Neuroimaging for Migraine: The American Headache Society Systematic Review and Evidence-Based Guideline

Randolph W. Evans, MD; Rebecca C. Burch, MD; Benjamin M. Frishberg, MD; Michael J. Marmura, MD; Laszlo L. Mechtler, MD; Stephen D. Silberstein, MD; Dana P. Turner, MSPH, PhD


Headache. 2020;60(2):318-336. 

In This Article


This systematic review included 23 articles which attempted to assess the value of neuroimaging in migraine. The methods for selecting subjects varied considerably among studies. Some specifically included subjects with worrisome features. Other studies retrospectively analyzed findings of CT or MRI after they were ordered by physicians. Only a few prospectively studied patients with migraine. In the few cases in which neuroimaging lead to the discovery of clinically meaningful abnormalities, many had abnormal exam findings such as homonymous hemianopsia,[42] progressive hemiparesis and focal seizures,[27] or previously diagnosed secondary causes of migraine including brain surgery.[43] While white matter abnormalities are common in those with long-standing migraine, they more likely represent a consequence of migraine, rather than a cause of the disorder.[22] Another limitation is that some of these studies were performed over 40 years ago. These studies predate recent headache classification and do not differentiate episodic and chronic migraine.

Another discrepancy was the definition of neuroimaging abnormalities. Early studies reported cerebral atrophy and white matter foci on MRI as significant abnormalities, but more recent studies have focused on clinically meaningful abnormalities that require observation or treatment. This review specifically focuses on neuroimaging for migraine, and excludes other common primary headache disorders such as the trigeminal autonomic cephalalgias and facial pain disorders. Only 9 of the studies exclusively studied MRI. Compared to CT, MRI does not pose a risk of radiation and may identify abnormalities commonly missed on CT located in the pituitary or posterior fossa, venous sinuses, and optic nerve.[44,45] For patients with disorders such as low or high cerebrospinal fluid pressure or Chiari malformation, a normal CT may be falsely reassuring. As MRI is widely available, carries no known biologic risk, and had significantly increased sensitivity, The American Headache Society and other organizations now recommend MRI over CT for patients presenting with subacute or chronic headache for those patients who need neuroimaging.[12] In patients at high-risk for having significant abnormalities, the judicious use of MRI may actually improve outcomes and decrease medical costs.[46]

Most of the studies in this review predate advances in MRI technology such as stronger magnet sizes, ultra-high-field magnetic resonance angiography, and the ability to obtain thinner slices for specific regions such the pituitary or brainstem.[48] These advances offer physicians more choices in selecting exams, and communication between referring providers and radiology can ensure more appropriate imaging. The studies in this review did not assess a role for non-invasive angiography or venography in the evaluation of migraine. Given its potential for toxicity,[49] there is no indication for the routine use of gadolinium contrast in the imaging of migraine, unless there is a high index of suspicion for another disorder such as multiple sclerosis or brain cancer.

In spite of these differences, the medical evidence to date appears fairly consistent. Subjects with concerning clinical or exam features frequently have abnormalities which require attention and should be imaged. Neuroimaging may be considered for presumed migraine for the following reasons: atypical in nature, prolonged or persistent aura, increasing frequency, severity, or change in clinical features, first or worst migraine, migraine with brainstem aura, migraine with confusion, hemiplegic migraine, late-life migrainous accompaniments, aura without headache, side-locked headache, and posttraumatic headache. While criteria have been promoted to guide recognition of secondary headache, so called "red flags" such as fever, immunosuppression, papilledema, or pregnancy, especially in combination increase the chances of secondary headache.[50] These signs and symptoms guide neuroimaging selections such as ordering angiography for suspected reversible cerebral vasoconstriction syndrome, or gadolinium enhancement for suspected low pressure headache.

However, there is no evidence that routine imaging for migraine meeting International Classification of Headache Disorders 3rd edition criteria (at least 5 attacking of migraine without aura and at least 2 attacks of migraine with aura)[51] is more likely to reveal meaningful abnormalities compared to the general healthy population in the absence of worrisome features. Several studies affirm that routine neuroimaging for migraine meeting the criteria is more likely to identify incidental abnormalities than identify serious problems, potentially creating anxiety or leading to further work-up.

Reducing the overutilization of neuroimaging is a high priority as we move toward value-based care delivery models.[52] In many cases, clinicians may overestimate the patient's desire to receive neuroimaging.[53] The threshold to perform neuroimaging varies considerably from provider to provider, even among specialists.[54]

While some clinicians may request neuroimaging hoping to ease the anxiety of their patients,[55] the initial reduction in anxiety is lost at 1 year follow-up in a study of chronic daily headache.[56] However, neuroimaging significantly reduced costs for patients with high levels of psychiatric co-morbidity. Rather than routine neuroimaging of migraine patients, an alternative approach is to establish a strong relationship and educate the patient about the low yield of neuroimaging. Informing and involving patients in the decision-making process may increase patient satisfaction and improve outcomes.

If the headache is resistant to migraine treatment or has changes in migraine character, reevaluation may be necessary. Reassure patients that neuroimaging can be performed at a later date should new symptoms or signs develop. Patients with potential warning signs of catastrophic headache (eg, thunderclap headache, neurologic deficits) in need of urgent attention rarely present in an outpatient setting and are an exception.

This review does not touch on emerging neuroimaging research such as functional neuroimaging,[57] or studies generally used to investigate secondary headache disorders such as CSF flow studies[58] or MRI venograms to assess for the presence of transverse sinus stenosis.[59]


It is not necessary to do neuroimaging in patients with headaches consistent with migraine who have a normal neurologic examination. Grade A (strong recommendation, high quality evidence).

  1. Neuroimaging may be considered for presumed migraine for the following reasons: unusual, prolonged, or persistent aura; increasing frequency, severity, or change in migraine clinical features, first or worst migraine, migraine with brainstem aura, confusional migraine, hemiplegic migraine, late-life migrainous accompaniments, migraine aura without headache, side-locked migraine, and posttraumatic migraine. Most of these are consensus based with little or no literature support. Grade C (strong recommendation, low quality evidence).