Participants
The Institutional Review Board of Shanghai Jiao Tong University approved this study for Human Research Protections (Ethics reference number: B2020027I). All experiments were performed according to the guidelines of the Experimentation Committee of Shanghai Jiao Tong University (Shanghai, China). 52 right-handed healthy 22- to 35-year-old (17 male and 35 females) participants were enrolled in this study. The participants were randomized to receive thermal stimulation (n = 8), mechanical stimulation (N = 18), or thermal plus mechanical stimulation (TMS, n = 26). Each participant receives the same treatment on bilateral LI4 or control points located in the hind leg at least three days apart in randomized order. All participants had signed up for the informed consent and were aware of the test procedure. After the experiments, every participant received a specific subsidy.
Non-invasive acupoints’ stimulations
The thermal stimulation device was developed from a moxibustion device (ijoouNJ01, Baichuanai Technology, Shenzhen, China). We added a 3D-printed heat-conducting gasket (photosensitive resin with a diameter of 2.5 cm and thickness of 0.25 cm) to the heater of the moxibustion device (Fig .1a). The mechanical stimulation device is made of non-woven and shaped into a rigid, blunt, hollow cone with a diameter of 1 cm and height of 0.8 cm (Fig. 1b). The TMS device was developed by adding the 3D-printed heat-conducting gasket with the mechanical stimulation device to the heater of the moxibustion device (Fig. 1c). The center of the protrusion is filled with tinfoil to ensure heat transfer. The heating part was controlled at 43 ± 1℃ during the TMS and thermal stimulation. Each participant receives the same treatment bilaterally on the LI4 or randomly selected point on the hind limb.
Electroencephalography recording
Electroencephalography (EEG) signals were recorded at 64 channels according to the 10–10 electrode system using a wireless EEG acquisition system (NSW364, Neuracle Changzhou Co., Ltd.). The electrode impedances were kept to less than 10 KΩ during recording. The sampling rate was 1000 Hz. Recording channels were connected to a reference electrode in the FCz position and the ground electrode in AFz. The signal from the EEG cap was transmitted through a Wi-Fi router by the wireless EEG amplifier to the recording computer.
Participants were instructed not to drink coffee, tea, or any other beverage containing stimulants two hours before the beginning of the recording session. Participants sat comfortably in an armchair in a sound-attenuated room kept with constant temperature at 25 ℃ ± 2 ℃ and humidity at 50% ± 5%. The baseline EEG was recorded for 5 minutes. Then, a 30-minute thermal, mechanical, or TMS stimulation to the bilateral LI4 or the control point was carried out, followed by another 5-minute EEG recording. During the EEG recording, participants were instructed to relax with open eyes.
EEG Data analysis
EEGLAB14_1_2b based on MATLAB R2020a was used to process data. An FIR filter filtered all data ranging from 0.5 Hz to 40 Hz. The artifacts contained in the data were manually identified using independent component analysis. We calculated power spectral density (PSD, dB) and divided it into five frequency bands through Welch’s function, including δ, θ, α, β, and γ bands. The θ/β ratio was also calculated. The command for calculating PSD is as follows:
$$\begin{array}{c}\left[pxx,f\right]=pwelch\left(x,window,noverlap,f,fs\right)\#\left(1.\right)\end{array}$$
Among them, pxx is the PSD estimate, x is the input signal, the window is the signal segments multiplied vector, the noverlap decides samples of overlap from segment to segment, the frequencies in f are in cycles per unit time, and fs is the sampling rate. Then, we convert the signal to a power value (dB) to obtain a normal distribution by the following formula:
$$\begin{array}{c} power value = 10\times {log}_{10} \left(signal\right)\#\left(2\right)\end{array}$$
Due to the abundance and complexity of epochs, we divided all epochs into several brain regions, including bilateral frontal-, parietal-, temporal- and occipital lobes and central lobe[41]. The central lobe was considered a separate epoch since this region spans many brain lobes.
Since all of our non-invasive stimulations were conducted onto the bilateral LI4s or control points, we combined the data from each left and right hemisphere lobe when comparing the EEG power spectrum and power changes between groups.
Psychomotor Vigilance Test
The mood score was tested with a 10-point Likert scale based on the level of drowsiness. The Psychomotor Vigilance Test (PVT) data were analyzed using PC-PVT 2.0 software (Biotechnology HPC Software Applications Institute, Frederick). The metric we analyzed is absolute response time. The PVT reaction is considered adequate if the response time is ≥ 100 ms, and the verse is considered wrong when the response time is < 100 ms. It is considered an error when the response time is ≥ 500 ms.
Statistical analysis
Statistical analyses and significance tests of EEG power values and ratios were done by MATLAB, with > Q3 + 1.5*IQR or < Q1–1.5*IQR as the basis for outlier judgments. Q3 is the upper quartile, Q1 is the lower quartile, and IQR is the interquartile range. For comparisons of two groups, analyses were performed with the data after excluding the outliers; Paired t-test, Student’s t-test, or non-parametric test was used, depending on whether the data met a normal distribution and had homogeneous variances with SPSS 22.0. Histograms and spectrum diagrams were plotted using GraphPad Prism 8.0.2.