Clinical Review

A guide to imaging for common neurological problems

BMJ 2010; 341 doi: (Published 16 August 2010) Cite this as: BMJ 2010;341:c4113
  1. Adam B Cohen, neurologist and fellow of neuroradiology1,
  2. Joshua P Klein, neurologist and fellow of neuroradiology1,
  3. Srinivasan Mukundan, neuroradiologist and section head of neuroradiology2
  1. 1Departments of Neurology and Radiology, Brigham and Women’s Hospital, Harvard Medical School, 45 Francis Street, Boston, MA 02115, USA
  2. 2Department of Radiology, Brigham and Women’s Hospital, Harvard Medical School
  1. Correspondence to: A B Cohen abcohen{at}

    Summary points

    • A careful clinical history, physical examination, and referral to evidence based guidelines when ordering imaging will reduce unnecessary studies

    • Although head computed tomography is less expensive, faster, and more readily available, magnetic resonance imaging is better at assessing most neurological conditions

    • Head computed tomography is recommended when urgent decisions are needed but has a low yield in transient neurological episodes

    • When vascular lesions are suspected, as in transient ischaemic attack, computed tomography angiography or magnetic resonance angiography of the head and neck is indicated

    • Treat most patients with low back and neck pain only conservatively; reserve imaging for those with red flag features and those who fail conservative treatment and are candidates for surgical intervention

    • Practitioners may need to assess the need for further testing and possible interventions in patients with incidental findings seen on neurological imaging

    Patients with headache, transient neurological episodes, symptoms after minor head trauma, and neck and low back pain often present to general practitioners and emergency room physicians. The examining doctor may be uncertain whether neurological imaging is needed. In this article, we discuss indications for imaging and tests that would be most useful in these scenarios. Table 1 summarises our approach to this problem. Figure 1 provides a walk through of representative images from patients with headache or minor head trauma, showing normal and abnormal findings. Figure 2 shows various examples of abnormal imaging findings relevant to the clinical scenarios described here.

    Occasionally, imaging leads to the incidental discovery of a lesion of unknown importance. We will also discuss the management and interpretation of such findings.


    Fig 1 (A) In acute headache or minor head trauma, head computed tomography may be recommended. All images are oriented axially. A normal head computed tomography is shown. The brain structures are evaluated for symmetry and size, including (i) the grey and white matter differentiation (single arrow), (ii) the ventricles (double arrows), (iii) sulcuses (asterisk), and (iv) additional cerebrospinal fluid spaces over the brain surface and basal regions (open arrow). (B) Pathology is detected by computed tomography by this approach: (i) loss of grey-white matter differentiation in herpes simplex virus encephalitis (single arrow), (ii) slightly compressed right lateral ventricle (double arrow) secondary to adjacent subdural haemorrhage (open circle), (iii) subarachnoid haemorrhage within sulcuses (asterisk), and (iv) subarachnoid haemorrhage within a basal cerebrospinal fluid space, anterior to the midbrain (double open arrow) along with intraparenchymal haemorrhage (open arrowhead) and minimal adjacent extra-axial haemorrhage. (C) Head magnetic resonance imaging with gadolinium is best approached with the same general concepts as computed tomography, but with special attention to high yield sequences; a normal magnetic resonance imaging is shown: (i) fluid attenuation inversion recovery sequences serve as a good screening sequence for many pathologies; (ii) many infectious, inflammatory, and neoplastic pathologies are bright on post-gadolinium T1, but not on (iii) pre-gadolinium T1 sequence; and (iv) diffusion weighted imaging, which shows acute cytotoxic injury, such as in infarction and other destructive processes, including various encephalitides. (D) Pathology is detected by magnetic resonance imaging with gadolinium by this approach; an example of herpes simplex virus encephalitis is shown: (i) Fluid attenuation inversion recovery (FLAIR) sequence shows extensive grey and white matter destruction and oedema (arrows); (ii) post-gadolinium T1 sequences show enhancement of the adjacent meninges (double arrow), which is not well visualised on (iii) pre-gadolinium T1 sequences; and (iv) diffusion weighted imaging shows bilateral low diffusivity, consistent with cytotoxic injury (asterisks)


