without assistance); 3, moderate disability (requiring some help, but able to walk without assistance); 4, moderately severe disability (unable to walk without assistance and unable to attend to own bodily needs without assistance); 5, severe disability (bedridden, incontinent, and requiring constant nursing care and attention); and 6, death.
Statistical Analysis
Statistical analysis was performed by using statistical software (SAS, version 8.0; SAS Institute, Cary, NC). Spearman rank correlation coefficients (r) were calculated to investigate possible correlations between one or more of the previously described MR imaging findings (number, volume, and location of shearing lesions) and the Glasgow Coma Scale and modified Rankin scores. The correlation between Glasgow Coma Scale score and modified Rankin score was also calculated. P values of less than .05 were considered to indicate a statistically significant difference. In addition, a multiple regression analysis (analysis of variance) was performed.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
Lesion Characterization
The total number and total volume of lesions depicted with each sequence, as well as the totals for the combination of all four MR imaging sequences, are presented in Table 1. A total of 488 traumatic lesions were counted in 26 patients. Diffusion-weighted imaging depicted the largest number of lesions (365 [75%] of 488), followed by FLAIR (297 [61%] of 488), T2-weighted fast SE (245 [50%] of 488), and T2*-weighted GRE (211 [43%] of 488) sequences (Fig 1). The majority of the lesions identified on diffusion-weighted images (239 [65%] of 365 lesions) showed decreased diffusion on the corresponding ADC maps. The highest total volume with signal-intensity abnormalities (all lesions depicted with diffusion-weighted imaging) was measured on diffusion-weighted images, followed by FLAIR images, diffusion-weighted images with decreased diffusion on corresponding ADC maps, T2-weighted fast SE images, and T2*-weighted GRE images. Among the 26 patients, 12 (46%) had lesions in the brainstem, 12 (46%) had lesions within the basal ganglia and/or thalamus, and 16 (61%) had injuries of the corpus callosum. Only five (19%) of 26 patients had traumatic lesions in all three locations.
|
|
|
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
The volume of shearing injuries depicted with diffusion-weighted imaging may demonstrate the strongest correlation with clinical scales for a number of reasons. First, diffusion-weighted imaging measures a fundamentally different physiologic parameter, compared with those measured by conventional MR imaging sequences. Image contrast in diffusion-weighted imaging is related to differences in the diffusion rates of water molecules in the brain rather than to a change in total tissue water. The exact mechanisms underlying the changes in diffusion rate associated with diffuse axonal injury are not yet fully understood. The results of one experimental head trauma study suggest that decreased ADC results from trauma-induced brain ischemia (23). In rats subjected to impact acceleration traumatic injury, lesions with decreased ADC occurred only when the rats were simultaneously subjected to a combination of hypoxia and hypotension. No changes in ADC were seen when the rats were subjected to trauma alone. Other explanations for the observed reduction in ADC include the process of trauma-induced axotomy with formation of retraction balls and concomitant cytoskeletal collapse along the severed axons (29). Lesions with increased ADC result from an increased amount of extracellular water (vasogenic edema). The results of experimental studies by Hanstock et al (30) and Yuan et al (31) show that these lesions occur when the trauma does not reduce cerebral blood flow enough to induce ischemia. While a small percentage of lesions have diffusion rates similar to or higher than those of normal brain tissue, these lesions can usually be visualized with diffusion-weighted imaging sequences secondary to the inherent T2 component ("T2 shine-through").
Second, the high contrast-to-noise ratio with diffusion-weighted imaging enhances lesion conspicuity. Investigators in a previous study (32) demonstrated a contrast-to-noise ratio for diffusion-weighted images in patients with acute stroke 20 times higher than that for proton-density and/or T2-weighted MR images. Third, the echo-planar sequences used for diffusion-weighted imaging are much more sensitive to magnetic susceptibility effects than are T2-weighted or FLAIR sequences. Thus, hemorrhagic lesions are difficult to visualize on T2-weighted and FLAIR images, but many hemorrhagic shearing lesions are identified on diffusion-weighted images. Also, while a greater number of hemorrhagic lesions are seen on T2*-weighted GRE images than on diffusion-weighted images, T2*-weighted GRE sequences are less sensitive in depicting nonhemorrhagic lesions than are diffusion-weighted, T2-weighted fast SE, and FLAIR sequences. Fourth, while most previous studies (33–35) were focused on the correlation between the number of lesions depicted with a single MR sequence or with a combination of MR sequences and clinical scores, measurement of the total volume of signal-intensity abnormality in patients with diffuse axonal injury may better represent the extent of brain injury: A single large lesion might involve the same amount of brain tissue as do multiple small lesions.
