If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Address correspondence to: Beatrice P. De Koninck, MS, Sports and Trauma Applied Research Lab, Montreal Sacred Heart Hospital, CIUSSS North-Montreal-Island, 5400 Boul Gouin O, Montreal, Quebec, H4J 1C5, Canada.
Sports and Trauma Applied Research Lab, Montreal Sacred Heart Hospital, CIUSSS North-Montreal-Island, Montreal, Quebec, CanadaUniversity of Montreal, Montréal, Quebec, Canada
Sports and Trauma Applied Research Lab, Montreal Sacred Heart Hospital, CIUSSS North-Montreal-Island, Montreal, Quebec, CanadaUniversity of Montreal, Montréal, Quebec, Canada
Sports and Trauma Applied Research Lab, Montreal Sacred Heart Hospital, CIUSSS North-Montreal-Island, Montreal, Quebec, CanadaUniversity of Montreal, Montréal, Quebec, Canada
Sports and Trauma Applied Research Lab, Montreal Sacred Heart Hospital, CIUSSS North-Montreal-Island, Montreal, Quebec, CanadaUniversity of Montreal, Montréal, Quebec, Canada
Transcranial alternating current stimulation (tACS) has been one of numerous investigation methods used for their potential to modulate brain oscillations; however, such investigations have given contradictory results and a lack of standardization.
Objectives
In this systematic review, we aimed to assess the potential of tACS to modulate alpha spectral power. The secondary outcome was the identification of tACS methodologic key parameters, adverse effects, and sensations.
Materials and Methods
Studies in healthy adults who were receiving active and sham tACS intervention or any differential condition were included. The main outcome assessed was the increase/decrease of alpha spectral power through either electroencephalography or magnetoencephalography. Secondary outcomes were methodologic parameters, sensation reporting, and adverse effects. Risks of bias and the study quality were assessed with the Cochrane assessment tool.
Results
We obtained 1429 references, and 20 met the selection criteria. A statistically significant alpha-power increase was observed in nine studies using continuous tACS stimulation and two using intermittent tACS stimulation set at a frequency within the alpha range. A statistically significant alpha-power increase was observed in three more studies using a stimulation frequency outside the alpha range. Heterogeneity among stimulation parameters was recognized. Reported adverse effects were mild. The implementation of double blind was identified as challenging using tACS, in part owing to electrical artifacts generated by stimulation on the recorded signal.
Conclusions
Most assessed studies reported that tACS has the potential to modulate brain alpha power. The optimization of this noninvasive brain stimulation method is of interest mostly for its potential clinical applications with neurological conditions associated with perturbations in alpha brain activity. However, more research efforts are needed to standardize optimal parameters to achieve lasting modulation effects, develop methodologic alternatives to reduce experimental bias, and improve the quality of studies using tACS to modulate brain activity.
Modifying brain electrical activity through noninvasive brain stimulation techniques (NIBS) has been a major target in the research field of cognitive neuroscience. The coupling of NIBS with electrophysiological recordings and computer-assisted behavioral changes allows the study of causal links between brain oscillations and cognitive function
Furthermore, clinical applications of this type of coupling have shown promising results in several neurological conditions such as depression and Parkinson’s disease.
In addition, brain oscillations have been studied as potential biomarkers for several neuropsychiatric disorders, suggesting the modulation of brain oscillation as a potential therapeutic avenue.
Progress in this field has led to the identification of several key stimulation parameters to optimize modulation of brain activity, such as the administration of transcranial alternating current stimulation (tACS) in a dimly lit and calm environment with eyes open, and the adjustment of the frequency of stimulation below the ceiling levels of the alpha band.
However, the lack of stimulation-protocol standardization still represents a major challenge in optimizing the therapeutic potential of NIBS, which has contributed to several contradictory results and reproducibility failures.
Both methods can be applied for different purposes because their mechanisms differ. Although TMS stimulates the brain through focal penetration of the scalp by high-intensity magnetic fields,
TES operates through a weak electrical current applied over the scalp. TES current is dissipated when in contact with different human tissues so that only a fraction of its total charge is transmitted to the brain.
The new modalities of transcranial electric stimulation: tACS, tRNS, and other approaches.
in: Brunoni A. Nitsche M. Loo C. Transcranial Direct Current Stimulation in Neuropsychiatric Disorders: Clinical Principles and Management. Springer International Publishing,
2016: 21-28
Although weakened, the remaining current alters the membrane potential of affected neurons, leading to a change in their propensity to fire action potentials.
In TES, tACS generates considerable research interest among neuroscientists, mainly because of its demonstrated ability to modify endogenous brain oscillations.
The new modalities of transcranial electric stimulation: tACS, tRNS, and other approaches.
in: Brunoni A. Nitsche M. Loo C. Transcranial Direct Current Stimulation in Neuropsychiatric Disorders: Clinical Principles and Management. Springer International Publishing,
2016: 21-28
tACS consists of the application of alternating sinusoidal currents between two or more electrodes. The brain oscillation modulation generated by tACS is achieved at a barely noticeable low intensity. These low-intensity current strengths appear convenient because they can remain below photic thresholds,1 thus decreasing discomfort and favoring blindness to stimulation conditions.
The new modalities of transcranial electric stimulation: tACS, tRNS, and other approaches.
in: Brunoni A. Nitsche M. Loo C. Transcranial Direct Current Stimulation in Neuropsychiatric Disorders: Clinical Principles and Management. Springer International Publishing,
2016: 21-28
showed that a 5-minute tACS stimulation over FP1 and O1 (10-10 International electroencephalography [EEG] System) induced single-neuron spike timing in a frequency-specific manner. The electric fields generated by the stimulation set at an intensity of ± 2 mA were strong enough to directly affect a deep brain structure, in this case, the hippocampal cells.
later showed the entrainment effects of tACS on neural spike timing but went a step further in investigating the effects of increasing tACS stimulation intensity from 0.5 to 1.5 mA in a dose-dependent study. The increase of stimulation electric field intensity elicited a quantitative augmentation of responding neurons.
tACS neuromodulation effects have mostly been attributed to two distinct yet interrelated action mechanisms. First, the online effects of tACS are thought to be based on the coordination of the endogenous rhythm with the applied external rhythm, referring to the synchronization phenomena.
