Neurologists diagnose migraine by the description of symptoms provided by the patient. We have not had an objective test to confirm that a person suffers from migraines.
A group of researchers led by Dr. Yiheng Tu in the department of psychiatry at Harvard Medical School developed an AI program that can diagnose migraine using fMRI (functional MRI) scanning. The AI program was first fed information on fMRIs of 116 individuals with migraines and then had this data compared to healthy controls.
The AI program had 93% sensitivity and 89% specificity. This means that it missed the diagnosis of migraine in only 7 out of 1oo patients and diagnosed migraine in 11% of patients who did not have it. These are very good numbers, but clearly, the method is not error-proof.
When they compared people with migraines to those with other types of pain, the sensitivity dropped to 78% and specificity, to 76%. This can be explained by the fact that similar functional changes in the brain probably occur with any type of pain.
A major obstacle to the wide use of fMRI scans is the cost. They are more expensive to perform than a regular MRI. Insurance companies are not likely to cover it since this is an experimental procedure. Another potential difficulty is that fMRI takes much longer to do than a regular MRI – an hour vs 20 minutes. During this time you have to lie inside a tube while trying not to move and hearing loud banging noises.
