WHITE PAPER

Imaging Evidence and Recommendations for Traumatic Brain : Advanced Neuro- and Neurovascular Imaging Techniques

M. Wintermark, P.C. Sanelli, Y. Anzai, A.J. Tsiouris, and C.T. Whitlow, on behalf of the American College of Radiology Institute

ABSTRACT SUMMARY: Neuroimaging plays a critical role in the evaluation of patients with , with NCCT as the first-line of imaging for patients with traumatic brain injury and MR imaging being recommended in specific settings. Advanced neuroimaging techniques, including MR imaging DTI, blood oxygen level–dependent fMRI, MR spectroscopy, perfusion imaging, PET/SPECT, and magnetoencephalography, are of particular interest in identifying further injury in patients with traumatic brain injury when conventional NCCT and MR imaging findings are normal, as well as for prognostication in patients with persistent symptoms. These advanced neuroimaging techniques are currently under investigation in an attempt to optimize them and substantiate their clinical relevance in individual patients. However, the data currently available confine their use to the research arena for group comparisons, and there remains insufficient evidence at the time of this writing to conclude that these advanced techniques can be used for routine clinical use at the individual patient level. TBI imaging is a rapidly evolving field, and a number of the recommendations presented will be updated in the future to reflect the advances in medical knowledge.

ABBREVIATIONS: BOLD ϭ blood oxygen level–dependent; FA ϭ fractional anisotropy; MD ϭ mean diffusivity; MEG ϭ magnetoencephalography; TBI ϭ traumatic brain injury

raumatic brain injury (TBI) is one of the most common neu- resulting in approximately 275,000 hospitalizations and 52,000 Trologic disorders, currently affecting 1.7 million Americans deaths each year. Furthermore, TBI contributes to one-third of all each year.1,2 The incidence of TBI, especially mild TBI, is under- injury-related deaths in the United States. The economic cost of TBI estimated3 because patients frequently dismiss their symptoms was estimated at $76.3 billion in 2010 ($11.5 billion in direct medical and never present to the emergency department or they believe costs and $64.8 billion in indirect costs such as lost wages, lost pro- that the admission of symptoms may compromise their work sit- ductivity, and nonmedical expenditures).6 Moreover, affected mili- 4 uation (eg, athletes, military ). Although most patients (nearly tary veterans generate an annual cost of $11,700 in medical treatment 80%) with diagnosed TBI are treated and released from the emer- per patient compared with $2,400 in TBI-free veterans.7 Leading 5 gency department, the remaining 20% have more severe causes of TBI in the general population include falls, motor vehicle collisions, assaults, and sports-related injuries. Advanced neuroimaging techniques, including MR imaging From the Division of Neuroradiology (M.W.), Stanford University, Palo Alto, Cali- fornia; Department of Radiology (P.C.S.), North Shore–LIJ Health System, Manhas- DTI, blood oxygen level–dependent (BOLD) fMRI, MR spectros- set, New York; Department of Radiology (Y.A.), University of Washington, Seattle, copy, perfusion imaging, PET/SPECT, and magnetoencephalog- Washington; Department of Radiology (A.J.T.), Weill Cornell Medical College, New York-Presbyterian Hospital, New York, New York; and Department of Radiology raphy (MEG), are of particular interest in identifying further in- and Translational Science Institute (C.T.W.), Wake Forest School of Medicine, Win- jury in patients with TBI when conventional NCCT and MR ston-Salem, North Carolina. imaging findings are normal, as well as for prognostication in M. Wintermark and P.C. Sanelli are co-first authors of this article. patients with persistent symptoms. Based on the National Insti- This article is endorsed by the American College of Radiology Head Injury Institute, the American Society of Neuroradiology, the American Society of Functional Neuroradiol- tute for Health and Care Excellence (http://www.nice.org.uk), ogy, and the American Society of Pediatric Neuroradiology. Another article addressing the evidence and recommendations for conventional neuroimaging techniques in adapted from the Oxford Centre for Evidence-Based Medicine traumatic brain injury is published in Wintermark M, Sanelli PC, Anzai Y, on behalf of (http://www.cebm.net) Levels of Evidence (2001), we indicated the the American College of Radiology Head Injury Institute. Imaging evidence and recom- mendations for traumatic brain injury: conventional neuroimaging techniques. JAm quality of publications for diagnostic tests and interventions by as- Coll Radiol; http://dx.doi.org/10.1016/j.jacr.2014.10.014. signing stratified and preferential levels of evidence (Table 1) and Please address correspondence to Max Wintermark, MD, MAS, Stanford Univer- classes of recommendations (Table 2). Overall, at the time of writing sity, Department of Radiology, Neuroradiology Division, 300 Pasteur Dr, Room S047, Stanford, CA 94305-5105; e-mail: [email protected] this article, there is insufficient evidence supporting the routine clin- Indicates open access to non-subscribers at www.ajnr.org ical use of advanced neuroimaging for diagnosis and/or prognostica- http://dx.doi.org/10.3174/ajnr.A4181 tion at the individual patient level (class IIb recommendation).

AJNR Am J Neuroradiol 36:E1–E11 Feb 2015 www.ajnr.org E1 Table 1: Levels of evidence for studies of the accuracy of described in terms of 3D space, for which that parameter is valid. a diagnostic tests There is a variety of ways that DTI can quantify the diffusion Levels of Evidence Type of Evidence ellipsoid, the most common of which is to compute scalar metrics, Ia Systematic review (with homogeneity)b of level-1 studiesc comprising fractional anisotropy (FA) for measuring ellipsoid Ib Level-1 studiesc shape and mean diffusivity (MD) for measuring ellipsoid size. II Level-2 studiesd TBI-associated changes in DTI scalar metrics have been well-doc- Systematic reviews of level-2 studies umented, FA in particular.9-14 In addition to measuring a variety III Level-3 studiese Systematic reviews of level-3 studies of DTI metrics, data processing and analysis techniques have been IV Consensus, expert committee reports or opinions and/or developed to utilize these data for visualizing abnormalities in clinical experience without explicit critical appraisal; 15,16 or based on physiology, bench research, or “first gray matter and white matter tracts. principles” With regard to white matter tracts, the diffusion tensor model a Adapted from The Oxford Centre for Evidence-Based Medicine Levels of Evidence is not capable of resolving multiple fiber orientations within a (2001) and the Centre for Reviews and Dissemination Report Number 4 (2001). Copy- voxel, and improved diffusion models and analysis methods are right National Institute for Health and Care Excellence February 2004. b Homogeneity means there are no or minor variations in the directions and degrees actively being developed. High-angular-resolution diffusion im- 17 of results between individual studies that are included in the systematic review. aging acquisitions use significantly more diffusion directions c Level-1 studies are studies: compared with many DTI protocols (for instance, 32 or higher). ● that use a blind comparison of the test with a validated reference standard. 18 ● in a sample of patients that reflects the population to whom the test would Diffusion kurtosis imaging uses a minimum of 3 b-values (in- apply. stead of the typical 2 for DTI, eg, 0, 1000, 3000), providing addi- d Level-2 studies are studies that have only 1 of the following: tional metrics related to non-Gaussian (kurtosis) diffusion that ● narrow population (the sample does not reflect the population to whom the test 19 would apply). can provide a more physiologic basis for white matter modeling 20 ● use a poor reference standard (defined as that where the “test” is included in the and for resolving fiber crossings. Diffusion spectroscopy imag- “reference,” or where the “testing” affects the “reference”). ing 21 requires a very comprehensive sampling scheme, making it ● the comparison between the test and reference standard is not blind. ● case-control studies. sensitive to intravoxel heterogeneities in diffusion directions e Level-3 studies are studies that have at least 2 or 3 of the features listed above. caused by crossing fiber tracts and thus allows more accurate mapping of axonal trajectories than DTI-based approaches. Q- Table 2: Classification of recommendationsa Ball imaging22 attempts a similar granular mapping resolution Class I Conditions for which there is evidence for and/or general agreement that a procedure or treatment is by using a much smaller diffusion sampling scheme. Newer beneficial, useful, and effective acquisition methods, such as multiband,23 and continued im- Class II Conditions for which there is conflicting evidence and/or a divergence of opinion about the provement in clinical scanner gradient performance, achieving usefulness/efficacy of a procedure or treatment maximum gradient strengths as high as 80 mT/m, are being de- Class IIa Weight of evidence/opinion is in favor of usefulness/ veloped that allow for faster scan times, making advanced diffu- efficacy Class IIb Usefulness/efficacy is less well-established sion techniques easier to implement as part of routine clinical Class III Conditions for which there is evidence and/or general protocols. agreement that a procedure/treatment is not useful/effective and in some cases may As mentioned above, DTI cannot resolve crossing fibers, and be harmful FA normally decreases in regions of crossing fibers. Methods for a From the American Heart Association Evidence-Based Scoring System. resolving crossing fibers include the neurite orientation disper- sion and density imaging model24 (which derives a measurement To date, advanced neuroimaging techniques have been uti- of intracellular volume fraction, a potential replacement of frac- lized primarily for research, by using a variety of statistical and tional anisotropy that should be less confounded by crossing fi- numeric approaches to assess for differences between study 22 bers), Q-ball imaging (which assumes a Gaussian model), and groups. Validated diagnostic tools to interpret advanced neuro- 25 constrained spheric deconvolution (which creates a model imaging techniques within a single patient do not presently exist. The based on the actual diffusion acquisition that is used to perform main challenges for developing these tools include the following: 1) deconvolution and produces very sharp fiber-orientation distri- lack of large-scale, age-stratified normal data with available advanced bution functions). Further studies, however, are required to de- neuroimaging techniques by using standardized protocols developed termine whether one of these approaches has a clear advantage with consensus by clinical and research communities, 2) lack of clear compared with the others. Improvements in data quality with patterns of injury that are predictive of clinical and neuropsycholog- 26 higher order eddy current correction, distortion correction, and ical deficits, and 3) poor definition of standard approaches to ac- high-order shimming beyond second order should further im- count for technical differences between clinical scanners that may prove data fidelity and increase the reliability of subsequent data introduce artifactual false-positives or -negatives into an assessment. processing.