This yielded a relationship model in a triangular upside-down shape called “Arnold tongue,” corresponding to regions of high synchrony between the injected external current and the endogenous brain oscillations.
Arnold tongue refers to the entrainment potential of an oscillator (ie, neural oscillations) that is coupled to a rhythmic driving force (ie, external sinusoidal current) according to its parameters of intensity and frequency.
For the entrainment to be observed with an applied low stimulation intensity, both the driving force (external injected current) and the endogenous frequency have to be proximate. In case of a greater difference between frequencies, entrainment could be observed with the targeted endogenous oscillations if the applied driving force is applied at a high intensity, although up to a certain level of mismatch.
Once applied, the external injected current would polarize the membrane potential of neurons. The frequency-specificity of tACS represents a clear advantage of tACS over other NIBS techniques.
The new modalities of transcranial electric stimulation: tACS, tRNS, and other approaches.
in: Brunoni A. Nitsche M. Loo C. Transcranial Direct Current Stimulation in Neuropsychiatric Disorders: Clinical Principles and Management. Springer International Publishing,
2016: 21-28
Alternatively, recent EEG studies point to the involvement of neuroplastic changes after tACS. The latter neuroplastic hypothesis is commonly used to account for the offline lasting aftereffects on EEG oscillation measures.
The alpha frequency band is also thought to play a crucial role in visual processing. This is supported by the close anatomical proximity of alpha generators to alpha-dominant visual occipital brain areas.
Furthermore, alpha-band activity represents a particularly appealing target for external brain stimulation given its inhibitory modulating effects on several cortical activities and functions.
therefore making it a versatile therapeutic avenue for alpha-dependent neurological conditions such as disorders of consciousness, neuropsychiatric conditions involving cognitive deficits, and even chronic pain.
Another interest in targeting modulation of alpha brain oscillations is the fact that they present numerous interactions with other frequency bands, which may therefore generate potential cross-frequency effects. Frequency bands are not isolated and independent but influence each other, as in a form of modulation.
Several studies report the influence exerted by the phase of the posterior alpha oscillations on the gamma amplitude, resulting in the synchronization of gamma bursts.
A few studies shed light on the phase-amplitude coupling between alpha and gamma frequency bands using NIBS techniques, such as tACS adjusted to the participant’s alpha frequency, which led to enhanced alpha-gamma coupling.
In contrast to other NIBS techniques, tACS offers the possibility to adjust the external current to the participant’s endogenous frequency within a specific frequency band. The latter methodologic manipulation has been associated with the modulation of optimized brain oscillations, especially at low intensities.
However, despite advances in using tACS as a proxy for modulation of brain oscillations, there is no consensus on the ideal combination of stimulation parameters (ie, frequency, intensity, stimulation location, electrode sizes/montages, and others) to optimize the alpha activity modulation. Several parameters seem to consistently lead to favorable results, notably the relevance of open eyes during the experiment, the use of vigilance tasks during administration of tACS to induce preferential states for neuromodulation, and an optimal environment for modulation with dim light to avoid excessive increase of alpha power.
Elicited visual phosphenes under alpha tACS could increase spectral power within the stimulation frequency band. The use of focused montage enables the isolation of the targeted stimulation mechanisms from this potential confounding factor.
Furthermore, it has been shown that the frequency of stimulation should be below the upper end of the alpha frequency range (8–12 Hz) to increase the likelihood of observing amplification by tACS. The eyes-open condition, although presenting a suppression of alpha peak, is seen as optimal to reduce chances of reaching an alpha-ceiling effect.
Nevertheless, sustained effort toward clarifying optimized stimulation parameters such as intensity and frequency of stimulation, electrode montage, and site of stimulation is warranted to improve reproducibility among study findings involving alpha tACS. Thus, we aimed to conduct a systematic review of the literature to assess whether tACS can modulate alpha power only in studies that paired stimulation with simple visual vigilance tasks to control for a preferential state of stimulation, while restricting potential confounding effects on neuromodulation, such as concomitant somatosensory stimulation or tasks soliciting higher cognitive processes. In this study, we first sought to shed light on this conundrum aiming to identify the overall effects of tACS on alpha-power modulation (increase/decrease). Secondarily, tACS different parameters and settings were assessed, and reported sensations and adverse effects were reviewed.
Materials And Methods
The protocol of this systematic review was registered through Open Science Framework Registries (10.17605/OSF.IO/GNJTC) in addition to PROSPERO (CRD42020104556).
were used. Eligibility criteria framed according to the PICO format were as follows:
—
Population: healthy adults (aged > 18 years).
—
Intervention: active tACS using any stimulation protocol, administered during vigilance tasks or no task. Studies were excluded that implicated tasks with complex stimuli paradigms soliciting higher-order processes such as discrimination of targets or working memory to measure accuracy. This is justified by the fact that the alpha spectral power modulation outcome targeted can be directly influenced by soliciting higher-order cognitive tasks, which can have a confounding effect.
One example of an optimal visual change detection task for modulation and a simple task specifically targeting alpha activity is the psychomotor vigilance task, a low-level sustained attention reaction-time task, in which the participant has to press a key as soon as a visual stimulus appears on arbitrary occurrences.
Comparison: sham tACS or other controlled conditions.
—
Outcome: primary outcomes were EEG or magnetoencephalography (MEG) alpha power changes (increase/decrease). Secondary outcomes were methodologic parameters (current intensity, frequency. density, and current modalities), sensation-reporting assessments, and adverse effects.
Controlled trial studies were included. Case reports, clinical observations, reviews, and animal studies were excluded. Studies that did not include any measurement of EEG/MEG alpha spectral power as an outcome were excluded.
The publication date was not restricted, and the languages of publications were limited to English, French, and Spanish.
