This study recruited healthy individuals who did not have medical or family histories of pain disorders and had not experienced any significant pain condition during the past year. Furthermore, patients with EM (monthly headache days: 1-14) were enrolled from the Headache Clinic of Taipei Veterans General Hospital (VGHTPE), and the diagnosis (code 1.1: migraine without aura) was according to the criteria of the International Classification of Headache Disorders, Third Edition . All participants were right-handed, denied having any history of systemic or major neuropsychiatric disease, and had normal physical and neurological examination results as well as normal brain MRI results. Participants who were taking any medication (eg, migraine preventive medications) on a daily basis were excluded. Furthermore, patients with migraine who overused medication were excluded. The hospital’s institutional review board (IRB) approved the study protocol (VGHTPE-IRB, 2015-10-001BC), and each participant provided written informed consent.
All participants underwent scheduled MPPT and MEG recordings (both detailed below) and were instructed not to take any analgesics or other medications within 3 days before the recordings. Patients with migraine were assessed during their interictal phase, which was defined as the absence of acute migraine within 2 days before (days −1 and −2) and after (days +1 and +2) the aforementioned assessment. We evaluated the severity of depression and anxiety using the Hospital Anxiety and Depression Scale (HADS) and functional disability in migraine using the Migraine Disability Assessment Scale (MIDAS).
Pain sensitivity measurement
Pain sensitivity measurement was performed in the chronic pain examination room (constant 20°C room temperature and no windows) in the Department of Neurology at VGHTPE. Quantitative measurements of the MPPT were defined as the lowest intensity perceived as painful for participants, and MPPT values were obtained as follows. MPTT was determined using the standard rigid electronic von Frey device (ALMEMO 2450, AHLBORN, Germany), which consists of a 1000-g internal load cell connected to a probe and a rigid tip (diameter: 0.8 mm; IITC Life Science Inc., USA). The system digitally measures and displays the values in grams, with a resolution of 0.1 g. During measurement, the tip was applied perpendicularly to the skin surface, with ascending stimulus intensity applied at 25 g/s . Participants were instructed to inform the examiner immediately upon experiencing a painful sensation. Assessment target areas included the left supraorbital (ie, the first branch of the trigeminal nerve dermatome, V1) and proximal medio-ventral forearm (ie, the first thoracic nerve dermatome, T1). Breaks were taken between each stimulus, and the order of stimulation sites was randomized. Sensory stimuli on each target area were delivered 5 times; the 2 extreme values were excluded, and the average values of the remaining 3 recordings were calculated for further analyses.
Resting-state MEG recording
A 5-minute resting-state MEG recording [33, 34] was obtained for each participant, each of whom sat comfortably with eyes closed but remained awake and relaxed. If the participant fell asleep or had excessive within-run head movement, the recording was stopped and then rerun. Furthermore, a 3-minute empty-room recording was conducted to capture sensor and environmental noises, which were applied to calculate the noise covariance for further source model analysis. MEG data were recorded with the digitization rate of 600 Hz using a whole-scalp 306-channel neuromagnetometer (Vectorview; Elekta Neuromag, Helsinki, Finland) composed of 102 identical triple sensor elements (1 magnetometer and 2 orthogonal planar gradiometers). In total, 4 coils representing the head position were placed on the participant’s scalp, specified by the nasion and 2 preauricular points using Cartesian coordinates and measured with a 3-dimensional (3D) digitizer. For accurate registration, approximately 50 additional scalp points were digitized. These head landmarks and points enabled further coordinate alignment between the MEG and MRI data. Additionally, electrooculography (EOG) and electrocardiography (ECG) activities during MEG recording were simultaneously recorded for data preprocessing. MRI images of individual brain structures were acquired using a 3T MR system (Siemens Magnetom Tim Trio), with a TR of 9.4 ms, TE of 4 ms, recording matrix of 256 × 256 pixels, field of view of 256 mm, and slice thickness of 1 mm.
To obtain intrinsic spontaneous cortical activities and reject the contaminations of nonbrain or environmental artifacts, MEG data were preprocessed as follows (Fig. 1): (1) MaxFilter from the Neuromag software system was applied to remove external noise from MEG recordings [35, 36], (2) the data segments containing artifacts from head movement, muscle activities or environmental noise were manually rejected, (3) notch filters (60 Hz and its harmonics) were used to remove powerline contaminations, (4) identified heartbeat and eye blinking events from ECG and EOG data were used to define the projectors through principal component analysis separately. The principal components meeting the artifact’s sensor topography were then manually excluded through orthogonal projection .
Furthermore, for further source modeling analysis, T1-weighted structural volumetric images were automatically reconstructed into a surface model using BrainVISA (4.5.0, http://brainvisa.info). The detailed geometric reconstruction of the scalp, brain gray and white matter, and tessellations provided a topographical 3D representation of the brain surface and was used to estimate the gray and white matter border.
Source-based spectral power and functional connectivity analysis
In combination with a structural surface model, the distributed source model of resting-state MEG data was estimated using depth-weighted minimum norm estimation analysis (Fig. 1). A forward model was established by overlapping sphere method for rapid calculation of realistic head shapes , which presented each cortical vertex as a current dipole and included approximately 15,000 vertices in the whole brain model. The inverse operator estimated the distribution of current sources that account for data recorded at the MEG sensors. For group analysis, the cortical source model of each participant was then morphed into a common source space defined by the Colin27 anatomy . Through dynamic source analysis, the current density of cortical activation in each individual could be obtained at any cortical region and any time point.
In this study, the regions of interest (ROIs) were defined in the T1 template volume using Mindboggle cortical parcellation  and were composed of 14 brain regions, including the bilateral ACC, medial orbitofrontal cortex (MOF), lateral orbitofrontal cortex (LOF), insula cortex, SI, MI, and PCC. These ROIs were involved in the sensory and affective aspects of cortical pain representation and were related to pain sensitivity in terms of the structural volume or thickness [12-14]. Besides, the bilateral auditory cortices, apart from the pain regions, were also selected as negative control. The time-varying source density of each ROI was individually derived from the averaged source density of each vertex within the ROI. To obtain the resting-state oscillations and functional connectivity, the dynamic source density of each ROI was further analyzed as follows. First, oscillatory power at each ROI was estimated using the Welch method (window duration: 5 s with 50% overlap) and defined as absolute power. Moreover, the oscillatory power was normalized to relative power through the division of the power at each frequency band by the total power, which has been reported to adequately reduce the interindividual variability of the oscillatory magnitude . Second, the amplitude envelope correlation analysis , which orthogonalized the signals to remove zero-lag interactions , was used to calculated the oscillatory functional connectivity between ROIs and then constructed the full 14 ´ 14 adjacency matrix. Oscillatory power and functional connectivity were categorized according to frequency bands: delta (2-4 Hz), theta (5-7 Hz), alpha (8-13 Hz), beta (14-29 Hz), and gamma (30-59 Hz), and averaged in each frequency range. The MEG data preprocessing and analysis were performed using Brainstorm .
The demographics and MPPT of the healthy individuals and EM were compared using Student’s t test or a chi-square test as appropriate. Pearson’s correlation was used to determine the correlation between the MPPT (V1 and T1) and cortical oscillations (absolute and relative) and synchronization (14 ´ 14 ROIs) at the delta to gamma frequency bands, as well as the oscillations and synchronization in the bilateral auditory cortex. Throughout the statistical analyses, false discovery rate correction (FDR) was used for multiple comparisons, and a P value of < 0.05 was considered statistically significant.