Advanced Diffusion Imaging Techniques, Including Evidence and Recommendations for Advanced Diffusion Imaging Diffusion Tensor Imaging Techniques, including Diffusion Tensor Imaging. Most studies of Diffusion tensor imaging8 is a well-established model using data TBI have reported decreases in FA and increases in MD, thought derived from directionally encoded MR diffusion-weighted imag- to be secondary to demyelination or disruption of tissue micro- ing, estimating the overall direction and degree of restriction of structural integrity.27-36 Some studies have reported both water diffusion. In DTI, each voxel has one or more pairs of pa- trauma-related decreases and increases in FA, particularly in the rameters: a rate of diffusion and a preferred direction of diffusion, subacute phase postinjury. New evidence is converging to suggest E2 Wintermark Feb 2015 www.ajnr.org a more complex pattern of trauma-related changes in the white ple spatially distributed brain regions. A variety of techniques can matter. Indeed many studies are reporting that FA values increase be used to study distributed whole-brain networks, such as inde- in both acute and chronic phases of injury.12,37-41 Variability in pendent component analysis/seed-based methods. Network the data regarding the direction of TBI-related change for com- properties of the brain can be further characterized by using graph mon DTI metrics may reflect heterogeneity within cohort studies, theoretic analysis to identify changes in network function, such as including variation in TBI severity, variability in spatial injury local and global efficiency, as well as small-worldness, among location, inconsistent imaging parameters, and marked differ- many other network connectivity metrics. ences between the time of injury and imaging.42 It is important to Evidence and Recommendations for Functional MR Imaging. fMRI note that alterations in DTI metrics including FA are not specific methods demonstrate great potential for evaluating brain subsys- to TBI and have also been documented in a wide variety of CNS tems that may underlie TBI-associated behavioral and cognitive disorders, particularly those that affect white matter.43 When impairment, including the detection of whole-brain changes in these issues are considered altogether, they suggest a need for functional connectivity across a variety of brain networks, as well more studies to translate this advanced imaging method into a as more focused task-specific changes in functional activity clinically useful and applicable tool for TBI. among targeted brain subregions. Only a few behavioral/cognitive Currently, there is evidence from group analyses that DTI can domains have been evaluated with task-based studies, and only a identify TBI-associated changes in the brain across a range of few functional networks, primarily the default mode, have been injury severity, from mild to severe TBI. Evidence also suggests evaluated with resting-state techniques, which limits the general- that DTI has the sensitivity necessary to detect acute and chronic izability of these methods to the breadth of TBI-associated se- TBI-associated changes in the brain, some of which correlate with quelae.41 Important limitations have been raised regarding the 44 injury outcomes. These data, however, are based primarily use of BOLD fMRI techniques for accurately characterizing upon group analyses, and there is insufficient evidence at the time changes in patients with TBI because of the possibility that brain of writing this article that DTI can be used for routine clinical injury might uncouple CBF and neural activity.51 Indeed, there is diagnosis and/or prognostication at the individual patient evidence that TBI is associated with reductions or increases in 41,44,45 level. Even though a few studies have reported z score CBF that have been described during the acute stages of methods for individual patient TBI evaluation, there remains in- injury.51,52 sufficient evidence at the time of writing to suggest that these There is insufficient evidence at the time of this writing that methods are valid, sensitive, and specific for routine clinical eval- fMRI based on BOLD techniques can be used for routine clinical uation of TBI at the individual patient level (class IIb TBI diagnosis and/or prognostication at the individual patient recommendation). level (class IIb recommendation).

Functional MR Imaging MR Spectroscopy Functional MR imaging relies on blood oxygen level–dependent MR spectroscopy is governed by the same physical principles of imaging and the coupling between CBF/metabolism and neuro- magnetism as MR imaging. While MR imaging data are analyzed 46-50 nal activity. As metabolic demand increases, there are local in the time domain to obtain T1 and T2 relaxation times and then increases in CBF, as well as dynamic and related metabolic processed to generate an anatomic image, MR spectroscopy data 46-50 changes in glucose and oxygen metabolism. Transient local are converted to frequency domain information and processed to increases in CBF and metabolism lead to changes in the ratio of form a spectrum of the signal intensities of different brain metab- oxygenated-to-deoxygenated hemoglobin, which, in turn, affects olites according to their Larmor resonance frequencies. 46-50 the MR imaging signal response. BOLD imaging offers milli- MR spectroscopy can be performed utilizing single-voxel meter-scale spatial resolution but poor temporal resolution due spectroscopy, or 2D or 3D multivoxel chemical shift techniques to the relatively slow hemodynamic response associated with neu- (MR spectroscopic imaging or chemical shift imaging). Single- ral activity. voxel spectroscopy has a greater signal-to-noise ratio and is more BOLD fMRI methods for investigating TBI have utilized task- robust, but only a single spectrum is obtained and the volume of based methods, particularly working memory paradigms. Task- interest placement is crucial. Chemical shift imaging limits sam- based methods require participation of the research subject to pling error by covering a much larger area, at the expense of a identify activation of brain regions thought to drive or be associ- lower signal-to-noise ratio and longer scan times. The 2 most ated with task performance. Task-free resting-state BOLD fMRI widely used MR spectroscopy techniques are point-resolved spec- techniques have also been used, with the advantage of being able troscopy sequence and the stimulated echo acquisition mode. to evaluate distributed whole-brain networks without requiring Commonly quantified brain metabolites with intermediate (TE ϭ overt behavioral output from subjects. These methods rely upon 144 ms) to long (TE ϭ 280 ms) TEs are N-acetylaspartate (NAA) identification of fluctuations in low-frequency BOLD waveforms for neuronal integrity, creatine (Cr) for cellular energy/attenua- throughout the brain, which are thought to reflect underlying tion, choline (Cho) for membrane turnover, and lactate for an- neural activity that occurs in the absence of active task perfor- aerobic metabolism. At shorter TEs (TE ϭ 35 ms), metabolites mance. Functional connectivity between brain regions is opera- with shorter T2 relaxation times can be detected, such as gluta- tionally defined as a strong interregional correlation in these low- mate/glutamine (Glx), which are excitatory amino acids released frequency BOLD waveforms, which is the basis for identifying after brain injury,53 and myo-inositol, thought to be a marker of functionally connected whole-brain networks comprising multi- astroglial proliferation.54 Each brain metabolite resonates at a AJNR Am J Neuroradiol 36:E1–E11 Feb 2015 www.ajnr.org E3 specific frequency and is plotted at a known chemical shift (mea- likely diffuse glial proliferation, corroborated by elevated myo- sured in parts per million) on a graphic spectrum. The area under inositol that persists for months after injury.82 More recent stud- each spectral peak corresponds to the metabolite concentration ies have shown changes in the energy marker Cr58,83;ifCrisaf- and can be reported as a ratio to Cr (such as NAA/Cr and Cho/ fected by mild TBI, metabolite ratio measurements would not be Cr); Cr has been used as an internal standard since it was thought accurate because it would be difficult to assess if changes were due to remain relatively constant. It is now known that Cr may change to the metabolite of interest or in Cr itself. in certain conditions55-57 including TBI58; however, Cr-based ratios A number of small prospective cross-sectional and case-con- remain useful when comparing serial measurements or data between trol longitudinal cohort studies have associated