The identification of lesions and the comparison of their location on images obtained with the various sequences allowed us to estimate the total number of lesions by integrating information from all sequences. The results of our study show that the correlation between the total number of lesions and score on the modified Rankin scale was higher compared with the correlation between the numbers of lesions depicted with individual sequences and the score on the modified Rankin scale. Previous studies have shown the respective value of each sequence that we evaluated; T2-weighted fast SE and FLAIR sequences are sensitive in depicting changes in total brain water content (eg, vasogenic edema, nonhemorrhagic lesions) (16,36), T2*-weighted GRE sequences are sensitive in depicting blood degradation products (hemorrhagic lesions) (17), and diffusion-weighted sequences depict changes in the diffusion rates of water molecules. Thus, the complementary sensitivity of the sequences for the different types of pathophysiologic processes increased the overall correlation of total number of lesions with the score on the modified Rankin scale. This integrative approach could not be used for establishing correlations with the total volume of lesions, however, because the sequences were not coregistered.
Despite the fact that T2*-weighted GRE sequences are very sensitive for detection of hemorrhagic lesions (17), these sequences had a limited correlation with clinical markers in our study. The only statistically significant correlation was found between the number of lesions on T2*-weighted GRE images and the score on the modified Rankin scale, and this correlation proved to be the weakest (r = 0.418) compared with that for the numbers of lesions visualized with other investigated MR sequences and modified Rankin score. Our findings match those of a previously published report (33) of a study in which no significant correlation was found between the total volume of lesions depicted on T2*-weighted GRE images and clinical outcome. Two reasons could explain this poor correlation: First, not all diffuse axonal injuries are hemorrhagic. Neuropathologic studies distinguish between three grades of diffuse axonal injuries. In grade 1 lesions, there is widespread axonal damage in the cerebral white matter, corpus callosum, and brainstem, without hemorrhage. Lesions of grades 2 and 3 show extensive white-matter damage with tissue-tear hemorrhages in the corpus callosum and the brainstem, respectively (37). Furthermore, in one MR imaging study, only 10%–30% of diffuse axonal injuries were hemorrhagic (6). Second, T2*-weighted GRE sequences are less sensitive than T2-weighted fast SE, FLAIR, and diffusion-weighted sequences for depicting nonhemorrhagic (edematous) lesions, because of the higher spatial resolution with T2-weighted fast SE and FLAIR sequences and the high sensitivity for depiction of changes in water diffusion with diffusion-weighted sequences. Consequently, T2*-weighted GRE sequences depict a relatively small percentage of diffuse axonal injuries. Our results suggest that hemorrhagic lesions are not an indicator of more severe injury but more likely represent a part of the spectrum of lesions seen in diffuse axonal injury.
While injuries to the brainstem, deep gray matter nuclei, and corpus callosum are believed to indicate a greater severity of injury, the presence of injury in these locations alone, in our cohort of patients, did not reliably predict coma at presentation or poor outcome at discharge. In fact, our results showed a relatively poor correlation or no correlation between injury in these specific locations and measures of clinical outcome. The strongest correlation was between injury of the corpus callosum and modified Rankin score (r = 0.513). A weak correlation was also identified between involvement of the deep gray matter and score on the Glasgow Coma Scale (r = –0.162). There were no correlations between involvement of the deep gray matter or brainstem and score on the modified Rankin scale or between involvement of the corpus callosum or brainstem and score on the Glasgow Coma Scale. The poor correlation of the presence of lesions in the brainstem, deep gray matter nuclei, and corpus callosum with clinical outcome scales could have occurred because we did not take lesion size, presence of hemorrhage, or associated brain injury into account. Also, it is possible that tiny lesions in the basal ganglia and brainstem could represent small infarcts from injury to perforating vessels (eg, thalamostriate, lenticulostriate, or pontine arteries) or transient edema rather than diffuse axonal injury.
In general, most of the studied sequences showed a stronger radiologic-clinical correlation with the score on the modified Rankin scale than with the initial Glasgow Coma Scale score. In addition, the volume of shearing injuries on diffusion-weighted images versus the score on the modified Rankin scale, and the number of shearing injuries on diffusion-weighted images, all images, and T2-weighted images versus modified Rankin score, had higher correlation coefficients than that for the score on the Glasgow Coma Scale versus the score on the modified Rankin scale. This finding correlates with the results from previous studies that have addressed the shortcomings of the initial Glasgow Coma Scale score, especially for predicting outcome in patients with middle-range Glasgow Coma Scale scores (38–40). In the hyperacute setting, the patient’s level of consciousness is a poor predictor of long-term outcome, because loss of consciousness is highly variable in the early hours after injury and can be substantially altered by the central nervous system side effects of acute medical treatment. In spite of these limitations, the Glasgow Coma Scale remains the most widely used scoring system in acute cerebral disorders because it is simple and reproducible and because its use requires little training. Conversely, the modified Rankin score is a functional global outcome score that indicates the degree of disability rather than the level of consciousness. Degrees of disability are better defined and may be easier to evaluate than levels of consciousness (25,41). In addition, assessments of disability at discharge tend to be more stable than initial assessments of level of consciousness, which can change during the early hospital course, depending on the timing of assessments and the state of the patient.