Search Strategy
The final search was performed in November 2021 in MEDLINE (OVID search engine) and then adapted to Embase (OVID search engine), Global Health (OVID search engine), PsychInfo, and Dissertation Abstracts and NICE evidence for gray literature. Ongoing trials and systematic reviews were searched through Cochrane Library
The search strategy was developed and guided by a trained librarian (Patrice Dupont) from the University of Montréal (Fig. 1). In addition, other potentially eligible studies were hand-searched using references of included tACS studies. Search results were imported to Zotero,
Two authors (Beatrice P. De Koninck and Samuel Guay) performed the title and abstract screening independently using Covidence. A calibration procedure was first executed with 50 randomly selected references, with each author reviewing independently. Substantial agreement was obtained between the two raters in the calibration process (96%, Cohen’s kappa coefficient = 0.65). The total screening of references was then conducted, in which both raters were blinded to each other during the screening. The agreement resulting from this process was substantial (98.7 %, Cohen’s kappa coefficient = 0.78). Consensus between both reviewers was sought if disagreements occurred, and if it was not reached, a third reviewer (Daphnée Brazeau) served as moderator. Then, the full-text articles were added on Covidence to proceed with assessment eligibility. This step was executed as the screening and by the same reviewers (Beatrice P. De Koninck and Samuel Guay), in a blinded manner.
Data Extraction
Data extraction was performed by two authors (Beatrice P. De Koninck and Daphnée Brazeau) and then revised by a third author (Samuel Guay) through Covidence’s extraction tool 1.0. Covidence is a workflow platform designed for primary screening and data extraction to facilitate the production of standard reviews.
Reconciliation between both reviewers was achieved in case of disagreement. Data extraction consisted of the number of participants (total and receiving each condition), age, sex, tACS parameters (stimulation duration, current frequency, current intensity, site of stimulation, current type, electrode size, and type), EEG/MEG recording site, recording state (eyes open/closed, online/offline), recording duration, sampling rate recording, sensation assessment, adverse effects, statistical analysis performed, and alpha-power increase/decrease.
Risk of Bias
Risk of bias and quality assessment were performed independently by two reviewers (Beatrice P. De Koninck and Daphnée Brazeau) using a tool adapted for the purpose of this systematic review, based on the Cochrane Risk of bias tool offered in Covidence.
The adapted version excluded the assessment of blinding of experimenters and outcome assessors. Our decision to adjust the tool was based on the known challenge with tACS studies to conduct adequate blinding for the experimenters. One factor is that the stimulation is usually programmed in a remote mode through MATLAB and fed through a digital-to-analog converter, which makes the blinding difficult to achieve. Another important factor is that for studies recording EEG during online tACS, the artifact generated on the signal by the stimulation is colossal, making active tACS easy to distinguish from sham.
In case of disagreement and absence of consensus, a third author (Samuel Guay) arbitrated. The criteria used for assessment were random sequence generation (selection bias), allocation sequence concealment (selection bias), blinding of participants, incomplete outcome data (attrition bias), selective outcome reporting (reporting bias), and other potential sources of bias. For each of those, a scale of low, unclear, or high risk was used. Random sequence generation was judged as low risk when the description included the use of a random component, such as a random number table, a computer random number generator, shuffling cards or envelopes, and more. In the use of a nonrandom component in the sequence generation process, such as the systematic approach following order of enrollment or rather, for example, a nonsystematic allocation based on the judgment of the clinician or participant’s preference, the risk of selection bias was judged as high. This criterion was scored as unclear when insufficient information was given to assess its risk. Allocation concealment was judged as low risk when the concealment methods described ensured no foreseeing before the assignment for participants and investigators. This criterion was judged as high risk when the method described represented a potential selection bias such as using unconcealed procedures, such as date of birth, participant ID, and so on. If the method of concealment was not described or not described in sufficient detail to allow judgment, it was stated as unclear. Selective outcome reporting was judged as low risk when the protocol and all prespecified outcomes were either reported or included all expected outcomes that were prespecified. This criterion was judged as high risk in the absence of reporting of prespecified primary outcomes, use of differential analyses from prespecified, or incomplete without relevant justification. Unclear risk of bias was attributed when not enough information was described to make a judgment. For the criterion of participant blinding, low risk was attributed when the absence of blinding was justified by the outcome not likely to be influenced by lack of blinding or when they were mentioned as strong single blind designs. High risk was attributed when the absence of blinding was likely to influence the outcome or when blinding was attempted but potentially compromised. Unclear risk corresponded to insufficient information available or if the study did not address this outcome. Other potential sources of bias implicated concerns that were not covered elsewhere in the assessment table and may represent potential sources of bias for the outcome, such as the absence of a control condition, the use of statistical models that lacked justification, lack of key methodologic descriptions such as the methods used to conduct spectral power analyses, or the specification of whether eyes open or eyes closed were used for analyses and tACS, or absence of blinding assessment. Once these results were acquired, a conversion was performed using the thresholds for Agency for Healthcare Research and Quality standards to obtain a quality of evidence (poor, fair, or good). Potential publication bias is a risk that was considered given the small sample sizes, which are characteristic within the field of NIBS. Approaches to facilitate the judgment of whether publication bias was an issue, such as the inspection of p value distribution
were unfortunately not possible owing to the difficulty of extraction of effect-size estimates from the type of outcome data collected for this systematic review (eg, spectral EEG data). In addition, a series of heterogeneous parameters prevented us from conducting a valid and reliable meta-analysis, including stimulation parameters (eg, type of montage, stimulation frequency, site of stimulation, modalities of stimulation), general methods (Table 1), settings (eg, discrepancies in blinding and randomization), and especially outcomes (eg, type of spectral analyses used), which may have led to misleading conclusions and interpretations.
Absence of alpha-tACS aftereffects in darkness reveals importance of taking derivations of stimulation frequency and individual alpha variability into account.
for which we could not simulate because the stimulation electrode sites were not specified. The current density modeling was computed according to the size, site, and intensity of the electrodes. It is important to mention that two studies
did not specify whether the stimulation intensities used were peak-to-peak or baseline-to-peak (disclosed in Table 1). For the sake of adequate comparisons between studies, we treated these as peak-to-peak when entered in ROAST, but the electric field magnitude values could differ by a factor of two if their reported amplitudes were baseline-to-peak. The first step of the processing pipeline in ROAST is to segment the uploaded magnetic resonance imaging into air cavities, skin, bone, cerebrospinal fluid (CSF), white matter, and gray matter. Next, the virtual electrodes are placed at the desired locations. A Finite Element Model mesh is generated and then solved for voltage and electric field distributions. The results are resampled into the original three-dimensional (3D) volume to create some interactive displays.