MR spectroscopy institutions. Methods for absolute metabolite quantitation are now metabolite levels with neuropsychological outcomes in the mod- used routinely, by using water as an internal reference or phantoms erate-to-severe TBI population. Most (but not all) of these studies containing known metabolite concentrations.59-61 When imaging have demonstrated decreased absolute NAA and NAA/Cr ratios children, normal age-matched controls must be used, since and decreased NAA to correlate with poor outcomes.54,67,84-86 brain metabolites change with maturation and myelination, Decreased NAA/Cr ratios have been detected in the splenium of rapidly in the first year of life, and then continue more slowly the corpus callosum and, to a lesser, extent in the lobar white through adolescence.62-66 matter.80 In pediatric patients with accidental and nonaccidental TBI, Evidence and Recommendations for MR Spectroscopy. There is MR spectroscopy has shown potential for providing early prog- significant heterogeneity in the literature that is, in part, related to nostic information regarding clinical outcomes.62,87-94 Reduced the inclusion of patients with variable TBI severity, spatial injury NAA has been correlated with impaired long-term neuropsycho- locations, differences between time of injury and imaging, and logical function in children.89,90,92,94 Elevated total Cho has also analysis with various MR spectroscopy techniques. There are cur- been described and may be related to diffuse axonal injury and/or rently no large prospective studies examining the sensitivity and repair.84,95 Studies have shown that elevated lactate levels are specificity of MR spectroscopy in the setting of TBI. A few small more common following nonaccidental TBI62,87 and are strongly studies in patients with moderate-to-severe TBI show promise in correlated with poor outcomes. In children, myo-inositol has the utility of MR spectroscopy as prognostic for outcomes.67,68 been shown to increase as a result of glial proliferation and has In mild TBI, the most common finding is a widespread de- also been correlated with poor outcomes after TBI.82,96 crease in gray matter and white matter NAA.69-75 However, dis- Currently, there remains insufficient evidence at the time of parity exists between studies in the quantification of Cho in TBI; writing to conclude that MR spectroscopy is sufficiently sensitive some studies have found increased Cho in various regions in the and specific for routine clinical use at the individual patient level brain parenchyma,70,73,76 while others have reported the absence (class IIb recommendation). of statistical changes in Cho.58,74,75,77 A few recent small longitu- dinal prospective controlled cohort studies have shown an initial Magnetoencephalography decrease in NAA, and increase in Cr and Glx in the white matter, In MEG, neuromagnetometers surround the head with hundreds but decreased Glx in the gray matter. In all cases, NAA, Cr, and of sensors connected to superconducting quantum interference Glx returned to normal levels by the end of the study, suggesting devices that are cooled to 4° Kelvin with liquid helium. The output recovery. However, in a recent pediatric study, initial NAA levels of each sensor is a waveform that reveals local magnetic fluctua- did not change following concussion.78 A recent controlled cross- tions and how they change with time. These fluctuations are the sectional prospective study that included patients in the early sub- result of synchronous postsynaptic intracellular electric currents acute, late subacute, and chronic stages of mild TBI revealed de- produced by pyramidal neurons of the cerebral cortex that are creasing trends in thalamic NAA/Cr levels at the early subacute accompanied by perpendicularly oriented magnetic fields.97,98 stage and decreases in Cho/Cr occurring in the thalamus and cen- Because the net current of pyramidal neuron depolarization is trum semiovale in the late subacute stage. Positive correlations oriented toward the cortical surface, MEG is most sensitive for between early subacute Cr levels in the centrum semiovale and detecting signals arising from sulcal walls, which are directed per- chronic cognitive performance on neuropsychiatric evaluation pendicular to the skull.98,99 MEG has intrinsically high temporal were noted. Although this provides some insights into brain bio- resolution, limited only by the sampling frequency of the elec- chemistry changes at a population level, the authors noted that the tronics,100 which is an advantage over other functional neuroim- applicability of these findings to an individual subject still requires aging methods, such as fMRI. However, MEG has inferior spatial careful examination of data from a larger population.79 resolution, largely because a single cortical source of signal can be Studies of MR spectroscopy in chronic mild TBI have found detected by multiple adjacent scalp sensors. As such, data must be decreases in NAA in the splenium of the corpus callosum,80 cen- processed to translate sensor signal back to the source space. trum semiovale,54 and frontal white matter.75 Chemical shift im- aging70,77,81 and whole-brain NAA studies69 have also demon- Evidence and Recommendations for Magnetoencephalogra- strated decreases in NAA in the white matter. Changes found in phy. There is insufficient evidence at the time of writing that Cho, a marker for cellular proliferation or tissue damage, may MEG can be used for routine clinical TBI diagnosis and/or prog- reflect diffuse axonal injury.54,70 In the acute phase of head injury, nostication at the individual patient level (class IIb recommenda- choline-containing metabolites may be released as a result of tion). Indeed, only a few studies have used MEG to evaluate the shear injury and damage to cell membranes and myelin.67 In effects of TBI101-104; therefore, more work is necessary to define chronic brain injury, the mechanism for increased Cho is more the utility and capabilities of this imaging technique. Very few E4 Wintermark Feb 2015 www.ajnr.org medical centers have modern neuromagnetometers, and analysis Perfusion imaging can also identify specific patterns, linked to of MEG data requires advanced expertise in signal processing, cerebral edema and intracranial hypertension.117,118 Finally, the which limits its widespread implementation. number of arterial territories with “low” regional cerebral blood volume (CBV) values on perfusion imaging is an independent Perfusion Imaging predictor of the functional outcome and is seen as early as on Brain perfusion in patients with TBI has been imaged by using dif- admission.116 The potential repercussions of perfusion imaging 105,106 ferent techniques, including stable xenon-enhanced CT, sin- on the clinical management of patients with severe TBI remain to 107-113 114,115 gle-photon emission CT (SPECT), PET, perfusion be evaluated. However, patients with altered brain perfusion 116-120 CT and perfusion-weighted MR imaging, including dy- might be considered for more aggressive and early treatment to 121,122 namic susceptibility contrast, and arterial spin-label- prevent intracranial hypertension, whereas patients with pre- 123,124 ing. These brain perfusion imaging techniques each have served brain perfusion might benefit from less invasive treatment. different drawbacks. SPECT and dynamic susceptibility contrast In patients with mild TBI and normal conventional brain im- only generate qualitative comparisons between the right and left aging findings, perfusion imaging studies have shown scattered 125 hemispheres. Stable xenon-enhanced CT, perfusion CT, and perfusion deficits, which are significantly correlated with neuro- 126,127 arterial spin-labeling are quantitatively accurate. psychological and neurocognitive impairment, as well as post- Stable xenon-enhanced CT requires specialized and expensive traumatic amnesia.109,110,112,119,122,132,133 equipment. A typical study is relatively long, approximately 10 minutes. Side effects, such as respiratory rate decrease, headaches, Evidence and Recommendations for Perfusion Imaging. All the nausea and vomiting, as well as convulsions, are observed in 4.4% studies above have involved only limited numbers of patients, and of patients. Consequently, stable xenon-enhanced CT is difficult further research is required to validate the findings described to perform, especially in the emergency setting.128 above and determine how relevant they are in the management of Perfusion-weighted MR imaging is also difficult to obtain in individual patients with TBI (class IIb recommendation). the acute setting due to scanner availability and patient contrain- dications but is an