2026-10-14T17:07:02Z https://www.scidb.cn/oai

10.57760/sciencedb.43516 2026-07-28T16:59:53Z

2026-07-28 EEG Dataset Recorded During 1-Hz rTMS Stimulation（1Hz-rTMS-EEG） doi:10.57760/sciencedb.43516 en The 1Hz-rTMS-EEG dataset contains preprocessed electroencephalography (EEG) recordings acquired concurrently during 1 Hz repetitive transcranial magnetic stimulation (rTMS) in healthy adults. The dataset is intended to characterize transient neurophysiological responses during low-frequency rTMS and to support studies of TMS-evoked potentials, stimulation-intensity effects, time-frequency features, and related computational modeling. This dataset contains only EEG data recorded during 1 Hz rTMS stimulation and does not include pre- or post-stimulation resting-state EEG, raw MRI images, or EMG data.The dataset was collected from 20 healthy adults, including 12 males and 8 females, aged 21-32 years (mean +/- SD: 25.1 +/- 2.29 years). All participants were right-handed, reported no history of neurological or psychiatric disorders, and met standard MRI and TMS safety criteria. A within-subject repeated-measures design was adopted. Each participant completed 4 stimulation conditions: Sham, T80 (80% resting motor threshold, RMT), T100 (100% RMT), and T120 (120% RMT). The session order was randomized and implemented in a single-blind manner. Sessions were separated by at least 1 week to minimize carryover effects. rTMS was delivered using a Magstim Rapid2 stimulator with a figure-of-eight coil at 1 Hz for 20 minutes, yielding 1200 pulses per session. Stimulation was applied over the left primary motor cortex (M1), with coil positioning guided by a neuronavigation system co-registered to each participant’s T1-weighted structural MRI. Structural MRI was acquired on a 3T GE SIGNA UHP MRI scanner using a 3D MP-RAGE sequence with isotropic spatial resolution of 1.0 x 1.0 x 1.0 mm3, serving neuronavigation and targeting purposes rather than as part of the main dataset.EEG was recorded using a TMS-compatible BrainAmp DC amplifier and an EasyCap system from Brain Products. Although the original setup was based on a 64-channel cap, the neuronavigation headband physically interfered with the outer-ring electrodes, resulting in unstable impedances and reduced signal reliability in those channels. Therefore, only 34 high-quality electrodes were retained for formal recording and analysis, primarily covering bilateral sensorimotor regions. The retained electrodes were F1, F3, F5, FC1, FC3, FC5, C1, C3, C5, CP1, CP3, CP5, P1, P3, P5, Fz, Cz, CPz, Pz, F2, F4, F6, FC2, FC4, FC6, C2, C4, C6, CP2, CP4, CP6, P2, P4, P6. The sampling rate was 5000 Hz, with FCz as ground and AFz as reference. In terms of temporal information, the dataset captures short-latency EEG responses time-locked to each TMS pulse. Before preprocessing, data were segmented into epochs from -300 ms to 950 ms relative to each pulse, covering both the prestimulus baseline and the poststimulus response period. After preprocessing, all data were downsampled to 1000 Hz, resulting in a final temporal resolution of 1 ms. Spatially, the dataset consists of discrete scalp recordings from 34 electrode locations rather than continuous voxel-wise brain imaging data.All data in this dataset have undergone a standardized offline preprocessing pipeline implemented in MATLAB using EEGLAB (version 2024.0) and TESA (version 1.1.1). TMS pulse onsets were first identified using the TESA toolbox with the Cz channel as the reference, applying a derivative threshold of at least 104 uV/ms and a refractory period of 3 ms to avoid duplicate detections. Continuous EEG data were then epoched from -300 ms to 950 ms relative to each pulse and baseline-corrected using the -300 ms to 0 ms interval. The large stimulation artifact segment from -10 ms to 28 ms was removed and interpolated. A first round of independent component analysis (ICA) was then applied to remove TMS-related and muscle-related artifacts, with an average of 21 +/- 4 components back-projected into sensor space. The data were subsequently downsampled to 1000 Hz, band-pass filtered using a fourth-order Butterworth filter (1-100 Hz), and notch-filtered at 48-52 Hz. A second ICA was then performed to remove ocular artifacts, blinks, and residual muscle activity, with an average of 16 +/- 4 components back-projected. Finally, all data were re-referenced to the average of the 34 retained electrodes to produce the final preprocessed dataset.With respect to data quality control, the dataset does not contain missing participants or entirely missing sessions. However, because concurrent TMS-EEG recordings are inherently susceptible to pulse artifacts, muscle activity, eye movements, and coil-related vibration, the number of retained artifact-free epochs varies across participants and conditions. After preprocessing, the mean numbers of retained artifact-free epochs were 976 +/- 91 for Sham, 1087 +/- 51 for T80, 1061 +/- 81 for T100, and 1029 +/- 77 for T120. The minimum retained numbers of epochs across participants were 770, 996, 874, and 878 for Sham, T80, T100, and T120, respectively. Therefore, the dataset does not contain missing values in the conventional tabular sense, but the number of valid trials is not identical across conditions because of stringent artifact rejection, which is common in TMS-EEG preprocessing.Potential sources of error or uncertainty mainly arise from the TMS-EEG paradigm itself, including the strong transient electromagnetic artifact induced by each TMS pulse, stimulation-related muscle activity and auditory responses, inter-individual differences in scalp impedance, differences in coil-to-cortex distance, and possible residual artifacts in the earliest poststimulus time window. Although all data have undergone standardized preprocessing and strict quality control, these factors may still affect signal amplitudes in specific time windows and certain spatial patterns. In addition, because only 34 electrodes were retained, the spatial coverage is mainly concentrated over sensorimotor-related regions. The dataset is therefore well suited for analyses of evoked responses, local topographies, and time-frequency dynamics, but it is not intended for high-density whole-brain source localization.If the released version is organized by participant or condition, each data file typically corresponds to the preprocessed EEG data from one participant under one stimulation condition. The data structure is generally represented as trials x channels x time points or channels x time points x trials, depending on the storage convention specified in the accompanying documentation. Trial labels indicate valid EEG epochs corresponding to individual TMS pulses, channel labels indicate electrode names, and the time axis indicates sampling points relative to the TMS pulse. Signal amplitude is typically expressed in uV, and time is expressed in ms. If the dataset also includes participant information tables, condition label tables, or channel description files, these files generally contain metadata such as participant ID, sex, age, stimulation condition code, and channel names.If the dataset is released in EEGLAB .set/.fdt format, it can be used within MATLAB together with the EEGLAB toolbox. If metadata are provided in .csv or .tsv format, they can be opened and processed using common software such as Excel, Python, or R. The official MATLAB website is https://www.mathworks.com/, the official EEGLAB website is https://sccn.ucsd.edu/eeglab/, and the TESA toolbox project page is https://github.com/nigelrogasch/TESA. EEG; TMS; TMS-evoked Potentials; Event-Related Spectral Perturbation 汪胜佩 RESTRICTED dataset Science Data Bank