The ROAST output, which consists of a 3D matrix containing the electric field magnitude value in each voxel, was used for further analyses. To obtain the value of the electric field in different locations of the cortex, a conversion of the seed Montreal Neurological Institute (MNI) coordinates
into the subject’s voxel coordinates was conducted using the transformation matrix provided by the ROAST output. Subsequently, it was possible to obtain the electric field value in the different seeds by conducting matrix indexing using the 3D matrix discussed previously. Matrix indexing also allowed us to identify the coordinates of the maximum electrical field value (Emax). To assess the focality of the configuration montages, a ratio of the number of brain voxels that have a value > 50% of the Emax on the total number of voxels in the cortex was calculated.
We were then able to obtain the percentage of the brain volume reaching an electric field > 50% of the Emax. Thus, the higher the percentage, the more the electric field is distributed in the brain. As performed by Khatoun et al,
we also calculated the focality using the average electric field for 5% of voxels (Emax5%), with the strongest field instead of Emax to compare the results while limiting the influence of voxels with extreme magnitude values. The code created for data extraction after ROAST simulation is available for use.
Our search resulted in 1429 references. After duplicate removal, 40 references were chosen for full-text review. The final selection process led to 20 references that met the inclusion criteria and were included in qualitative synthesis.
Absence of alpha-tACS aftereffects in darkness reveals importance of taking derivations of stimulation frequency and individual alpha variability into account.
The main reasons for exclusion were participants with medical conditions, use of cognitive tasks other than a vigilance task (state-dependent effect), and outcomes not following inclusion criteria (Fig. 2).
Figure 2PRISMA 2020 flow diagram for new systematic reviews, which included searches of data bases, registers, and other sources.
Characteristics of the 20 studies are described in Table 2. Eleven studies were conducted in Germany, of which nine were held by the same research group.
Absence of alpha-tACS aftereffects in darkness reveals importance of taking derivations of stimulation frequency and individual alpha variability into account.
Study designs were crossover (k = 13) and parallel (k = 7). Most of the participants were young adults (k = 19), and sex distribution was balanced for most studies (k = 19). Among the inclusion criteria, half the studies (k = 10) only recruited right-handed adults, given potential hemispheric differences in cortical excitability.
phase 1: 20 phase 2: 11 (those who tolerated higher intensities in phase 1)
phase 1: 25.40 ± 3.73 phase 2: 26.09 ± 7.78
phase 1: 10 females/10 males phase 2: 6 females
healthy adults
medication intake, neurological disorder, psychiatric illness, evidence of a developmental learning disability or ADHD, alcohol and/or substance abuse, clinically relevant anxiety at the time of testing. Contraindication to tACS.
neurological or psychiatric disorders, head injury, cardiac pacemaker, intracranial metal implant, tinnitus, seizures, fainting, substance abuse. contraindications to tACS
Absence of alpha-tACS aftereffects in darkness reveals importance of taking derivations of stimulation frequency and individual alpha variability into account.
Alpha power modulation was achieved through the administration of a longer-lasting continuous tACS protocol for most studies (13 of 20). However, there is a definite challenge in identifying optimal stimulation conditions, given the various methodologic differences observed between studies that achieved alpha power modulation (Table 3).
Table 3EEG Recording Methodology and Main Outcomes.
Study
Setting conditions
EEG/MEG recording montage
EEG/MEG recording setting (online / offline) and time (min)
Pz, P7, P8, O1 and O2 and referenced to the right earlobe
online and offline, 3 min pre/post/postII
both (3 min eyes-open, then 3 min eyes-closed)
500S/s
starting 50 ms post tACS. raw EEG (rest intervals) segmented into 1-sec epochs. All EEG-epochs were processed as follows:epochs were transformed FFT transformation. Referenced to the Pz electrode.
computed using a wavelet transform with Hanning windows of 200 ms displaced in steps of 10 ms. Power at θ = [5, 8] Hz, α = [8,13] Hz, β = [13,25] Hz, low-γ = [30−40] Hz, and γ = [60−80] Hz bands was computed via trapezoidal integration of PSD of EEG-epochs.
GLMM was used to test for differences in the EEG power across tACS protocols. In the GLMM, the session (tACS10, tACS70 and control), EEG intervals (pre, post and postII) are fixed effects, while experimental block (from 1 to 4) and the subject number (1 to 23) are random effects.
α-power increased after tACS10 (p < 0.05) as compared to baseline. No statistically significant change in α-power is observed after tACS70 or control compared to baseline.α and γ-power increases after tACS10 observed at post0 and at 18-min post stim.
comfortable chair in a sound-proof, temperature-controlled, and electrically shielded room with constant dim light.
C3, C4, Cp1, Cp2, Cp5, Cp6, Cz, F3, F4, F7, F8, Fc1, Fc2, Fc5, Fc6, Fp1, Fp2, Fz, O1, O2, Oz, P3, P4, P7, P8, Pz, T7 and T8 locations of the international 10-10 system
offline, 5 min
closed
250 Hz (0.1 μV steps resolution)
high-pass filter, time constant of 1 s; low pass filter at 70 Hz. notch filtering at 50 Hz (band-width of 5 Hz, symmetrical around to the 50 Hz frequency). Filters implemented as phase shift-free Butterworth filters.