attractive imaging technique for the subacute PET and SPECT and chronic phases, especially because it can be combined with Most studies utilizing PET and SPECT to investigate TBI have other MR imaging sequences sensitive to TBI lesions. focused on cerebral blood flow and metabolism and demon- 134-138 On the other hand, perfusion CT is more readily available in strated TBI-associated decreases in cohort studies. Recent 139 the emergency/acute setting.120 It can be implemented in all hos- literature reviews summarizing the possible utility of PET and 140 pital institutions equipped with CT units and injectors, which are SPECT in TBI, particularly mild TBI, have drawn this cohort usually available 24/7. It necessitates neither specialized technol- data largely from cross-sectional designs with a smaller number of ogists nor extra material, but only requires dedicated postprocess- longitudinal studies also reviewed. Very few randomized con- ing software. It affords real-time postprocessing, with a complete trolled trials were discovered when examining the literature. Also, set of parametric maps typically generated within 5 minutes of various levels of analytic rigor have been used to identify TBI- completing data acquisition.129,130 Perfusion CT can easily be associated changes with these imaging methods, including quan- performed as a complement to conventional noncontrast CT titative and visual-based qualitative measures, which are known (NCCT) of the head and CT angiography (CTA) of the cerebral to have interrater variability. Additionally, very few PET and vasculature and does not interfere with the contrast-enhanced SPECT studies to date have gone beyond CBF and metabolic thoracoabdominal CT survey performed in patients with severe mapping to incorporate specific ligands for exploring TBI-related TBI.116 Patients with severe TBI typically undergo contrast-en- changes at the receptor level, which may open the door to a new hanced chest, abdomen, and pelvis CT routinely to evaluate for generation of sensitive and specific molecular biomarkers of brain 141 aortic injuries. The contrast material administration is performed injury. Recent PET amyloid imaging, however, suggests that 142 even in obtunded patients unable to report about possible previ- amyloid levels rise in persons with TBI. A small study with a ous contrast reactions or without knowing about the renal func- combined amyloid tau (␶) tracer also showed higher levels in tion because the risk associated with these conditions is out- those with TBI.143 The condition, sometimes referred to as weighed significantly by the risk of a missed traumatic aortic “chronic traumatic encephalopathy, ” is believed to be progressive injury. The dose of contrast material added for the perfusion CT is tauopathy, seen most commonly in persons with repetitive con- minor compared with the dose used for the chest, abdomen, and cussive brain injuries, such as professional athletes and military pelvis. personnel.144 Beta (␤) amyloid deposition, while seen in 40%– In patients with severe TBI, perfusion imaging affords insight 45% of chronic traumatic encephalopathy, is not as common as ␶ into regional brain perfusion alterations due to TBI, with the ma- accumulation.145 As TBI is a recognized risk factor for cognitive jor advantage being detection of regional heterogeneity.116 Perfu- decline and dementia,146 advanced neuroimaging will likely con- sion imaging has a higher sensitivity for the diagnosis of cerebral tinue to provide key insights into this disease process and possible contusions at baseline when compared with admission NCCT, avenues for routine and personalized clinical use. At present, with a sensitivity reaching 87.5% versus 39.6% for conventional more work must be conducted to translate these powerful neuro- NCCT.116,120,121 Perfusion imaging can show changes related to imaging methods into useful, routine clinical tools at the individ- the mass effect caused by an epidural or subdural hematoma and ual subject level, for which there is insufficient evidence at the the resolution of these changes after hematoma evacuation.131 time of writing this article (class IIb recommendation). AJNR Am J Neuroradiol 36:E1–E11 Feb 2015 www.ajnr.org E5 Summary trauma and who have no indication for immediate operative in- 160 1) Advanced neuroimaging techniques, including MR imaging tervention (class IIa recommendation). CTA can be used to DTI, BOLD fMRI, MR spectroscopy, perfusion imaging, grade (in the acute setting) and follow-up (in the subacute and PET/SPECT, and MEG, are of particular interest for pa- chronic settings) blunt neurovascular injuries, providing signifi- tients with mild TBI when conventional NCCT and MR cant prognostic information and influencing management 161 imaging findings are negative. These advanced neuroimag- decisions. ing techniques have shown promising results in group com- The application of duplex ultrasonography, MR imaging/an- parison analyses. giography, and transesophageal echocardiography has been de- 2) At the time of writing, there is insufficient evidence supporting scribed, but these imaging modalities are usually more difficult the routine clinical use of advanced neuroimaging for diagno- to obtain in the acute setting in patients involved in severe sis and/or prognostication at the individual patient level (class trauma (evidence level III) due to availability of techniques IIb recommendation). and patient contraindications. For the same reason, 4-vessel DSA is only used in the acute setting in patients with inconclu- Imaging of Neurovascular Trauma sive CTA interpretation or when an endovascular intervention Traumatic neurovascular injuries are a rare but known occur- is considered.150,160 rence associated with certain types of to the face and The description above pertained to blunt neurovascular inju- head. They occur in approximately 0.1%–2% of patients with ries but also applies to penetrating neck injury. A comprehensive severe trauma, with carotid and vertebral arteries being equally physical examination, combined with CTA of the neck extended 147,148 affected. to include the circle of Willis, is adequate for to effectively The natural history of these lesions is still being elucidated and identify or exclude vascular and aerodigestive injury after penetrating optimal treatment plans formulated. A grading scheme, called the neck trauma (evidence level III), with endoscopy and angiography 149 Denver grading scale, has been proposed by Biffl et al : serving as second-line evaluation modalities (class IIa recommenda- Ͻ Grade I: intimal irregularity with 25% narrowing tion).162 Traumatic neurovascular injuries are treated medically and, Grade II: dissection with intramural hematoma and Ͼ25% when needed, by endovascular approaches (coil embolization, stent luminal narrowing placement), with a low rate of immediate and delayed neurovascular Grade III: pseudoaneurysm complications (evidence level II).163,164 Grade IV: occlusion Grade V: transection with extravasation, as well as arterio- Summary venous fistula. Undetected, traumatic neurovascular injuries can cause isch- 1) Traumatic neurovascular injuries should be suspected in pa- emic stroke and are associated with high mortality rates. Imple- tients with a high-velocity mechanism, low Glasgow Coma mentation of the screening protocol can virtually eliminate in- Scale score, high , AND mandible fracture, jury-related strokes in patients without primary thromboembolic complex skull fractures, basilar skull fractures (including carotid neurologic deficits.150 canal fractures), scalp degloving, any type of cervical spine injury, Several other injuries and injury patterns can be used to iden- tify patients with a high likelihood of concurrent traumatic neu- and/or TBI with thoracic injuries, and/or thoracic vascular imag- rovascular injuries, and these patterns can be used as indications ing, as well as in patients with penetrating neck injury (class I to screen for traumatic neurovascular injuries. Patients with blunt recommendation). trauma with a high-velocity mechanism, low Glasgow Coma Scale 2) CTA of the neck extended to include the circle of Willis should score, high injury severity score, mandible fracture, complex skull be the first-line investigation for all patients with suspected fractures, basilar skull fractures (including carotid canal frac- vascular trauma and who have no indication for immediate tures), scalp degloving, any type of cervical spine injury, traumatic operative intervention (class IIa recommendation). brain injury with thoracic injuries, and/or thoracic vascular im- 3) A relatively high proportion of traumatic neurovascular inju- aging are