    Fig 2 Findings on neuroimaging of common neurological problems. All images are oriented axially unless indicated otherwise. (A) Head computed tomography shows traumatic epidural haematoma in the right temporal region (double arrow) as well as adjacent pneumocephalus (arrow), indicating temporal bone fracture. (B) Neck magnetic resonance angiography (oblique view) shows stenosis (arrow) of the proximal internal carotid artery and normal left internal carotid artery (double arrow). (C) Brain magnetic resonance diffusion weighted imaging shows a small right parietal infarct (arrow). (D) Brain magnetic resonance T1 weighted post-gadolinium imaging shows a right parafalcine meningioma (arrow). (E) Brain magnetic resonance coronal fluid attenuation inversion recovery (FLAIR) imaging shows left sided mesial temporal sclerosis (arrow). (F) Cervical spine computed tomography (sagittally oriented, soft tissue window) shows disc herniation (arrow). (G) In the same patient (as F), cervical spine magnetic resonance T2 weighted imaging (sagittally oriented) shows multilevel disc herniations (arrows), a well delineated spinal cord surrounded by cerebrospinal fluid, cord distortion, and mild signal abnormality within the cord (double arrows). (H) Cervical spine computed tomography imaging (coronally oriented) shows a C2 odontoid fracture (arrows). (I) Cervical spine magnetic resonance T2 weighted imaging (sagittally oriented) shows a large intradural extramedullary lipoma (arrow), compressing and displacing the spinal cord (double arrow). (J) Brain magnetic resonance FLAIR imaging shows old infarctions in the bilateral frontal lobes (arrows). (K) Brain magnetic resonance angiography (lateral view of internal carotid artery) shows a small ophthalmic artery aneurysm (arrow). (L). Brain magnetic resonance FLAIR imaging shows scattered abnormal T2 hyperintensities (arrows) in the white matter

    Table 1

     Assessing the need for imaging in patients with common neurological problems

    View this table:

    Sources and selection criteria

    We searched PubMed for reviews and clinical trials using the terms “head trauma”, “concussion”, “transient isch(a)emic attack”, “seizure”, “headache”, “back pain”, “neck pain”, “magnetic resonance imaging”, “computed tomography”, “imaging”, “neuroimaging”, and “neurologic(al) imaging” from January 1995 to May 2010. Search results were individually reviewed and cross referenced manually. We also searched the Cochrane Library and Clinical Evidence databases, reviewed guidelines from the American Academy of Neurology, European Stroke Organisation, and American College of Radiology, and used references from our personal collections. Although randomised studies and meta-analyses are available for many of the problems discussed, data are still emerging. Definitive quality and outcome driven data are lacking for some of the subentities discussed.

    When is neuroimaging needed after minor head trauma?

    Minor head trauma and post-concussion syndrome are common (see box 1 for defining criteria). In the United States, 128 per 100 000 population present to emergency rooms with minor head trauma each year.1 2 Doctors worry about fracture, contusion, and space occupying haemorrhage or enlarging haemorrhage—lesions that might signify the need for neurocritical care or neurosurgical intervention (figs 1Bii-iv and 2A). Large prospective cohort studies show that the probability of neurologically important conditions or problems requiring neurosurgical intervention is less than 10% and 1%, respectively,3 4 5 so not all patients with minor head trauma need imaging, and those who do need to be identified.