FFT for 28 scalp locations in 0.5–29 Hz range (1-Hz bin resolution but for the 0.5–1 Hz bin), averaged across epochs. Spectral power values averaged across canonical EEG bands: delta (1–4 Hz), theta (5–7 Hz), alpha (8–12 Hz), beta (13–24 Hz)
paired two-tailed t-tests (Δ θ-tACS vs Δ sham). threshold-free cluster enhancement (TFCE) for multiple comparisons
θ-tACS increased α-power at C4, F4, Fc2, Fc6; in the α-range main result is represented by an extended effect of the stimulation at 10 Hz frequency bin, encompassing a large portion of the scalp with a bilateral peak at central areas.
seated comfortably in a dimly lit, sound-attenuated room
day 1: Fp1, Fp2, Fpz, C3, C4, Cz, P3, P4, Pz, POz, PO7, PO8, O1, O2 and Oz. day 2: Fp2, Fpz, F3, F4, C3, C4, Cz, P3, P4, Pz, POz, O1, O2 and Oz.
offline, pre/post 10 min 5 min at 60 and 120 min post tACS
open
256 Hz
high-pass filter at 0.1 Hz and 60 Hz notch filter. Recordings were segmented into 2-s epochs.The first 3-min, eyes-open, at-rest recordings (pre/post) were used. 5-min post60 and post 120 were used.
mean spectral power was obtained using Welch’s method by frequency bands; theta (4–8 Hz), alpha (8–12 Hz) and beta (8–32 Hz). Results were then averaged across all epochs for each frequency band.
LMMs to compare effects of frequency, site and intensity of stimulation on α-power modulation directly after tACS as well as over time (60 and 120 min after stimulation)
low intensity posterior IAF tACS induced significantly greater aftereffects on α-power when tested at both 60- and 120-min post-stimulation. the posterior condition IAF tACS condition induced an overall dominance of a specific α-power increase, specifically near the stimulation site.
electrically shielded, sound-proof, and dimly lit room.
P7, P3, PZ, P4 and P8 according to the 10–20 System.
online and offline, 5 min pre and post
open
5000 Hz
1 s segments. First 200 artifact free 1 s segments for pre, online and post for each condition were baseline corrected by subtracting the mean, multiplied by a hanning window.
Pz (P4 for some participants) used. Averaged FFT, maxn (8–14 Hz). α amplitude: mean of the range 2 Hz from peak. Value normalized relative to average amplitude from corresponding 5 min pre-measurement
normalized α amplitude values: RM ANOVA; one factor of condition and three levels (conditions).post data RM ANOVA; one factor of condition and four levels (conditions). Post hoc pairwise t-tests with Bonferroni correction.
online data: significant difference between positive ramp sawtooth and sham (p = 0.0059), but no significant differences between any other conditions (p > 0.1)
subjects were seated in an upright position in the MEG shielded room
102 positions, each a channel triplet of one magnetometer and two orthogonal planar gradiometers, yielding 306 sensors overall.
online, 2 min repeated during blocks
both
1000 Hz
offline high-pass filtered above 1 Hz and then downsampled to 512 Hz. then EO and EC resting-state data: nonoverlapping epochs of 2 s aligned to the phase of the stimulation.
Fourier coefficients were estimated for each epoch and in reconstructed activity in brain sources. multitaper spectral estimation with a fixed smoothing window of ± 2 Hz, with a 1 Hz resolution for the frequencies between 1 and 40 Hz and 2 Hz for those between 42 and 84 Hz.
percent Δ power (strong tACS vs sham baseline) within subject and condition. nonparametric cluster-based permutation dependent-samples T-statistics.
contrast for weak tACS: No state-dependent power activity. Strong stimulation: significant difference (pcluster < 0.006) at 10 Hz, max effect in posterior cingulate, where largest state-dependent effect in α power increases (EC vs EO).
52 passive electrodes setup mounted in an elastic cap based on the standard international 10-10 system. Electrodes CP3, CP5, P3, P5, CP4, CP6, P4 and P6 were omitted.
offline, 2 min pre/post
open
2500 Hz
EEG data recalculated to average reference and cut into 120 segments of 1 s each.
amplitude spectra calculated via FFT and averaged for each time window. Mean amplitude values of mu-α oscillations (mu-α peak frequency ± 1 Hz) extracted at C3 and C4
t-test against 0: Potential Δ amplitude values (tACS and sham). paired t-test: difference tACS-sham. paired t-test on baseline difference in pre-stim amplitude between tACS-sham.
significant difference between tACS-sham-related modulation of mu-α oscillations (p = 0.044); significative negative modulation of mu-α amplitude after tACS block (central electrodes, bilaterally), (p = 0.008)
dimly lit room with participants seated in a recliner in front of a computer screen at a distance of approximately 100 cm. sessions started either at 9 am or 2 pm
10 sintered Ag-AgCl electrodes mounted in elastic cap, placed at five frontal and five parietal positions around Fz and Pz, following the international 10–20 system layout.
offline, 3 min baseline EEG; 90 min post
open
250 Hz
high-pass filtered at 0.3 Hz, low-pass filtered at 100 Hz segmented into 3 min blocks resulting in one baseline block prior to and 30 blocks after tACS
FFT spectra (Hanning window, 2 s zero-padding) computed and averaged for first 120 artifact free epochs (1 s) in each 3 min block. From FFT, IAF (IAF ± 2 Hz) power obtained and averaged for each bloc (post-stim IAF band power normalized with respect to pre-stim baseline).
RM ANOVAs factor time (10 levels) and between subject factor group (stim vs sham). separate two-sided t-tests for stim and sham group against baseline computed on IAF band power. Bonferroni and Greenhouse-Geisser applied
1st and 2nd 30 min block post-tACS: power in IAF band increased in stim group vs sham. 3rd 30 min block post-tACS: no significant difference in group vs sham. Further analysis (α power calculated for each 3 min blocks) shows that aftereffect vanish around 70 min
seated underneath the sensor array in upright position (60° dewar orientation)
306-channel whole-head MEG system with 102 magnetometer and 204 orthogonal planar gradiometer sensors
offline, 10-min immediately before and after stimulation
open
resampled to 250 Hz
signals were subsequently imported to Matlab and resampled to 250 Hz. A 4th-order forward-backward Butterworth filter introducing zero phase-shift between 1 Hz and 40 Hz was applied.