at increased risk for blunt neurovascular injuries (evi- ries are treated by endovascular approaches (coil emboliza- dence level II).148,151-154 In patients identified as having these tion, stent placement), with a low rate of immediate and de- types of concurrent injuries, the overall rate of blunt neurovascu- layed neurovascular complications. lar injuries increases up to 27%–30%.147 Although digital subtraction angiography is the diagnostic ref- erence standard for detecting blunt neurovascular injuries, a CONCLUSIONS number of retrospective studies and meta-analyses indicate that A number of advanced neuroimaging techniques are currently multidetector CT angiography is an accurate, rapid, noninvasive under investigation in an attempt to optimize them and sub- diagnostic alternative. CTA of the neck extended to include the stantiate their clinical relevance in individual patients; how- circle of Willis has a sensitivity and specificity of 90%–100% for ever, the data currently available confine their use to the re- diagnosing vascular trauma in blunt vascular injury within the search arena for group comparisons. TBI imaging is a rapidly neck (evidence level II).155-159 Based on the evidence available, evolving field, and a number of the recommendations pre- CTA of the neck extended to include the circle of Willis should be sented will be updated in the future to reflect the advances in the first-line investigation for all patients with suspected vascular medical knowledge. E6 Wintermark Feb 2015 www.ajnr.org APPENDIX David W. Wright, MD, FACEP, Department of Emergency American College of Radiology Head Injury Institute Medicine, Emory University School of Medicine, Atlanta, Significant contributors: T. Jason Druzgal, MD, PhD, Depart- Georgia; Robert D. Zimmerman, MD, Department of Radiol- ment of Radiology, University of Virginia, Charlottesville, Vir- ogy, Weill Cornell Medical College, New York-Presbyterian ginia; Alisa D. Gean, MD, Department of Radiology, University of Hospital, New York, New York. California, and Brain and Spinal Injury Center, San Francisco General Hospital, San Francisco, California; Yvonne W. Lui, MD, ACKNOWLEDGMENTS Department of Radiology, New York University School of Medi- We would like to thank Laura Coombs and Barbara Seedah from cine, New York, New York; Alexander M. Norbash, MD, Depart- the American College of Radiology for the support they provided ment of Radiology, Boston University School of Medicine, Bos- in the preparation of this article. ton, Massachusetts; Cyrus Raji, MD, Department of Radiology, University of California, Los Angeles, Los Angeles, California; Da- Disclosures: Max Wintermark—UNRELATED: Grants/Grants Pending: GE Health- vid W. Wright, MD, FACEP, Department of Emergency Medi- care,* Philips Healthcare.* Apostolos J. Tsiouris—UNRELATED: Consultancy: Bio- Clinica, Parexel, Comments: BioClinica and Parexel are contract research organiza- cine, Emory University School of Medicine, Atlanta, Georgia; Mi- tions for which I perform blinded expert imaging interpretations and reports for chael Zeineh, MD, Department of Radiology, Stanford oncology pharmaceutical trials; Expert Testimony: I perform expert imaging reviews University, Palo Alto, California. for medicolegal cases (approximately 5–10 per year). *Money paid to the institution. This article was reviewed and approved by: Joseph M. Aulino, MD, Department of Radiology, Vanderbilt University, Nashville, REFERENCES Tennessee; Jeffrey T. Barth, PhD, Department of Psychiatry and 1. Faul M, Xu L, Wald M, et al. Traumatic Brain Injury in the United Neurobehavioral Sciences, University of Virginia, Charlottesville, States: Emergency Department Visits, Hospitalizations and Deaths Virginia; Michael Brant-Zawadzki, MD, FACR, Hoag Hospital, 2002–2006. Atlanta: Centers for Disease Control and Prevention, National Center for Injury Prevention and Control; 2010 Newport Beach, California; Asim F. Choudhri, MD, Department 2. Marin JR, Weaver MD, Yealy DM, et al. Trends in visits for trau- of Radiology, University of Tennessee Health Science Center, matic brain injury to emergency departments in the United States. Memphis, Tennessee; Amanda Corey, MD, Department of Radi- JAMA 2014;311:1917–19 ology, Emory University School of Medicine, Atlanta, Georgia; 3. Ilie G, Boak A, Adlaf EM, et al. Prevalence and correlates of trau- Rebecca S. Cornelius, MD, FACR, Departments of Radiology and matic brain injuries among adolescents. JAMA 2013;309:2550–52 4. Gean AD. Brain Injury: Applications from War and Terrorism. Lip- Otolaryngology, University of Cincinnati Medical Center, Cin- pincott Williams & Wilkins; 2014 cinnati, Ohio; Cameron Craddock, PhD, Center for the Develop- 5. National Center for Injury Prevention and Control. Report to Con- ing Brain, Child Mind Institute, New York, New York; Burton P. gress on Mild Traumatic Brain Injury in the United States: Steps to Drayer, MD, Department of Radiology, Mount Sinai Hospital, Prevent a Serious Public Health Problem. Atlanta: Centers for Disease Control and Prevention; 2003 New York, New York; Steven Flanagan, MD, Department of Re- 6. Centers for Disease Control and Prevention. Injury Prevention and habilitation Medicine, New York University Langone Medical Control: Traumatic Brain Injury. http://www.cdc.gov/traumatic Center, New York, New York; Magali Haas, MD, PhD, Orion braininjury/. Updated March 6, 2014. Accessed October 30, 2014 Bionetworks; Joseph A. Helpern, PhD, Department of Radiology, 7. Congressional Budget Office. The Veterans Health Administration’s Medical University of South Carolina, Charleston, South Caro- Treatment of PTSD and Traumatic Brain Injury Among Recent Combat Veterans. http://www.cbo.gov/sites/default/files/cbofiles/ lina; Ramona Hicks, PhD, Program Director, Repair and Plastic- attachments/02-09-PTSD.pdf. February 2012. Accessed October 20, ity, National Institutes of Health/National Institute of Neurolog- 2014 ical Disorders and Stroke, Bethesda, Maryland; Randy Horton, 8. Basser PJ, Mattiello J, LeBihan D. Estimation of the effective self- American College of Radiology Head Injury Institute, Reston, diffusion tensor from the NMR spin echo. J Magn Reson B Virginia; Patricia Hudgins, MD, Department of Radiology, Em- 1994;103:247–54 9. Davenport ND, Lim KO, Armstrong MT, et al. Diffuse and spatially ory University School of Medicine, Atlanta, Georgia; Carolyn Ci- variable white matter disruptions are associated with blast-re- dis Meltzer, MD, FACR, Department of Radiology and Imaging lated mild traumatic brain injury. Neuroimage 2012;59:2017–24 Sciences, Emory University School of Medicine, Atlanta, Georgia; 10. Mayer AR, Ling JM, Yang Z, et al. Diffusion abnormalities in pedi- David Mikulis, MD, PhD, Department of Radiology, University atric mild traumatic brain injury. J Neurosci 2012;32:17961–69 of Toronto, Toronto, Ontario, Canada; Sumit N. Niogi, MD, 11. Ling JM, Pen˜a A, Yeo RA, et al. Biomarkers of increased diffusion anisotropy in semi-acute mild traumatic brain injury: a longitudi- PhD, Department of Radiology, New York-Presbyterian Hospi- nal perspective. Brain 2012;135:1281–92 tal, Weill Cornell Medical Center, New York, New York; Michael 12. Wilde EA, McCauley SR, Hunter JV, et al. Diffusion tensor imaging Rothman, MD, Bethlehem, Pennsylvania; David Seidenwurm, of acute mild traumatic brain injury in adolescents. Neurology MD, Department of Radiology, Sutter Health, Sacramento, Cali- 2008;70:948–55 13. Chu Z, Wilde EA, Hunter JV, et al. Voxel-based analysis of diffu- fornia; Martha Shenton, PhD, Department of Psychiatry, sion tensor imaging in mild traumatic brain injury in adolescents. Brigham and Women’s Hospital, Harvard Medical School, AJNR Am J Neuroradiol 2010;31:340–46 Boston, Massachusetts; James R. Stone, MD, PhD, Department 14. Mayer AR, Ling J, Mannell MV, et al. A prospective diffusion tensor of Radiology, University of Virginia, Charlottesville, Virginia; imaging study in mild traumatic brain injury. Neurology Gordon Sze, MD, Department of Radiology, Yale University, 2010;74:643–50 15. Smith SM, Jenkinson M, Johansen-Berg H, et al. Tract-based spatial New Haven, Connecticut; Elizabeth A. Wilde, PhD, Depart- statistics: voxelwise analysis of multi-subject diffusion data. Neu- ments of Physical Medicine and Rehabilitation, Neurology, roimage 2006;31:1487–505 and Radiology, Baylor College of Medicine, Houston, Texas; 16. Bouix S, Pasternak O, Rathi Y, et al. Increased gray matter diffusion AJNR Am J Neuroradiol 36:E1–E11 Feb 2015 www.ajnr.org E7 anisotropy in patients with persistent post-concussive symptoms changes in corpus callosum white matter indices. Ann Biomed Eng following mild traumatic brain injury. PLoS