Frequency-Specific Alternations in the Amplitude of Fluctuations in Tension-Type Headache: A Machine Learning Study.

Brain neural signal at different frequency bands relates to different functions. However, the frequency-specific properties of spontaneous brain activity in tension-type headache (TTH)-the most rampant primary headache-remain largely unknown. We investigated the local neural activity of 33 TTH patients and 31 healthy controls (HCs) in the conventional frequency band and two sub-frequency bands (slow-4 and slow-5 frequency band), employing fractional amplitude of low-frequency fluctuations (fALFF), percent amplitude fluctuations (PerAF) and Wavelet-ALFF analytic methods. Using age as covariate, we performed two sample t-test to compare the between-group differences of each metrics in each frequency band. Support vector machine (SVM) was conducted to classify TTH patients and HCs on the basis of altered spontaneous brain activities. TTH patients showed lower fALFF values in the left cerebellar lobule X, left parahippocampal gyrus, and right supplementary motor area in slow-5 band. TTH patients showed lower PerAF in the left fusiform and cerebellar regions in three bands. Altered Wavelet-ALFF values in the right thalamus, left anterior cingulum gyrus, superior parietal gyrus and middle and parietal frontal regions in three frequency bands were detected. And the SVM classifier obtained an overall accuracy of 77.38%, 82.38%, and 95% based on fALFF, PerAF, and Wavelet ALFF values, respectively. TTH patients exhibited abnormal neural activity in various brain regions. The abnormal brain activities serve as powerful features for distinguishing TTH patients. This preliminary exploration provides a novel insight into the underlying mechanism of TTH.