FFTs (4-s zero-padding, Hanning window) were computed on each of the segments. The resulting power spectra were averaged across the first 260 artifact-free segments in each experimental block.
a comparison of the source-projected α-power increase from the pre- and post-stimulation blocks of the two experimental groups by means of an independent samples random permutation cluster t-test
a significantly larger power increase in the tACS group as compared to sham (permutation cluster t-test, pcluster = 0.013)
seated in a comfortable chair in a sound-attenuated room
recorded from 59 scalp positions using NetStation 4.4.2 (EGI Software, Eugene, OR, USA), an EGI NetAmps 300 amplifier with a 24-bit analog-to-digital converter, and the HydroCel Geodesic sensor net
offline, 5 min pre/post
closed
1000 Hz
1–100 Hz ban-pass filter. Artifacts exceeding ± 100 μV were excluded from all channels. Fifteen randomized artifact-free epochs (epoch length: 4096 s) were determined for each participant.
EEG data were reanalyzed using Matlab 2017 software (MathWorks, Natick, MA, USA), including a fast Fourier transform with a 1–50 Hz filter to calculate absolute power for all frequency bands
to compare differences in changes of the EEG power value before and after tES intervention in the three groups, statistical significance was tested by MANOVA at each band frequency individually.
postintervention activation increased significantly in the fronto-central and parietal regions at the FCZ, FC2, and P4 electrodes in the tACS group compared to the sham group.
during experimental sessions, participants were conformably seated in a chair located in an electrically shielded cabin.
signal was obtained from eight electrodes (F3, F4, Fz, C3, C4, Cz, P3, P4) mounted on a neoprene headcap in accordance with the international 10–20 EEG system. Online electrical reference earclips consisted of two opposed Ag/AgCl pellets of 8 mm diameter on the right ear.
offline, 5 min pre/20 min post
open
500 Hz
EEG segments were separated into five-minute periods. The first segment corresponded to baseline (T0; pre-stimulation), and the four others to post-stimulation recordings (T1; 0–5 min, T2; 5–10 min, T3; 10–15 min, T4; 15–20 min). All segments were then split into 1 s epochs, and segments contaminated by eye blinks or muscle movements were excluded using a semiautomatic artifact detection algorithm (min–max 100 μV criterion).
FFT: 1 Hz frequency resolution (Hanning window function [10%]). Epochs were averaged for each time interval and condition and the mean power activity (μV2) was extracted for α and β frequencies
RM ANOVA, with factors Condition (10 Hz, 20 Hz, sham) and Time (T0, T1, T2, T3, T4) was used separately to test for changes in α and β EEG power. This analysis was conducted for C3-C4 at each of the two frequency bands.
For α-power at electrode site C3, there was no main effect of Condition nor Time and no Condition × Time interaction.For α-power at electrode site C4 there was no main effect of Condition, no main effect of Time and no Condition × Time interaction.
post-EEG into 10 epochs of 3 min each. Each pre-EEG, 10 × post-EEG split into 180, 1 s segments. first 120 segments without artefacts were used for further analysis. mean value substracted of each segment avoiding DC distortion of the spectra at 0 Hz.
FFT on Pz (each segment): 120 spectra for each data averaged. Individual mean spectral powers calculated. Power data normalized to α-power of pre-EEG. Analysis (lower band, IAF −5 to −3 Hz) and above (upper band, IAF +3 to +5 Hz)
tACS after-effect: normalized spectral power/coherence via a Two-Way RM ANOVA;(between-subject factor group (2 levels) and within subject factor time (10 levels). one sample t-tests against 1, for α power increase after tACS and sham stimulation compared to baseline IAF power
in E1 (EC) no difference between stim and sham in any frequency band. E2 (EO): α-power enhanced after tACS, but not after sham stimulation, also visible over time
EC: only for the Two-Way ANOVA with repeated measurements on the normalized alpha coherence, η2 EO: η2
2 sessions: ambient illumination and without (“light” and “dark”). “dark”-session: spotlight turned off, only monitor. “light”-session: spotlight turned on and produced 500 lux at 1 m. 3 days between sessions
23 head locations according to the 10/10-system
offline, 15 min pre and 30 min post tACS
open
10 kHz
down-sampled to 500 Hz, high-pass filtered at 1 Hz and low-pass filtered at 100 Hz, re-referenced to a combined Fp1/Fp2. cut into 2 baseline blocks of 5 min and 30 blocks poststimf 1 min each. divided in 1-s segments.
FFT using Hanning window with 2 s zero-padding, averaged across all segments for each block. IAF poststimulation power by scanning power peak (8–12 Hz) at Pz, averaged over 30 min after stimulation
Δ α power over time by using a generalized additive mixed regression model (GAMM) to account for inter-subject variability and for time being a continuous, multilevel variable
α power increase in both sham and stimulation conditions directly after stim (Dark: 20%; Light: 7%). Within 30 min post stim, for sham: Dark: remained stable; Light: increase 0.337% per min. Stim condition: Dark: increase 0.366% per min; Light : increase 0.7% per min.
Eta squared (η2) for the pre-stim alpha-increase only
Absence of alpha-tACS aftereffects in darkness reveals importance of taking derivations of stimulation frequency and individual alpha variability into account.
participants were seated in a dark room, with a monitor as a sole light source
23 active electrodes according to 10-10 system
offline, 3 min pre tACS and 10 min after each tACS block
open
10 kHz
down-sampled to 1000 Hz, high-pass filtered above 0.5 Hz and low-pass filtered 48 Hz. segments starting and ending 30 s before and after stimulation, resulting in 3 min baseline block and four segments of 9 min length for both stimulation groups
first 66% of artifact free trials of each segment used to compute the mean α-power for each block using a Hanning window with 2-s zero padding. for poststimulation power analysis, data normalized to power in baseline-segment
RM ANOVA for Δ α-power poststimulation; between subject factor group (stim/sham) and within-factor time (observation windows 1, 2, 3, 4) for both groups against sham corresponding time-segments. Greenhouse-Geisser applied
significant effect of stimulation on alpha power increase following 10 min of α-tACS in the increasing sequence
seated in a dimly lit room in front of a light emitting diode (LED) in 50 cm distance centered between their eyes
25 sintered Ag-AgCl electrodes fitted in an elastic cap. A standard 10–20 layout was applied with a vertical EOG-electrode, referenced to the tip of the nose.