    Box 1 Definitions2345

    Minor head trauma
    • Blunt head trauma within the past 24 hours and Glasgow coma scale score of 13 or more, together with resultant loss of consciousness or amnesia or disorientation

    Post-concussion syndrome
    • Less than four weeks between head trauma with loss of consciousness and development of symptoms plus at least one symptom from three of the following categories:

      • Headache, dizziness, fatigue, noise intolerance

      • Irritability, depression, anxiety, emotional lability

      • Reduced subjective concentration or memory, or intellectual difficulties without neuropsychological evidence of marked impairment

      • Insomnia

      • Reduced alcohol tolerance

      • Preoccupation with above symptoms and fear of brain damage, with hypochondriacal concern and adoption of sick role


    Two large prospective studies evaluated patients who underwent head computed tomography after minor head trauma.3 4 They each produced a set of clinical decision rules aimed at detecting those patients who need neurosurgical intervention or are harbouring brain injuries (box 2). The presence of any of the findings in either rule set identified all at risk patients (100% sensitivity for each set).3 4 5 Many experts therefore advocate head computed tomography only for patients with positive findings from either set of rules.

    Box 2 Decision rules from New Orleans and Canadian studies345

    New Orleans criteria (initial Glasgow coma scale score 15)*
    • Headache

    • Vomiting

    • Age >60 years

    • Drug or alcohol intoxication at the time of trauma or evaluation

    • Persistent anterograde amnesia

    • Trauma above the clavicle

    • Seizure

    Canadian head computed tomography rule (initial Glasgow coma scale score 13-15)*
    • Glasgow coma scale score <15 two hours after injury

    • Suspected open skull fracture

    • Signs of basal skull fracture

    • Two or more episodes of vomiting

    • Age >65 years

    • Cannot remember anything that happened for at least 30 minutes before the trauma

    • Dangerous mechanism

    • *The Canadian rule has a higher specificity than the New Orleans criteria and can prevent more unnecessary head computed tomography studies5

    The lack of studies on neuroimaging in post-concussion syndrome probably means that clinicians use a similar approach to the one they use with acute minor head trauma. Imaging is avoided if brain injury is not apparent on clinical examination and the diagnosis of post-concussion syndrome is clear.

    What modalities are used?

    Computed tomography of the head is faster than magnetic resonance imaging (5-10 minutes v ≥20 minutes) and detects more pathology. For example, a retrospective analysis of 100 patients showed that 58% of small traumatic subdural haematomas were detected only by magnetic resonance imaging.6 Although the improved prognostic information and data altered medical management, they did not necessarily change surgical management.6

    Advanced imaging techniques

    A variety of new imaging modalities can improve the detection of intracranial lesions and help explain physical findings. This may be particularly valuable in trauma patients, in whom imaging results may be normal despite a poor clinical state. Diffusion tensor imaging measures the directionality of water movement along white matter tracts. Diffusion tensor imaging abnormalities may correlate with destruction of white matter pathways and help in prognosis.7 A variety of functional and perfusion imaging modalities may show focal or regional abnormalities even when standard imaging is normal.7

    How should patients be evaluated after a transient neurological episode?

    When a neurological episode is truly transient, in that no residual symptoms, signs, or deficits remain, history taking is the key to distinguishing high risk patients from low risk ones. Table 2 offers an evidence based approach where data are available and gives answers to important questions that may help establish a diagnosis. The initial investigation of a patient with neurological symptoms that have resolved often focuses on distinguishing between two potential causes of the episode: transient ischaemic attack and epileptic seizure. A meta-analysis found a 15% risk of infarction within three months of a transient ischaemic attack, half of which occurred within 48 hours.8 9 A seizure may point to a serious underlying lesion or systemic illness, or it may herald status epilepticus.10

    Table 2

     Imaging recommendations in transient neurological episodes9-15

    View this table:

