Resting-state and task EEG
Use Welch PSD and band power to assess frequency content; specify window length and absolute/relative power.
Consulting and data analysis
From research ideas to meaningful data.
Alongside device selection and experimental design, we offer consulting and hands-on training for EEG, fNIRS, ECG, EDA, EMG and eye-tracking data analysis. Let’s plan acquisition, preprocessing and analysis around your research question.
Acquisition → analysis → research findings
EEG
ECG
EDA
EMG
Eye tracking
BCI
fNIRS
Multiple signals. An analysis approach for your research.
Schematic illustration · Not actual measurements or analysis results.01 / Analysis workflow
Collect the signal. Explore each step.
Choose a signal to explore acquisition, quality checks, preprocessing, feature extraction and evaluation. Each study’s analysis plan depends on its experimental design and data quality.
Analysis lab
From raw recording to result, every step computed.
Choose a signal and the analysis runs from start to finish on a synthetic recording built with the noise and artefacts real recordings carry. Every curve, map and number on screen is computed from that data at that step: how much the filter reduced the noise, how much of the blink ICA removed, how the ERP emerges as trials accumulate.
Generating the recording and computing the analysis…
Synthetic recording · every value computed from it
The recordings are synthetic: generated with the known properties of real signals (spectrum, artefacts, response shapes) and the same on every visit. The methods are computed the way their counterparts in common tools (MNE-Python, EEGLAB, NeuroKit2, the HRV Task Force standards, Homer3) compute them. The results belong to a single synthetic participant; they are not a study's finding, a clinical assessment or a device's performance.
Analysis guide
Which analysis answers your question?
These workflows are starting points. Filters, thresholds and analysis windows depend on the research question, sampling rate and recording conditions. One pipeline does not fit every dataset.
EEGFrom raw recordings to spectral, ERP and BCI analysis
- 01
EEG recording and event markers
What to doDocument channel locations, sampling rate, reference and event codes; retain an unchanged raw-data copy.
Quality checkpointDo event markers match stimulus timing? Check clock drift in multimodal recordings.
Step outputRecording inventory and verified event table.
- 02
Channels and artifacts
What to doInspect raw traces and spectra; mark bad channels, eye/muscle artifacts, mains interference and gaps.
Quality checkpointVisually verify automatic flags and log rejected channels and durations.
Step outputChannel and recording-segment quality report.
- 03
Filtering and recording preparation
What to doChoose appropriate filters and reference. If needed, review ICA components for artifact handling; epoch around events and define an appropriate baseline.
Quality checkpointDocument filter cutoffs/design, check that relevant neural activity is preserved, and specify epoch-rejection criteria.
Step outputProcessed data, epochs and parameter log.
- 04
Spectral and event-related analysis
What to doUse PSD/band power for resting-state recordings, ERP for stimulus-locked tasks, or time–frequency analysis for temporal changes.
Quality checkpointDo windows, frequency ranges, electrode/ROI choices and usable trial counts match the question?
Step outputParticipant- and condition-level feature tables and figures.
- 05
Comparing conditions
What to doCompare conditions with design-appropriate statistics; report effect sizes, uncertainty and multiple-comparison corrections where needed.
Quality checkpointDo not treat trials from one participant as independent participants. Prevent train/test leakage in BCI models.
Step outputReproducible analysis files, methods and research report.
Analysis options
Event-related potentials · ERP
Event-based epoching, baseline correction and trial averaging; assess amplitude/latency within predefined electrode and time windows.
Time–frequency and BCI
Time–frequency analysis for task-related rhythms; feature extraction and cross-validation for BCI. Split by trial, session or participant according to the intended use.
ECGFrom R peaks to heart rate and HRV
- 01
ECG recording and experiment timing
What to doDocument electrodes, sampling rate, recording length, task and rest periods.
Quality checkpointVerify event alignment and document posture, movement and respiratory conditions.
Step outputRecording metadata and analysis periods.
- 02
Beats and noise
What to doReview movement, signal loss and noise; select segments suitable for reliable beat detection.