online (intermittent 8-s recordings) and offline (10 min pre and post tACS conditions)
open
250 Hz
data rearranged into a 10 min pre-stimulation block, 149 intermittent 7 s epochs between stimulation epochs, and a 10 min poststimulation block. 1 Hz high pass filter and a 100 Hz low pass filter, using a two-pass Butterworth filter of sixth order
blocks divided into 1 s trials and Fourier transformed, using a 5 s zero padding and a Hanning-taper. Alpha peak power in each block was determined by identifying the peak α-power (maximum between 6.5 and 13 Hz) at electrode Pz in the averaged spectrum of each block.
kruskal Wallis test on the relative increase in peak power between the groups. the average α-power during the nonstimulated epochs within the stimulation-measurement relative to the prestimulation power with a was also tested using a Kruskal Wallis test.
significant difference of α-power changes between the groups, fixed-stimulation showed significantly increased power compared to sham, the comparison between closed loop and sham was not significant. No difference between groups in α-power during the stimulation.
25 sintered Ag-AgCl electrodes in an elastic cap. Standard 10–20 system layout
1.5 s pre/post recording between tACS blocks, offline
open
5000 Hz
pre-tACS epoch from −1100 to −100 ms relative to tACS onset and post-tACS epoch from 1500 to 2500 ms relative to tACS onset. data were sampled down to 250 Hz afterwards.
P3 and P4 analyzed. Oscillatory power and ITC calculated for pre-tACS and the post-tACS epochs by means of FFT. Power and ITC were calculated for six bins of 100 consecutive trials each.
mean power at IAF ± 1 Hz: 2 × 2 RM ANOVA with factors session (IAF vs control) and epoch (pre vs post)
significant main effect for time indicating a power change during session. However, not attributed to tACS at IAF. Increased α activity in IAF and control group over time probably reflecting fatigue
P3 and P4; 16-channel V-Amp amplifier and ActiCap BP active electrodes, according to 10/20 system
1-min preoffline recording; online recording
eyes open for the 1 min control, then eyes closed for stimulation protocol
not reported
triple-sweeps of 100th order zero phase-lag high-pass finite impulse response filter (f = 2 Hz) in MATLAB.
FFT smoothed using moving average filter (width = 2 Hz), 120–140 Hz. For time-resolved spectral analysis, spectra calculated with multitaper FFT on 1-s segments, whitened by multiplying each frequency by frequency value (1/f method)
paired t-test (Bonferroni correction); for group statistics, mean α amplitudes near stimulus peak (−135° to −45°) and near stimulus (45° to 135°) at P3 and P4 current intensity
significant modulation of LFP amplitude by TES phase at intensities of 4.5, 6, and 7.5 mA at each hemisphere when the preferred current direction was applied. All intensities were tested against 0 mA condition
segments 1 s-epochs, two 1 s-epochs cut at the beginning of each 2.3 s-intervals, divided into blocks of early and late epochs, respectively
FFT for 1–20 Hz (0.5 Hz resolution) for ind. epochs via Hanning window and 2 s zero-padding. Spectra were averaged across epochs and α-bands per condition. Mean α-power normalized relative changes from pre-test to post-test in dB
α-After effect with intermittent α-tACS: Friedman test to compare conditions. Wilcoxon Signed Rank Tests (two-tailed)
Significant α-enhancement for long conditions vs sham. Α-enhancement after active (LongCo > LongDi > ShortCo) not significantly different between conditions. Long intermittent tACS significantly enhanced α-power (vs sham) irrespective of phase-continuity between trains.
C3/4, Cz, P3/4, Pz, POz, Oz, according to 10/10 system
offline, 8 min pre and post tACS
eyes closed (3 min) and eyes open (5 min) for pre-post EEG
5 kHz
signal amplified to a range of ± 3.2768 mV at 0.1 μV resolution, bandpass-filtered online between 0.1-1000 Hz. Pre- and post-EEGs were re-referenced offline to electrode C, after 1 s epochs.
FFT (4-30 Hz, Hanning window, 4 s zero padding, 0.25 Hz resolution) on 120 epochs (2 min) each set. α-power resulting spectra as mean power across trials and frequencies (IAF ± 2 Hz) for all pre- and post-tests in each condition
nonparametric Friedman test for the effect on relative power change at electrode POz
α-power changes not statistically discernible at the group level between conditions. LowCont: least effective in producing α-increase from pre- to post-test
electrically shielded, sound-attenuated, and dimly lit cabin
CPz, Pz, and POz, according to 10–20 System
offline, 3 min pre and post tACS
open
500 Hz
online band-pass filter of 0.016–200 Hz with IIR filter with attenuation of 12 dB/octave. FIR high-pass filter with a cut-off frequency of 0.5 Hz (60 dB attenuation of direct current (DC) signals) was applied off-line, with 1 s segments
absolute spectra computed via FFT for each segment. the resulting 150 spectra were averaged to evaluate tACS-induced cortical modulation
mean spectral amplitude (IAF) ± 2 Hz; RM ANOVA between factor Group (tACS-group/Sham-group) and within factor measurement (Pre/Post). post-hoc t-statistics performed
individual α power increased from pre- to post with tACS, but not with sham. significant difference between pre- and post-stimulation in the tACS-group but not in sham (tACS specific effect)
participants were subsequently instructed to sit relaxed, avoid movements, and later to keep their eyes opened or closed cued block-wise.
31 Ag-AgCl electrodes mounted in a passive EEG EasyCap using a standard 10–20 system
offline : pre-stimulation and post-stimulation alpha-band power were extracted from 500 ms before and 500 ms after each tACS interval
eyes open and closed
500 Hz
the 1 s long pre- and post-stimulation data intervals for each stimulation interval were extracted and bandpass-filtered with a 5–40 Hz fourth order Butterworth zero-phase filter
FFT of the detrended EEG data at POz and parieto-occipital cluster. individual α-power values were calculated by averaging power values in the range of IAF−1 Hz to IAF+1 Hz for pre/post-stimulation time windows and averaging across trials of each condition.