    Suspected transient ischaemic attack

    For patients with suspected transient ischaemic attack, urgent imaging of the head and neck vasculature is needed to determine the subsequent risk of stroke and to guide decision making about interventions such as carotid endarterectomy. Head magnetic resonance imaging and head and neck magnetic resonance angiography, or head computed tomography and head and neck computed tomography angiography (fig 2B), should be performed according to evidence based recommendations from the American Stroke and Heart Associations and the European Stroke Organisation.9 11 Magnetic resonance imaging is often preferred because it is more likely to detect pathology. An aggregate of 19 studies (1117 patients) showed that magnetic resonance diffusion weighted imaging detected small areas of infarction in 39% of patients (fig 2C), even when symptoms and signs had resolved and the patient met clinical criteria for transient ischaemic attack.9

    Suspected seizure

    Magnetic resonance imaging can identify pathology that may underlie a seizure. Head computed tomography and magnetic resonance imaging both yield abnormal findings in 10% of patients with seizure, and both techniques are recommended by the American Academy of Neurology and American Epilepsy Society).10 Magnetic resonance imaging with gadolinium is more sensitive than computed tomography, however, for detecting a wide range of causes including tumour (fig 2D), haemorrhage, infection, inflammation, and developmental anomalies (fig 2E).10

    Transient neurological episode of uncertain cause

    In many cases, the differential diagnosis remains broad even after a thorough history and examination. Such patients might report a variety of symptoms not related to either transient ischaemic attack or seizure, such as dizziness, vertigo, presyncope, and syncope. In this setting, imaging is not recommended when the most likely diagnoses include migraine, benign forms of vertigo, psychogenic spells, and cardiogenic or neurogenic (for example, vasovagal) syncope. Head computed tomography alone is unlikely to be useful in such patients, although evidence to guide imaging practice is lacking.

    In certain circumstances, especially when history and examination are incomplete, incidental imaging findings may steer the clinician away from the correct diagnosis. For example, a patient with transient neurological symptoms found to have a chronic cerebral infarction by imaging might be improperly diagnosed with a transient ischaemic attack. This patient may not have had a proper cardiac evaluation, which might have shown that the neurological symptoms were caused by a cardiac arrhythmia.

    Is imaging always needed for severe headache?

    Large prevalence studies have shown that primary headache disorders, such as migraine, affect more than 10% of the population.16 17 A large literature review and a prospective evaluation of patients with non-acute headache found intracranial lesions—such as haemorrhage, raised pressure, infection, or tumour—in less than 1% of patients with a normal neurological examination.12 13 Box 3 outlines features of the patient’s history that should alert a clinician to serious pathology. Expert reviews suggest that investigations are needed only if the history indicates a worrisome entity or the neurological examination is abnormal.12 17 A systemic review showed that the presence of four of the following features associated with headache was strongly suggestive of migraine and made imaging unnecessary: pulsatile quality, duration four to 72 hours, unilateral location, nausea and vomiting, and headache that is disabling.13

    Box 3 Clinical features of headache that warrant imaging121314

    • Acute thunderclap* headache, cluster-type headache, or undefined (not a known primary headache disorder) quality of headache

    • Accelerating (or new) pattern of headache, including during pregnancy

    • Aggravation of headache by exertion or Valsalva-like manoeuvre

    • Onset of headache after age 50

    • Non-classic visual symptoms or other aura

    • Vomiting

    • Other focal neurological symptoms or signs (such as aphasia, neglect, or hemiparesis) or an abnormal neurological examination*

    • Current or alternative medical systemic illness or signs (such as sinusitis or mastoiditis, HIV, cancer, fever, rash)

    • *Particularly predictive of abnormal neuroimaging.12 13

    The presence of any one symptom must be considered in the context of the entire clinical picture. For example, vomiting associated with other symptoms of migraine should reassure the clinician, whereas vomiting associated with non-migrainous headache is a cause for concern (box 3).

    If the history and examination do not support a diagnosis of primary headache disorder or if red flags are raised, the American Academy of Neurology and the American College of Radiology ( recommend imaging to exclude pathology.14 15 The results of a randomised controlled trial suggested that neuroimaging reduces anxiety in patients with headache, which may reduce subsequent costs.18 This might encourage doctors to request imaging in some patients even if the examination does not suggest pathology.