Quality checkpointAn automatic quality score is insufficient on its own; inspect questionable traces.
Step outputUsable durations and exclusion log.
- 03
Preparing signals and beat intervals
What to doFilter appropriately, detect R peaks and check missed/double detections. Prepare normal-to-normal (NN) intervals for HRV.
Quality checkpointDistinguish raw RR from cleaned NN intervals; document ectopic-beat/artifact handling and correction rate.
Step outputVerified beat times, NN series and correction log.
- 04
Heart rate and HRV
What to doSelect heart rate and time-domain metrics such as RMSSD/SDNN; add frequency-domain or Poincaré analysis when duration and signal conditions permit.
Quality checkpointCheck comparable durations, resampling, stationarity and respiratory effects for spectral analysis.
Step outputCondition-level heart rate and selected HRV metrics.
- 05
Comparing physiological responses
What to doCompare task/rest periods using the study design; report quality, duration and recording conditions.
Quality checkpointLF/HF alone is not a definitive measure of sympathetic–parasympathetic balance. Do not interpret research metrics as a clinical diagnosis.
Step outputHRV report with methods and limitations.
Analysis options
Heart rate and event-related response
Derive heart rate from R peaks and assess changes within stimulus/task windows.
Time-domain HRV
Calculate RMSSD/SDNN from NN intervals; report duration, correction rate and experimental conditions.
Frequency-domain and nonlinear HRV
When suitable, assess LF/HF band power or Poincaré SD1/SD2; state methods, recording length and interpretation limits.
EDAFrom skin conductance to tonic and phasic responses
- 01
Skin conductance recording
What to doDocument sensor placement, conductance units, sampling rate, environment and stimulus timing.
Quality checkpointCheck electrode contact, initial stabilization and alignment of stimulus markers.
Step outputRecording metadata and stimulus table.
- 02
Contact and motion effects
What to doMark sudden movement changes, contact loss, saturation and gaps.
Quality checkpointDistinguish physiological responses from contact artifacts and identify unusable durations.
Step outputClean analysis periods and artifact log.
- 03
Tonic and phasic components
What to doChoose suitable filtering and decomposition; assess slow tonic levels separately from phasic responses.
Quality checkpointDocument decomposition methods/parameters and compare signals before and after processing.
Step outputTonic SCL and phasic component series.
- 04
Skin conductance responses
What to doExtract SCR amplitude, response count, rise time and stimulus-related latency within study-specific windows.
Quality checkpointDefine response thresholds/windows and account for overlapping responses to closely spaced stimuli.
Step outputStimulus- or period-level EDA feature table.
- 05
Comparing tasks and stimuli
What to doCompare conditions, accounting for individual baselines, environment and movement.
Quality checkpointEDA alone cannot definitively identify a specific emotion or stress; interpret alongside behavior and experimental context.
Step outputEDA report documenting context and methods.
Analysis options
Tonic level · SCL
Compare slow skin conductance levels across rest/task periods.
Stimulus-related phasic response · SCR
Assess SCR amplitude and latency in defined post-stimulus windows; document nonresponse trials and overlapping responses.
Period-level response frequency
Compute response counts/rates during tasks, accounting for duration and detection thresholds.
EMGFrom muscle activity to envelopes, RMS and movement phases
- 01
EMG recording and movement alignment
What to doDocument target muscles, electrode orientation/placement, sampling rate and movement phases.
Quality checkpointCheck consistent placement/protocol and synchronization with movement sensors.
Step outputMuscle, channel and movement-event inventory.
- 02
Checking muscle signal quality
What to doInspect contact, motion artifacts, mains noise, saturation and crosstalk from adjacent muscles.
Quality checkpointCan rest and contraction be distinguished? Visually verify questionable raw segments.
Step outputChannel quality and usable movement cycles.
- 03
Filtering, rectification and envelope
What to doFilter appropriately; plan full-wave rectification and envelope or RMS estimation. Select protocol-appropriate normalization if needed.
Quality checkpointDocument windows and filters; if using MVC normalization, ensure a suitable and reliable reference measurement.