One 2013;8:e66205 2012;40:127–40 17. Tuch DS, Reese TG, Wiegell MR, et al. High angular resolution 38. Henry LC, Tremblay J, Tremblay S, et al. Acute and chronic changes diffusion imaging reveals intravoxel white matter fiber heteroge- in diffusivity measures after sports concussion. J Neurotrauma neity. Magn Reson Med 2002;48:577–82 2011;28:2049–59 18. Jensen JH, Helpern JA. MRI quantification of non-Gaussian water 39. Bazarian JJ, Zhong J, Blyth B, et al. Diffusion tensor imaging detects diffusion by kurtosis analysis. NMR Biomed 2010;23:698–710 clinically important axonal damage after mild traumatic brain 19. Fieremans E, Benitez A, Jensen JH, et al. Novel white matter tract injury: a pilot study. J Neurotrauma 2007;24:1447–59 integrity metrics sensitive to Alzheimer disease progression. AJNR 40. Bazarian JJ, Zhu T, Blyth B, et al. Subject-specific changes in brain Am J Neuroradiol 2013;34:2105–12 white matter on diffusion tensor imaging after sports-related con- 20. Jensen JH, Helpern JA, Tabesh A. Leading non-Gaussian correc- cussion. Magn Reson Imaging 2012;30:171–80 tions for diffusion orientation distribution function. NMR 41. Eierud C, Craddock RC, Fletcher S, et al. Neuroimaging after mild Biomed 2014;27:202–11 traumatic brain injury: review and meta-analysis. Neuroimage Clin 21. Wedeen VJ, Hagmann P, Tseng WY, et al. Mapping complex tissue 2014;4:283–94 architecture with diffusion spectrum magnetic resonance imag- 42. Hulkower M, Poliak D, Rosenbaum S, et al. A decade of DTI in ing. Magn Reson Med 2005;54:1377–86 traumatic brain injury: 10 years and 100 articles later. AJNR Am J 22. Tuch DS. Q-ball imaging. Magnet Reson Med 2004;52:1358–72 Neuroradiol 2013;34:2064–74 23. Feinberg DA, Moeller S, Smith SM, et al. Multiplexed echo planar 43. Fakhran S, Yaeger K, Alhilali L. Symptomatic white matter changes imaging for sub-second whole brain FMRI and fast diffusion im- in mild traumatic brain injury resemble pathologic features of aging. PLoS One 2010;5:e15710 early Alzheimer dementia. Radiology 2013;269:249–57 24. Zhang H, Schneider T, Wheeler-Kingshott CA, et al. NODDI: prac- 44. Shenton M, Hamoda H, Schneiderman J, et al. A review of magnetic tical in vivo neurite orientation dispersion and density imaging of resonance imaging and diffusion tensor imaging findings in mild the human brain. Neuroimage 2012;61:1000–16 traumatic brain injury. Brain Imaging Behav 2012;6:137–92 25. Tournier J, Calamante F, Connelly A. Robust determination of the 45. Niogi SN, Mukherjee P. Diffusion tensor imaging of mild trau- fibre orientation distribution in diffusion MRI: non-negativity matic brain injury. J Head Trauma Rehabil 2010;25:241–55 constrained super-resolved spherical deconvolution. Neuroimage 46. Logothetis NK, Pauls J, Augath M, et al. Neurophysiological inves- 2007;35:1459–72 tigation of the basis of the fMRI signal. Nature 2001;412:150–57 26. Xu D, Maier JK, King KF, et al. Prospective and retrospective high 47. Heeger DJ, Ress D. What does fMRI tell us about neuronal activity? order eddy current mitigation for diffusion weighted echo planar Nat Rev Neurosci 2002;3:142–51 imaging. Magnet Reson Med 2013;70:1293–305 48. Arthurs OJ, Boniface S. How well do we understand the neural 27. Wilde EA, Ramos MA, Yallampalli R, et al. Diffusion tensor imag- origins of the fMRI BOLD signal? Trends Neurosci 2002;25:27–31 ing of the cingulum bundle in children after traumatic brain in- 49. Attwell D, Iadecola C. The neural basis of functional brain imaging jury. Dev Neuropsychol 2010;35:333–51 signals. Trends Neurosci 2002;25:621–25 28. Arfanakis K, Haughton VM, Carew JD, et al. Diffusion tensor MR 50. Nair DG. About being BOLD. Brain Res Brain Res Rev imaging in diffuse axonal injury. AJNR Am J Neuroradiol 2005;50:229–43 2002;23:794–802 51. Jantzen KJ. Functional magnetic resonance imaging of mild trau- 29. Kumar R, Gupta RK, Husain M, et al. Comparative evaluation of matic brain injury. J Head Trauma Rehab 2010;25:256–66 corpus callosum DTI metrics in acute mild and moderate trau- 52. Freeman JR, Barth JT, Broshek DK, et al. Sports injuries. In: Silver matic brain injury: its correlation with neuropsychometric tests. JM, McAllister TW, Yudofsky SC, eds. Textbook of Traumatic Brain Brain Inj 2009;23:675–85 Injury. Washington, DC: American Psychiatric Publishing Inc; 30. Miles L, Grossman RI, Johnson G, et al. Short-term DTI predictors 2005:453–76 of cognitive dysfunction in mild traumatic brain injury. Brain Inj 53. Bullock R, Zauner A, Woodward JJ, et al. Factors affecting excit- 2008;22:115–22 atory amino acid release following severe human head injury. 31. Newcombe VF, Williams GB, Nortje J, et al. Concordant biology J Neurosurg 1998;89:507–18 underlies discordant imaging findings: diffusivity behaves differ- 54. Garnett MR, Blamire AM, Rajagopalan B, et al. Evidence for cellular ently in grey and white matter post acute neurotrauma. Acta Neu- damage in normal-appearing white matter correlates with injury rochir Suppl 2008;102:247–51 severity in patients following traumatic brain injury: a magnetic 32. Newcombe VF, Williams GB, Nortje J, et al. Analysis of acute trau- resonance spectroscopy study. Brain 2000;123(pt 7):1403–09 matic axonal injury using diffusion tensor imaging. Br J Neurosurg 55. Hattingen E, Raab P, Franz K, et al. Prognostic value of choline and 2007;21:340–48 creatine in WHO grade II gliomas. Neuroradiology 2008;50:759–67 33. Wozniak JR, Lim KO. Advances in white matter imaging: a review 56. Inglese M, Li BS, Rusinek H, et al. Diffusely elevated cerebral cho- of in vivo magnetic resonance methodologies and their applicabil- line and creatine in relapsing-remitting multiple sclerosis. Magn ity to the study of development and aging. Neurosci Biobehav Rev Reson Med 2003;50:190–95 2006;30:762–74 57. Mun˜oz Maniega S, Cvoro V, Armitage PA, et al. Choline and crea- 34. Wozniak JR, Krach L, Ward E, et al. Neurocognitive and neuroim- tine are not reliable denominators for calculating metabolite ra- aging correlates of pediatric traumatic brain injury: a diffusion tios in acute ischemic stroke. Stroke 2008;39:2467–69 tensor imaging (DTI) study. Arch Clin Neuropsychol 58. Gasparovic C, Yeo R, Mannell M, et al. Neurometabolite concen- 2007;22:555–68 trations in gray and white matter in mild traumatic brain injury: 35. Aoki Y, Inokuchi R, Gunshin M, et al. Diffusion tensor imaging an 1H-magnetic resonance spectroscopy study. J Neurotrauma studies of mild traumatic brain injury: a meta-analysis. J Neurol 2009;26:1635–43 Neurosurg Psychiatry 2012;83:870–76 59. Danielsen ER, Michaelis T, Ross BD. Three methods of calibration 36. Brandstack N, Kurki T, Tenovuo O. Quantitative diffusion-tensor in quantitative proton MR spectroscopy. J Magn Reson B tractography of long association tracts in patients with traumatic 1995;106:287–91 brain injury without associated findings at routine MR imaging. 60. Jansen JF, Backes WH, Nicolay K, et al. 1H MR spectroscopy of the Radiology 2013;267:231–39 brain: absolute quantification of metabolites. Radiology 37. McAllister TW, Ford JC, Ji S, et al. Maximum principal strain and 2006;240:318–32 strain rate associated with concussion diagnosis correlates with 61. Provencher SW. Estimation of metabolite concentrations from lo- E8 Wintermark Feb 2015 www.ajnr.org calized in vivo proton NMR spectra. Magn Reson Med 82. Ashwal S, Holshouser B, Tong K, et al. Proton spectroscopy de- 1993;30:672–79 tected myoinositol in children with traumatic brain injury. Pediatr 62. Holshouser BA, Ashwal S, Luh GY, et al. Proton MR spectroscopy Res 2004;56:630–38 after acute central nervous system injury: outcome prediction in 83. Yeo RA, Gasparovic C, Merideth F, et al. A longitudinal proton neonates, infants, and children. Radiology 1997;202:487–96 magnetic resonance spectroscopy study of mild traumatic brain 63. Horska´ A, Kaufmann WE, Brant LJ, et al. In vivo quantitative pro- injury. J Neurotrauma 2011;28:1–11 ton MRSI study of brain development from childhood to adoles- 84. Brooks WM, Stidley CA, Petropoulos H, et al. Metabolic and cog- cence. J Magn Reson Imaging 2002;15:137–43 nitive response to human traumatic brain injury: a quantitative 64. Kreis R, Ernst T, Ross BD. Development of the human brain: in- proton magnetic resonance study. J Neurotrauma 2000;17:629–40 vivo quantification of metabolite and water-content with proton 85. Friedman SD, Brooks WM, Jung RE, et al. Quantitative proton magnetic-resonance spectroscopy. Magnet Reson Med 1993;30: MRS predicts outcome after traumatic brain injury. Neurology 424–37 1999;52:1384–91 65. Kreis R, Hofmann L, Kuhlmann B, et al. Brain metabolite compo- 86. Friedman SD, Brooks WM, Jung RE, et al. Proton MR spectroscopic sition during early human brain development as measured by findings correspond to neuropsychological function in traumatic quantitative in vivo H-1 magnetic resonance spectroscopy. Mag- brain injury. AJNR Am J Neuroradiol 1998;19:1879–85 net Reson Med 2002;48:949–58 87. Aaen GS, Holshouser BA, Sheridan C, et al. Magnetic resonance 66. Pouwels PJW, Brockmann K, Kruse B, et al. Regional age depen- spectroscopy predicts outcomes for children with nonaccidental dence of human brain metabolites from infancy to adulthood as trauma. Pediatrics 2010;125:295–303 detected by quantitative localized proton MRS. Pediatr Res 88. Ashwal S, Holshouser BA, Shu SK, et al. Predictive value of proton 1999;46:474–85 magnetic resonance spectroscopy in pediatric closed head injury. 