    What imaging technique is most useful?

    The best test to choose depends on the suspected diagnosis (table 1). In patients with acute headache who need rapid diagnosis because of suspected intracranial haemorrhage or cerebral aneurysm, head computed tomography and computed tomography angiography are preferred to evaluate the brain parenchyma and vasculature, respectively. In most other situations, head magnetic resonance imaging with gadolinium is preferred because it provides better anatomical detail and special sequences to probe various pathologies (see figs 1 and 2). The addition of gadolinium increases the sensitivity of disease detection and lesion description,19 particularly when intracranial infection, tumour, or other inflammatory processes are suspected.

    Advanced imaging techniques

    Recent studies have shown syndrome specific alterations in regional cerebral metabolism and blood flow that may offer further insight into the aetiology of headache.20

    When is imaging needed in patients with neck and low back pain?

    Neck and low back pain is ubiquitous and affects most people at some point in their life. A large population based study found a 31% three month prevalence in adults in the US.21 In such patients, imaging often shows non-specific or non-diagnostic findings, such as mild to moderate degenerative spine disease (fig 2F-G).22 A meta-analysis of randomised controlled trials found that lumbosacral spine imaging does not improve outcomes in patients with isolated low back pain without an indication of a serious underlying condition.23 Although studies evaluating diagnostic imaging for neck pain are lacking, the same conclusion would probably hold true. The American College of Physicians and the American Pain Society recommend conservative treatment for patients with neck pain and back pain that is not associated with neurological signs or concerning features.24

    Imaging should be reserved for patients with red flag features on history or physical examination. These include previous trauma; constitutional symptoms such as unexplained fever or weight loss; systemic disease or cancer; and motor, sensory, or sphincteric deficits.25 26 Table 3 shows clinical features that might prompt a neuroimaging evaluation in patients with back and neck pain, adapted from guidelines from the American College of Physicians, the American Pain Society, and an algorithmic approach.24 25 In the absence of concerning features, imaging is generally deferred for at least six weeks while conservative treatment is instituted, because most patients improve during this time.25

    Table 3

     Imaging guidance for neck or low back pain24 25

    View this table:

    Choice of imaging modality

    Magnetic resonance imaging optimally evaluates the spinal cord, nerve roots, intervertebral discs, ligaments, bony elements, and soft tissues—features that are not well delineated by computed tomography (fig 2F, 2G, and 2I). Gadolinium enhanced magnetic resonance imaging is indicated when neoplastic, infectious, or inflammatory conditions are suspected (fig 2I). Computed tomography may be useful when bone integrity needs to be assessed, such as in a patient with a history of trauma, osteomyelitis, or metastases (fig 2H). Although plain radiographs allow evaluation of spinal alignment and stability (with flexion and extension images) and the basic integrity of the bony elements, they have limited use in many settings because of poor anatomical and spatial resolution.25 26 27

    How should incidental findings be interpreted and managed?

    Magnetic resonance imaging detected incidental findings in about 14% of brain imaging studies in a large adult Western population.28 Common incidental findings include asymptomatic infarcts (fig 2J), benign tumours (fig 2D), aneurysms (fig 2K), and small white matter lesions (fig 2L). Practitioners must decide whether to disregard such findings as clinically unimportant, obtain additional images, refer to a specialist, or manage directly. The direct and indirect costs of incidental findings, especially in patients in whom imaging is overused, is unknown.