Step outputProcessed EMG, envelope/RMS series and parameter log.
- 04
Measuring muscle activation
What to doAssess amplitude, activation onset/offset and movement-phase patterns; examine frequency features in suitable protocols.
Quality checkpointDefine onset thresholds/minimum duration; use appropriate unrectified data for frequency analysis.
Step outputMovement-phase activation features.
- 05
Assessing movements and conditions
What to doCompare tasks/phases, reporting normalization, electrode placement and load conditions.
Quality checkpointMedian-frequency changes alone do not diagnose fatigue; account for contraction type, force and movement.
Step outputMuscle-activation figures, feature tables and methods report.
Analysis options
Amplitude and activation envelope
Use RMS or rectified-signal envelopes across movement phases; specify windows and normalization.
Activation onset and offset
Use baseline-appropriate thresholds and duration criteria; visually verify detections.
Frequency features
Assess median/mean frequency in suitable contraction protocols alongside amplitude and load information.
Eye trackingGaze and visual attention
- 01
Gaze recording and calibration
What to doAfter calibration and validation, record gaze coordinates, timestamps and stimulus metadata. Plan clock synchronization for concurrent EEG recordings.
Quality checkpointCheck calibration accuracy and EEG synchronization.
Step outputCalibration and synchronization record.
- 02
Tracking loss and accuracy checks
What to doReview calibration error, blinks, invalid samples and tracking loss. Define usable periods and acceptance criteria.
Quality checkpointDocument invalid-sample rates and predefined exclusion criteria.
Step outputParticipant- and period-level quality report.
- 03
Preparing gaze data
What to doTransform coordinates, flag invalid samples and filter if appropriate. Consider interpolation only for justified short gaps.
Quality checkpointDo not interpolate long gaps or turn saccades into artificial fixations.
Step outputValid gaze samples and preprocessing parameters.
- 04
Fixation, saccade and AOI analysis
What to doUse velocity- or dispersion-based methods to identify fixations and saccades. Extract dwell time, time to first fixation and transitions for study-specific areas of interest.
Quality checkpointDocument velocity/dispersion thresholds, minimum duration and AOI geometry.
Step outputFixation, saccade and AOI feature tables.
- 05
Comparing visual attention patterns
What to doCompare scanpaths, AOI metrics and conditions. Heatmaps are descriptive; report invalid-sample rates, accuracy and participant-level differences.
Quality checkpointHeatmaps alone are not statistical evidence; account for tracking loss and AOI size.
Step outputScanpath figures, AOI comparisons and methods report.
Analysis options
Fixations and saccades
Assess durations, amplitudes and transitions using a documented detection method and thresholds.
Areas of interest · AOI
Assess dwell time, visits and time to first fixation in the context of AOI geometry and task.
EEG and eye tracking
Align gaze events with EEG epochs. Address eye-movement artifacts and overlapping events in fixation-related potentials.
BCIBrain–computer interface
- 01
BCI paradigm and labelled recordings
What to doChoose motor imagery, SSVEP or P300 around the project goal. Plan task labels, event timing and participant-specific calibration recordings.
Quality checkpointCheck task labels and paradigm suitability.
Step outputParadigm protocol and labelled EEG trials.
- 02
Trial quality and class balance
What to doReview artifacts, usable trials and class distribution. Plan participant- or session-level train/test splits according to the intended application.
Quality checkpointMatch train/test independence to the intended participant/session generalization.
Step outputQuality report and data-splitting plan.
- 03
Preprocessing BCI recordings
What to doFilter and epoch recordings. Fit scaling and learned transforms only on training data, then apply the same transforms to test data.
Quality checkpointFit scaling, CSP and feature selection only within training folds.
Step outputLeakage-safe processing pipeline.
- 04
Feature extraction and model training
What to doConsider band power/CSP for motor imagery, spectral/CCA approaches for SSVEP, or event-related features for P300. Select models and parameters using training data.
Quality checkpointMatch CSP, CCA or event-related features to the paradigm; avoid test-data tuning.
Step outputTrained model and validation plan.