67. Ross BD, Ernst T, Kreis R, et al. 1H MRS in acute traumatic brain Pediatr Neurol 2000;23:114–25 injury. J Magn Reson Imaging 1998;8:829–40 89. Babikian T, Freier MC, Ashwal S, et al. MR spectroscopy: predicting 68. Shutter L, Tong KA, Holshouser BA. Proton MRS in acute trau- long-term neuropsychological outcome following pediatric TBI. J matic brain injury: role for glutamate/glutamine and choline for Magn Reson Imaging 2006;24:801–11 outcome prediction. J Neurotrauma 2004;21:1693–705 90. Brenner T, Freier MC, Holshouser BA, et al. Predicting neuropsy- 69. Cohen BA, Inglese M, Rusinek H, et al. Proton MR spectroscopy chologic outcome after traumatic brain injury in children. Pediatr and MRI-volumetry in mild traumatic brain injury. AJNR Am J Neurol 2003;28:104–14 Neuroradiol 2007;28:907–13 91. Holshouser BA, Tong KA, Ashwal S. Proton MR spectroscopic im- 70. Govindaraju V, Gauger GE, Manley GT, et al. Volumetric proton aging depicts diffuse axonal injury in children with traumatic spectroscopic imaging of mild traumatic brain injury. AJNR Am J brain injury. AJNR Am J Neuroradiol 2005;26:1276–85 Neuroradiol 2004;25:730–37 92. Hunter JV, Thornton RJ, Wang ZJ, et al. Late proton MR spectros- 71. Henry LC, Tremblay S, Boulanger Y, et al. Neurometabolic changes copy in children after traumatic brain injury: correlation with in the acute phase after sports concussions correlate with symp- cognitive outcomes. AJNR Am J Neuroradiol 2005;26:482–88 tom severity. J Neurotrauma 2010;27:65–76 93. Makoroff KL, Cecil KM, Care M, et al. Elevated lactate as an early 72. Henry LC, Tremblay S, Leclerc S, et al. Metabolic changes in con- marker of brain injury in inflicted traumatic brain injury. Pediatr cussed American football players during the acute and chronic Radiol 2005;35:668–76 post-injury phases. BMC Neurol 2011;11:105 94. Yeo RA, Phillips JP, Jung RE, et al. Magnetic resonance spectros- 73. Sarmento E, Moreira P, Brito C, et al. Proton spectroscopy in pa- copy detects brain injury and predicts cognitive functioning in tients with post-traumatic headache attributed to mild head in- children with brain injuries. J Neurotrauma 2006;23:1427–35 jury. Headache 2009;49:1345–52 95. Holshouser BA, Tong KA, Ashwal S, et al. Prospective longitudinal 74. Vagnozzi R, Signoretti S, Cristofori L, et al. Assessment of meta- proton magnetic resonance spectroscopic imaging in adult trau- bolic brain damage and recovery following mild traumatic brain matic brain injury. J Magn Reson Imaging 2006;24:33–40 injury: a multicentre, proton magnetic resonance spectroscopic 96. Garnett MR, Corkill RG, Blamire AM, et al. Altered cellular metab- study in concussed patients. Brain 2010;133:3232–42 olism following traumatic brain injury: a magnetic resonance 75. Vagnozzi R, Signoretti S, Tavazzi B, et al. Temporal window of met- spectroscopy study. J Neurotrauma 2001;18:231–40 abolic brain vulnerability to concussion: a pilot 1H-magnetic res- 97. Hari R, Forss N. Magnetoencephalography in the study of human onance spectroscopic study in concussed athletes—part III. Neu- somatosensory cortical processing. Philos Trans R Soc Lond B Biol rosurgery 2008;62:1286–95; discussion 1295–96 Sci 1999;354:1145–54 76. Govind V, Gold S, Kaliannan K, et al. Whole-brain proton MR 98. Hari R, Salmelin R. Magnetoencephalography: from SQUIDs to spectroscopic imaging of mild-to-moderate traumatic brain in- neuroscience—Neuroimage 20th anniversary special edition. jury and correlation with neuropsychological deficits. J Neu- Neuroimage 2012;61:386–96 rotrauma 2010;27:483–96 99. Hamalainen M, Hari R, Ilmoniemi RJ, et al. Magnetoencephalog- 77. Kirov I, Fleysher L, Babb JS, et al. Characterizing ‘mild’ in trau- raphy—theory, instrumentation, and applications to noninvasive matic brain injury with proton MR spectroscopy in the thalamus: studies of the working human brain. Rev Mod Phys 1993;65:413–97 initial findings. Brain Inj 2007;21:1147–54 100. Stam CJ. Use of magnetoencephalography (MEG) to study func- 78. Maugans TA, Farley C, Altaye M, et al. Pediatric sports-related con- tional brain networks in neurodegenerative disorders. J Neurol Sci cussion produces cerebral blood flow alterations. Pediatrics 2010;289:128–34 2012;129:28–37 101. Tormenti M, Krieger D, Puccio AM, et al. Magnetoencephalo- 79. George EO, Roys S, Sours C, et al. Longitudinal and prognostic graphic virtual recording: a novel diagnostic tool for concussion. evaluation of mild traumatic brain injury: a 1H-magnetic reso- Neurosurg Focus 2012;33:E9 nance spectroscopy study. J Neurotrauma 2014;31:1018–28 102. Huang M-X, Nichols S, Robb A, et al. An automatic MEG low- 80. Cecil KM, Hills EC, Sandel ME, et al. Proton magnetic resonance frequency source imaging approach for detecting injuries in mild spectroscopy for detection of axonal injury in the splenium of the and moderate TBI patients with blast and non-blast causes. Neu- corpus callosum of brain-injured patients. J Neurosurg roimage 2012;61:1067–82 1998;88:795–801 103. Huang MX, Theilmann RJ, Robb A, et al. Integrated imaging ap- 81. Cimatti M. Assessment of metabolic cerebral damage using pro- proach with MEG and DTI to detect mild traumatic brain injury in ton magnetic resonance spectroscopy in mild traumatic brain in- military and civilian patients. J Neurotrauma 2009;26:1213–26 jury. J Neurosurg Sci 2006;50:83–88 104. Tarapore PE, Findlay AM, Lahue SC, et al. Resting state magneto- AJNR Am J Neuroradiol 36:E1–E11 Feb 2015 www.ajnr.org E9 encephalography functional connectivity in traumatic brain in- 124. Kim J, Whyte J, Patel S, et al. Resting cerebral blood flow alterations jury. J Neurosurg 2013;118:1306–16 in chronic traumatic brain injury: an arterial spin labeling perfu- 105. Verweij BH, Muizelaar JP, Vinas FC. Hyperacute measurement of sion FMRI study. J Neurotrauma 2010;27:1399–411 intracranial pressure, cerebral perfusion pressure, jugular venous 125. Peters AM, Gunasekera RD, Henderson BL, et al. Noninvasive mea- oxygen saturation, and laser Doppler flowmetry, before and dur- surement of blood flow and extraction fraction. Nucl Med Com- ing removal of traumatic acute subdural hematoma. J Neurosurg mun 1987;8:823–37 2001;95:569–72 126. Wintermark M, Thiran JP, Maeder P, et al. Simultaneous measure- 106. Adelson PD, Clyde B, Kochanek PM, et al. Cerebrovascular re- ment of regional cerebral blood flow by perfusion CT and stable sponse in infants and young children following severe traumatic xenon CT: a validation study. AJNR Am J Neuroradiol 2001;22: brain injury: a preliminary report. Pediatr Neurosurg 1997;26: 905–14 200–07 127. Kudo K, Terae S, Katoh C, et al. Quantitative cerebral blood flow 107. Nedd K, Sfakianakis G, Ganz W, et al. 99mTc-HMPAO SPECT of measurement with dynamic perfusion CT using the vascular-pixel the brain in mild to moderate traumatic brain injury patients: elimination method: comparison with (H2O)-O-15 positron compared with CT: a prospective study. Brain Inj 1993;7:469–79 emission tomography. AJNR Am J Neuroradiol 2003;24:419–26 108. Kinuya K, Kakuda K, Nobata K, et al. Role of brain perfusion single- 128. Latchaw RE, Yonas H, Pentheny SL, et al. Adverse reactions to xe- photon emission tomography in traumatic head injury. Nucl Med non-enhanced CT cerebral blood flow determination. Radiology Commun 2004;25:333–37 1987;163:251–54 109. Gowda NK, Agrawal D, Bal C, et al. Technetium Tc-99m ethyl cys- 129. Wintermark M, Reichhart M, Thiran JP, et al. Prognostic accuracy teinate dimer brain single-photon emission CT in mild traumatic of cerebral blood flow measurement by perfusion computed to- brain injury: a prospective study. AJNR Am J Neuroradiol mography, at the time of emergency room admission, in acute 2006;27:447–51 stroke patients. Ann Neurol 2002;51:417–32 110. Pavel D, Jobe T, Devore-Best S, et al. Viewing the functional conse- 130. Wintermark M, Reichhart M, Cuisenaire O, et al. Comparison of quences of traumatic brain injury by using brain SPECT. Brain