ANOVARM with factors time (pre- vs post-stimulation), state (EO vs EC) and stimulation (in-phase vs anti-phase). ANOVARM with factors block, state, and stimulation. Pre-to-post modulations of α-power for the CSP(pre) data were modeled with an RM ANOVA comprising the factors time, state and stimulation.
α-power are not significantly modulated by tACS. analysis of the spectra revealed visible differences between pre/post-stimulation time for signals extracted from both CSP components for eyes closed and open. there was no systematic change of pre to post α-power modulations across time.
Generalized Eta Squared (η2G) and CI
n.a., not applicable; PSD, power spectral density.
Two tACS experiments stimulated at frequencies outside the alpha band. One study stimulated at a theta frequency that resulted in increased alpha-power peaks over right fronto-central EEG electrodes and an extended effect at 10 Hz bilaterally over central areas (focalityEmax5% = 42.96%).
The other study chose another avenue administering tACS at a gamma frequency of 40 Hz, which resulted in an increase of alpha-band activity in the fronto-central and parietal regions compared with the sham stimulation condition (focalityEmax5% = 28.06%).
For studies strictly comparing active tACS with sham without any manipulation of other experimental factors (k = 6), individual alpha frequency (IAF) band power increased for the stimulation condition compared with sham (focalityEmax5% = 55.31%
). Equivalent alpha modulation in the active alpha sinusoidal stimulation condition relative to a sham condition was also observed in six other studies (focalityEmax5% = 55.31%
Absence of alpha-tACS aftereffects in darkness reveals importance of taking derivations of stimulation frequency and individual alpha variability into account.
). However, a study comparing 10-minute ramp sawtooth, sine wave tACS and sham revealed only a significant difference in alpha-power change between the positive ramp sawtooth and sham stimulation conditions.
For one study, when comparing tACS stimulation effects with eyes closed (EC) relative to eyes open (EO), the weak stimulation intensity condition (0.5 mA) did not induce a state-dependent effect. However, when stimulating at higher intensities (maximum intensity of 1.5 mA), a state-dependent effect was found maximal over the posterior cingulate. Alpha power was further increased by tACS under the high-intensity condition (focalityEmax5% = 59.43%).
also compared different intensity strengths through a stimulating condition of low intensity (maximum of 1 mA peak-to-peak) and a condition of much higher intensities (4–6 mA peak-to-peak) (focalityEmax5% = 26.88%).
Much higher and longer-lasting alpha-power increases of up to 120 minutes after tACS were observed under the 1 mA stimulation condition than under the high-intensity simulation condition. This type of increase of alpha-power modulation under tACS at 1 mA is comparable with the high-intensity condition of up to 1.5 mA from the previous study.
compared a closed-loop condition (readjustment of frequency alternating between eight-second recordings and eight-second tACS) and a fixed IAF frequency condition with a sham condition. Only a significant difference in alpha-power increase was observed for the fixed-frequency condition compared with sham.
Certain studies investigated the effect of manipulating specific parameters that were not related to current modalities. On the subject of EO vs EC, a study reported increased alpha power with EO.9 Furthermore, the effects of light intensity on tACS modulation of alpha power were investigated by Stecher et al
using a comparison of a dark tACS condition, a light tACS condition, and a sham stimulation condition. Directly after the tACS, no differences were observed between the sham and the stimulation condition for both light and dark conditions. However, within 30 minutes, both tACS stimulation conditions showed a greater increase of alpha power than did the sham condition; the stimulation under the dark condition induced an increase in alpha power of 0.366% per minute relative to no change for the sham condition, whereas the tACS under the light condition induced an increase of 0.7% per minute relative to an increase of 0.337% per minute for the sham condition.
Accordingly, light dimming during recording was found in half the studies (k = 10).
The phase parameter was manipulated in two studies. One study presented various intermittent tACS conditions with different phase modalities. Results showed no significant difference between all three active tACS conditions, namely long-phase–continuous, short-phase–continuous, and long-phase–discontinuous.
Another study also experimented with the adjustment of phase using a closed loop to apply tACS. The comparison between in-phase and antiphase showed no difference and no modulation of alpha power.
However, a significant alpha power increase was observed only under long intermittent tACS conditions when compared with sham, irrespective of phase continuity. Finally, in one study, tACS-induced negative alpha oscillations amplitude modulation compared with sham, which was observed over central electrodes bilaterally.
The main tACS parameters are provided in Table 1. Fixation material varied between conductive gel (k = 7) and conductive paste (k = 10), when reported (k = 3 did not report it). The use of analgesic topic agents during tACS was not reported. In most studies, visual vigilance tasks (k = 13) were used. One study alternatively presented an auditory detection task as a vigilance task
Stimulation montage electrodes among studies mainly consisted of electrodes positioned over the medial axis on Cz and Oz (k = 10) (Fig. 3a–d,f–i,m,n). As for remaining studies, electrodes were placed bilaterally (k = 8), on PO7/9–PO8/10 (k = 4) (Fig. 3j,o–q,t), in the central region C3-C4/CP3–CP4 (k = 2) (Fig. 3k,l), in the frontal region F3–F4 (k = 1) (Fig. 3s) or between F7–F8 and T7–T8 (k = 1) (Fig. 3e). One study did not disclose the exact electrode location.
Electrodes were either rectangular (k = 15) or circular (k = 5), and size varied. Most studies (k = 16) stimulated using large electrodes, with surface areas varying from 16 cm2 to 35 cm2.
Absence of alpha-tACS aftereffects in darkness reveals importance of taking derivations of stimulation frequency and individual alpha variability into account.
Figure 3Current density simulation of tACS according to electrode placement, size, and intensity. a–e. Intensity < 1 mA. f–l. Intensity = 1 mA. m–s. Intensity = 1 mA, 2 mA. t and u. Intensity > 4 mA. a. Fuscà et al,
Electrical current administered through active tACS was continuous in most studies: 16 involved continuous sinusoidal tACS current for at least one active condition, and four studies included a protocol involving short, intermittent, repeating sinusoid trains.
01407-6&id=gr1.jpg){kind=link}
01407-6&id=gr2.jpg){kind=link}
01407-6&id=gr3.jpg){kind=link}