Abstract
The purpose of the present study was to explore whether and to what extent the neuroimaging markers could predict the relief of the symptoms of patients with migraine without aura (MWoA) following a 4-week acupuncture treatment period. In study 1, the advanced multivariate pattern analysis was applied to perform a classification analysis between 40 patients with MWoA and 40 healthy subjects (HS) based on the z-transformed amplitude of low-frequency fluctuation (zALFF) maps. In study 2, the meaningful classifying features were selected as predicting features and the support vector regression models were constructed to predict the clinical efficacy of acupuncture in reducing the frequency of migraine attacks and headache intensity in 40 patients with MWoA. In study 3, a region of interest–based comparison between the pre- and post-treatment zALFF maps was conducted in 33 patients with MwoA to assess the changes in predicting features after acupuncture intervention. The zALFF value of the foci in the bilateral middle occipital gyrus, right fusiform gyrus, left insula, and left superior cerebellum could discriminate patients with MWoA from HS with higher than 70% accuracy. The zALFF value of the clusters in the right and left middle occipital gyrus could effectively predict the relief of headache intensity (R2 = 0.38 ± 0.059, mean squared error = 2.626 ± 0.325) and frequency of migraine attacks (R2 = 0.284 ± 0.072, mean squared error = 20.535 ± 2.701) after the 4-week acupuncture treatment period. Moreover, the zALFF values of these two clusters were both significantly reduced after treatment. The present study demonstrated the feasibility and validity of applying machine learning technologies and individual cerebral spontaneous activity patterns to predict acupuncture treatment outcomes in patients with MWoA. The data provided a quantitative benchmark for selecting acupuncture for MWoA.
Highlights
Migraine is a chronic paroxysmal neurological disorder characterized by multiphase attacks of moderate or severe headache and reversible neurological and systemic symptoms [1]
Using the z-transformed amplitude of low-frequency fluctuation (zALFF) values of voxels in these five clusters as inputted features, the overall linear kernel Support Vector Classification (SVC) achieved an accuracy of 86.25% (p = 0.0002), area under the curve (AUC) of 0.9213 (p = 0.0002), sensitivity of 87.5%, and specificity of 85%
The results of the voxel-based correlation analysis illustrated that the zALFF value of the right middle occipital gyrus exhibited a positive correlation with baseline monthly migraine days (MMDs), the zALFF value of the left superior cerebellum positively correlated with the baseline VAS scores, and the zALFF value of the right fusiform gyrus correlated positively with the duration in patients with migraine without aura (MWoA) (Figure 3A)
Summary
Migraine is a chronic paroxysmal neurological disorder characterized by multiphase attacks of moderate or severe headache and reversible neurological and systemic symptoms [1]. Its clinical efficacy in relieving headache intensity and reducing the frequency of migraine attacks has been verified in several clinical trials [8,9,10]. A recent systematic review suggested that acupuncture could reduce the frequency of migraine attacks, and could be considered as a treatment option for migraine [11]. Acupuncture is effective for migraine, certain differences have been noted with regard to its efficacy across different subjects, suggesting that patients’ responses to acupuncture treatment may vary considerably. Identifying migraineurs who may benefit from acupuncture before treatment can improve the efficacy of acupuncture and reduce the waste of medical resources
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