admission perfusion computed tomography and qualitative diffu- Cogn 2006;60:211–13 sion- and perfusion-weighted magnetic resonance imaging in 111. Shin YB, Kim SJ, Kim IJ, et al. Voxel-based statistical analysis of acute stroke patients. Stroke 2002;33:2025–31 cerebral blood flow using Tc-99m ECD brain SPECT in patients 131. Amorim RL, Bor-Seng-Shu E, Gatta´s GS, et al. Decompressive with traumatic brain injury: group and individual analyses. Brain craniectomy and cerebral blood flow regulation in head injured Inj 2006;20:661–67 patients: a case studied by perfusion CT. J Neuroradiol 112. Uruma G, Hashimoto K, Abo M. A new method for evaluation of 2012;39:346–49 mild traumatic brain injury with neuropsychological impairment 132. Metting Z, Cerliani L, Rodiger LA, et al. Pathophysiological con- using statistical imaging analysis for Tc-ECD SPECT. Ann Nucl cepts in mild traumatic brain injury: diffusion tensor imaging re- Med 2013;27:187–202 lated to acute perfusion CT imaging. PLoS One 2013;8:e64461 113. Bavetta S, Nimmon CC, White J, et al. A prospective study compar- 133. Grossman EJ, Jensen JH, Babb JS, et al. Cognitive impairment in ing SPET with MRI and CT as prognostic indicators following se- mild traumatic brain injury: a longitudinal diffusional kurtosis vere closed head injury. Nucl Med Commun 1994;15:961–68 and perfusion imaging study. AJNR Am J Neuroradiol 2013;34:951– 114. Abu-Judeh HH, Parker R, Singh M, et al. SPET brain perfusion 57, S1–3 imaging in mild traumatic brain injury without loss of conscious- 134. Stamatakis EA, Wilson JTL, Hadley DM, et al. SPECT imaging in ness and normal computed tomography. Nucl Med Commun head injury interpreted with statistical parametric mapping. 1999;20:505–10 J Nucl Med 2002;43:476–83 115. Yamaki T, Imahori Y, Ohmori Y, et al. Cerebral hemodynamics and 135. Lorberboym M, Lampl Y, Gerzon I, et al. Brain SPECT evaluation of metabolism of severe diffuse brain injury measured by PET. J Nucl amnestic ED patients after mild head trauma. Am J Emerg Med Med 1996;37:1166–70 2002;20:310–13 116. Wintermark M, van Melle G, Schnyder P, et al. Admission perfu- 136. Lewine JD, Davis JT, Bigler ED, et al. Objective documentation of sion CT: prognostic value in patients with severe head trauma. traumatic brain injury subsequent to mild head trauma: multi- Radiology 2004;232:211–20 modal brain imaging with MEG, SPECT, and MRI. J Head Trauma 117. Wintermark M, Chiolero R, van Melle G, et al. Relationship be- Rehabil 2007;22:141–55 tween brain perfusion computed tomography variables and cere- 137. Chen SHA, Kareken DA, Fastenau PS, et al. A study of persistent bral perfusion pressure in severe head trauma patients. Crit Care post-concussion symptoms in mild head trauma using positron Med 2004;32:1579–87 emission tomography. J Neurol Neurosurg Psychiatry 118. Wintermark M, Chiolero R, Van Melle G, et al. Cerebral vascular 2003;74:326–32 autoregulation assessed by perfusion-CT in severe head trauma 138. Peskind ER, Petrie EC, Cross DJ, et al. Cerebrocerebellar hypome- patients. J Neuroradiol 2006;33:27–37 tabolism associated with repetitive blast exposure mild traumatic 119. Metting Z, Rodiger LA, de Jong BM, et al. Acute cerebral perfusion brain injury in 12 Iraq war veterans with persistent post-concus- CT abnormalities associated with posttraumatic amnesia in mild sive symptoms. Neuroimage 2011;54(suppl 1):S76–82 head injury. J Neurotrauma 2010;27:2183–89 139. Byrnes KR, Wilson CM, Brabazon F, et al. FDG-PET imaging in 120. Soustiel JF, Mahamid E, Goldsher D, et al. Perfusion-CT for early mild traumatic brain injury: a critical review. Front Neuroenergetics assessment of traumatic cerebral contusions. Neuroradiology 2014;5:13 2008;50:189–96 140. Raji CA, Tarzwell R, Pavel D, et al. Clinical utility of SPECT neuro- 121. Garnett MR, Blamire AM, Corkill RG, et al. Abnormal cerebral imaging in the diagnosis and treatment of traumatic brain injury: blood volume in regions of contused and normal appearing brain a systematic review. PLoS One 2014;9:e91088 following traumatic brain injury using perfusion magnetic reso- 141. Bigler ED. Neuroimaging biomarkers in mild traumatic brain in- nance imaging. J Neurotrauma 2001;18:585–93 jury (mTBI). Neuropsychol Rev 2013;23:169–209 122. Liu W, Wang B, Wolfowitz R, et al. Perfusion deficits in patients 142. Hong YT, Veenith T, Dewar D, et al. Amyloid imaging with carbon with mild traumatic brain injury characterized by dynamic sus- 11-labeled Pittsburgh compound B for traumatic brain injury. ceptibility contrast MRI. NMR Biomed 2013;26:651–63 JAMA Neurol 2014;71:23–31 123. Ge YL, Patel MB, Chen Q, et al. Assessment of thalamic perfusion 143. Small GW, Kepe V, Siddarth P, et al. PET scanning of brain tau in in patients with mild traumatic brain injury by true FISP arterial retired National Football League players: preliminary findings. spin labelling MR imaging at 3T. Brain Inj 2009;23:666–74 Am J Geriatr Psychiatry 2013;21:138–44 E10 Wintermark Feb 2015 www.ajnr.org 144. Stein TD, Alvarez VE, McKee AC. Chronic traumatic encephalopathy: analysis of risk factors for blunt cerebrovascular injury. J Vasc Surg a spectrum of neuropathological changes following repetitive 2010;51:57–64 brain trauma in athletes and military personnel. Alzheimers Res 155. Chokshi FH, Munera F, Rivas LA, et al. 64-MDCT angiography of Ther 2014;6:4 blunt vascular injuries of the neck. AJR Am J Roentgenol 145. Gavett BE, Stern RA, McKee AC. Chronic traumatic encephalopathy: 2011;196:W309–15 a potential late effect of sport-related concussive and subconcus- 156. Desai NK, Kang J, Chokshi FH. Screening CT angiography for pe- sive head trauma. Clin Sports Med 2011;30:179–88, xi diatric blunt cerebrovascular injury with emphasis on the cervical 146. Shively S, Scher AI, Perl DP, et al. Dementia resulting from trau- “seatbelt sign.” AJNR Am J Neuroradiol 2014;35:1836–40 matic brain injury: what is the ? Arch Neurol 2012; 157. Dhillon RS, Barrios C, Lau C, et al. Seatbelt sign as an indication for 69:1245–51 four-vessel computed tomography angiogram of the neck to diag- 147. McKinney A, Ott F, Short J, et al. Angiographic frequency of blunt nose blunt carotid artery and other cervical vascular injuries. Am cerebrovascular injury in patients with carotid canal or vertebral Surg 2013;79:1001–04 foramen fractures on multidetector CT. Eur J Radiol 2007;62: 158. Fleck SK, Langner S, Baldauf J, et al. Blunt craniocervical artery injury in cervical spine lesions: the value of CT angiography. Acta 385–93 Neurochir (Wien) 2010;152:1679–86 148. Franz RW, Willette PA, Wood MJ, et al. A systematic review and 159. Liang T, McLaughlin PD, Louis L, et al. Review of multidetector meta-analysis of diagnostic screening criteria for blunt cerebro- computed tomography angiography as a screening modality in the vascular injuries. J Am Coll Surg 2012;214:313–27 assessment of blunt vascular neck injuries. Can Assoc Radiol J 149. Biffl WL, Moore EE, Offner PJ, et al. Blunt carotid arterial injuries: 2013;64:130–39 implications of a new grading scale. J Trauma 1999;47:845–53 160. Patterson BO, Holt PJ, Cleanthis M, et al. Imaging vascular trauma. 150. Lo¨hrer L, Vieth V, Nassenstein I, et al. Blunt cerebrovascular inju- Br J Surg 2012;99:494–505 ries in acute trauma care: a screening protocol. Eur Spine J 161. Sliker CW. Blunt cerebrovascular injuries: imaging with multide- 2012;21:837–43 tector CT angiography. Radiographics 2008;28:1689–708; discus- 151. Cothren CC, Moore EE, Ray CE Jr, et al. Cervical spine fracture sion 1709–10 patterns mandating screening to rule out blunt cerebrovascular 162. Shiroff AM, Gale SC, Martin ND, et al. Penetrating neck trauma: a injury. Surgery 2007;141:76–82 review of management strategies and discussion of the ‘No Zone’ 152. Kopelman TR, Leeds S, Berardoni NE, et al. Incidence of blunt cere- approach. Am Surg 2013;79:23–29 brovascular injury in low-risk cervical spine fractures. Am J Surg 163. Kansagra AP, Cooke DL, English JD, et al. Current trends in endo- 2011;202:684–88; discussion 688–89 vascular management of traumatic cerebrovascular injury. J Neu- 153. Burlew CC, Biffl WL, Moore EE, et al. Blunt cerebrovascular rointerv Surg 2014;6:47–50 injuries: redefining screening criteria in the era of noninvasive di- 164. Hassan AE, Zacharatos H, Souslian F, et al. Long-term clinical and agnosis. J Trauma Acute Care Surg 2012;72:330–35; discussion 336– angiographic outcomes in patients with cervico-cranial dissec- 37, quiz 539 tions treated with stent placement: a meta-analysis of case series. 154. Berne JD, Cook A, Rowe SA, et al. A multivariate logistic regression J Neurotrauma 2012;29:1342–53

AJNR Am J Neuroradiol 36:E1–E11 Feb 2015 www.ajnr.org E11