The study had a number of limitations. Routine clinical management was not altered for the purposes of this observational study. Thus, the time of initial and follow-up imaging was variable, and follow-up imaging was performed with computed tomography (CT) in some cases, instead of MR imaging. The evolution of the signal intensity changes and their possible dependence on the elapsed time between trauma and imaging could interfere with the proper identification of shearing lesions, especially on images acquired with diffusion-weighted sequences. While we excluded patients with a history of hypertension, diabetes, vascular disease, and other chronic medical problems, it is possible that lesions thought to represent diffuse axonal injury were preexisting lesions. The 6-mm sections depicted on diffusion-weighted images were slightly thicker than the sections depicted with other sequences (5 mm). This difference might have favored increased detection on diffusion-weighted images because of increased signal-to-noise ratio, or increased detection on other images because of slightly increased spatial resolution. Susceptibility effects are prominent at the skull base on diffusion-weighted images and might have obscured lesions in the inferior frontal and anterior temporal lobes. Susceptibility effects associated with hemorrhagic lesions could have obscured adjacent lesions on diffusion-weighted images. Punctate lesions may be difficult to confirm on ADC maps because of low signal-to-noise ratio, and we may have underestimated the number of lesions with decreased ADC. Seven of the 26 patients had long-bone fractures, and some lesions that were thought to represent diffuse axonal injury could have resulted from fat emboli. None of the lesions could be verified with histopathologic examination, because all patients survived. We were unable to determine whether the lesions had resolved, because most follow-up imaging was performed with CT. Patients with the most severe degree of head injury were likely not to be included, because these patients are not stable enough to undergo emergency MR imaging. Also, we have only short-term clinical follow-up results, whereas the correlation between imaging and long-term follow-up may be of more clinical importance. Many factors, such as infection, hypoxemia, or hypotension, may be present before or after neurologic imaging and may influence clinical outcome in traumatic brain injury. Finally, our results are limited to the acute setting, since restricted diffusion is not observed in patients with a history of chronic head injury.
Among all of the sequences investigated, diffusion-weighted sequences showed the strongest correlation between lesion volume and subacute modified Rankin score at discharge. This likely was the strongest correlation because (a) the majority of diffuse axonal injuries identified were characterized by changes in water diffusivity; (b) diffusion-weighted imaging also demonstrates increased contrast-to-noise ratio, compared with conventional sequences; and (c) diffusion-weighted imaging is vulnerable to susceptibility effects and, therefore, depicts many hemorrhagic lesions. The volume of diffuse axonal injuries likely reflects aggregate injury better than does the number of lesions. Accordingly, lesion volume on diffusion-weighted images was more indicative of short-term neurologic outcome than was lesion number on any or all sequences. Future prospective studies with initial and follow-up imaging at predefined time points are necessary. In the future, diffusion-weighted imaging may be useful for determining treatment strategies for patients with acute head injury.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
P.W.S. and T.A.G.M.H. contributed equally to this work.
Authors stated no financial relationship to disclose.
Author contributions: Guarantors of integrity of entire study, P.W.S., T.A.G.M.H.; study concepts and design, all authors; literature research, T.A.G.M.H.; clinical studies, P.W.S., T.A.G.M.H., L.H.S.; data acquisition, P.W.S., T.A.G.M.H.; data analysis/interpretation, P.W.S., T.A.G.M.H., L.H.S.; manuscript preparation, P.W.S., T.A.G.M.H.; manuscript definition of intellectual content, editing, revision/review, and final version approval, all authors
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  D.R. Rutgers, P. Fillard, G. Paradot, M. Tadie, P. Lasjaunias, and D. Ducreux Diffusion Tensor Imaging Characteristics of the Corpus Callosum in Mild, Moderate, and Severe Traumatic Brain Injury AJNR Am. J. Neuroradiol., October 1, 2008; 29(9): 1730 - 1735. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. Ghajar and R. B. Ivry The Predictive Brain State: Timing Deficiency in Traumatic Brain Injury? Neurorehabil Neural Repair, May 1, 2008; 22(3): 217 - 227. [Abstract] [PDF]
|
|
|
|
 |
|
 E. Park PhD, J. D. Bell BSc, and A. J. Baker MD Traumatic brain injury: Can the consequences be stopped? Can. Med. Assoc. J., April 22, 2008; 178(9): 1163 - 1170. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 I. Kharatishvili, R. Immonen, O. Grohn, and A. Pitkanen Quantitative diffusion MRI of hippocampus as a surrogate marker for post-traumatic epileptogenesis Brain, December 1, 2007; 130(12): 3155 - 3168. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. P. Coles Imaging after brain injury Br. J. Anaesth., July 1, 2007; 99(1): 49 - 60. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. Zhang, L.A. Heier, R.D. Zimmerman, B. Jordan, and A.M. Ulug Diffusion anisotropy changes in the brains of professional boxers. AJNR Am. J. Neuroradiol., October 1, 2006; 27(9): 2000 - 2004. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. Scheid, K. Walther, T. Guthke, C. Preul, and D. Y. von Cramon Cognitive sequelae of diffuse axonal injury. Arch Neurol, March 1, 2006; 63(3): 418 - 424. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