    Small white matter lesions

    White matter disease comprises a diverse set of cerebral pathologies, but it is usually chronic cerebral ischaemic microangiopathy when found incidentally in elderly people and those with risk factors for cerebrovascular disease.29 The incidence of this type of white matter disease increases with age and, when moderate to severe, increases the risk of stroke in the next four years fivefold; it is also associated with a decline in cognitive function.30

    Asymptomatic infarcts

    Asymptomatic infarcts increase the risk of subsequent stroke threefold.30 Their presence should prompt an evaluation of risk factors for cerebrovascular disease, such as hypertension, hyperlipidaemia, and diabetes, and the need for antiplatelet treatment should be considered. Vascular and cardiac studies, such as non-invasive angiography of the head and neck, echocardiography, and ambulatory electrocardiography, may also be considered.

    Benign tumours

    The most common incidental benign tumours are meningiomas and pituitary adenomas; if smaller than 1 cm in diameter they do not usually become problematic.28 Although controversial, small meningiomas (<2 cm) do not generally warrant subsequent imaging unless they lie in potentially high risk locations (such as parasellar or at the cerebellopontine angle). In contrast, one small study found that pituitary adenomas less than 1 cm in diameter carry a 15% risk of enlargement, so they may warrant further evaluation (neuroendocrinological studies and neurological investigations such as formal visual field testing).31


    A large prospective study reported that small aneurysms in the anterior circulation (anterior cerebral, middle cerebral, and internal carotid derived arteries; <7 mm) have a 0% risk of subsequent rupture at five years.32 However, in a retrospective review of 152 patients with aneurysmal subarachnoid haemorrhage, the ruptured aneurysms were less than 7 mm in diameter in 100 (65.7%) of these patients.33 Thus, controversy exists regarding the best management of patients with small aneurysms discovered incidentally, and whether it is good practice to perform serial imaging.

    Until consensus emerges it is reasonable to avoid serial imaging in such patients with the caveat that an irregularly shaped aneurysm, hypertension, and young age (<50 years) may increase the risk of subsequent rupture.33 Patients with any other type of cerebral aneurysm are best referred to a specialty clinic where the patient’s history and aneurysmal factors will guide risk assessment and subsequent investigation and management.

    Additional educational resources

    Resources for healthcare professionals
    • Krumholz A, Wiebe S, Gronseth G, Shinnar S, Levisohn P, Ting T, et al. Practice parameter: evaluating an apparent unprovoked first seizure in adults (an evidence-based review): report of the quality standards subcommittee of the American Academy of Neurology and the American Epilepsy Society. Neurology 2007;69:1996-2007

    • Vosa PE, Battistinb L, Birbamerc G, Gerstenbrandc F, Potapovd A, Prevec T, et al. EFNS guideline on mild traumatic brain injury: report of an EFNS task force. Eur J Neurol 2002;9:207-9.

    • Frishberg BM, Rosenberg JH, Matchar DB, McCrory DC, Pietrzak MP, Rozen TD, et al; US Headache Consortium. Evidence-based guidelines in the primary care setting: neuroimaging in patients with nonacute headache. 2000.

    Resources for patients
    • UpToDate for Patients ( —Provides general explanations and diagnostic approaches for mild head trauma, transient ischaemic attack, seizures, neck and low back pain, and several of the incidental findings discussed in this article

    • Patient UK (—Provides general explanations and diagnostic approaches to various headache disorders, transient ischaemic attack, and seizures


    Cite this as: BMJ 2010;341:c4113


    • Contributors: All authors helped to conceive and design the study and draft and critically revise the manuscript; ABC and JPK acquired the data; ABC and SM supervised the study. ABC is guarantor.

    • Competing interests: All authors have completed the Unified Competing Interest form at (available on request from the corresponding author) and declare: no support from any organisation for the submitted work; ABC and JPK have no financial relationships with any organisations that might have an interest in the submitted work in the previous three years; SM received a fee from the American Society of Neuroradiology for speaking on experience with the Toshiba 320 Detector computed tomography at a meeting in 2009 and has received research funds from General Electric and Siemens for neuroimaging research; all authors declare no other relationships or activities that could appear to have influenced the submitted work.

    • Provenance and peer review: Commissioned; externally peer reviewed.