- 05
Independent validation and online testing
What to doAssess balanced accuracy, confusion matrices and, where appropriate, ITR on untouched test data. Check latency, false commands and feedback separately during online testing.
Quality checkpointReport balanced accuracy, class-level results and confusion matrices; document ITR assumptions and online latency.
Step outputIndependent test report and online prototype evaluation.
Analysis options
Motor imagery BCI
Band power, ERD/ERS and, when appropriate, CSP-based classification; test participant/session generalization according to the intended use.
SSVEP and P300 BCI
Assess target-frequency relationships for SSVEP or stimulus-locked features for P300; verify stimulus and event timing.
Interactive student and research projects
Prototype robot, drone or virtual-environment commands in simulation first; assess feedback, latency and false commands.
fNIRSHemodynamic response
- 01
Optode placement and intensity recording
What to doRecord source–detector distances, wavelengths, sampling rate and task markers. Plan a common time axis for EEG–fNIRS.
Quality checkpointVerify wavelengths, optode geometry and event synchronization.
Step outputOptode map and task timeline.
- 02
Optode contact and motion checks
What to doAssess saturation, coupling, motion artifacts and channel quality; identify usable channels and periods.
Quality checkpointCheck coupling, saturation and motion-affected segments.
Step outputChannel quality and artifact report.
- 03
From optical density to HbO and HbR
What to doConvert light intensity to optical density, apply appropriate motion correction and use the modified Beer–Lambert law for relative HbO/HbR changes. Document parameters.
Quality checkpointDocument optical path-length factors and distances; use a consistent motion-correction/conversion sequence.
Step outputOptical density and relative HbO/HbR series.
- 04
Task-related hemodynamic analysis
What to doChoose task averaging or a general linear model (GLM). Account for short-separation channels, when available, and physiological effects appropriately.
Quality checkpointConsider short-separation channels, drifts and systemic effects; document GLM design and contrasts.
Step outputTask responses or GLM coefficients and contrasts.
- 05
Channel and condition comparisons
What to doAssess HbO/HbR together and report effects and multiple comparisons across channels/ROIs. Hemodynamics are not a direct measurement of neural activity.
Quality checkpointDocument channel/ROI selection and multiple-comparison correction; interpret HbO and HbR together.
Step outputHbO/HbR figures, comparisons and methods report.
Analysis options
Task-averaged response
Epoch around events and define baselines to assess relative HbO/HbR; report trial counts and exclusions.
General linear model · GLM
Build a design matrix from task/nuisance regressors; choose a hemodynamic model, noise structure and contrasts.
EEG–fNIRS and multimodal analysis
Align electrophysiology and hemodynamics while accounting for different response latencies.
Make the method as clear as the result.
Start with the research question
Define primary outcomes/comparisons before analysis; distinguish later exploratory analyses.
Record quality decisions
Rejected channels, durations, trials and corrected beats explain how results were produced.
Match statistics to the design
Respect participants, sessions and repeated measures; report effect sizes and uncertainty.
Make analysis reproducible
Preserve raw data and organize code, software versions, parameters and outputs consistently.
Method references
The processing steps draw on open MNE-Python and NeuroKit2 method documentation. These general workflows do not replace review of your data and protocol.
02 / Support for your research
Plan together. Learn together.
Data analysis consulting
Let’s assess your recording setup, data quality and research question, then develop a project-specific plan for preprocessing, feature extraction, statistical evaluation and reporting.
Request analysis consultingHands-on analysis training
Learn to explore EEG, ECG, EDA or EMG data step by step. We plan training around your team’s experience, software and research goals.
Request analysis trainingBCI project consulting
From brain signals to interactive projects.
From student projects to research prototypes, plan paradigms, devices, data analysis and validation for motor imagery, SSVEP and P300 BCI studies.
Motor imagery
Movement-related EEG rhythms, feature extraction and participant-specific model evaluation.
SSVEP and P300
Visual stimulus design, event timing, target selection and online feedback.
Robots, drones and virtual environments
Integrating BCI commands into robotic or simulation systems; first assess latency and false commands in a virtual environment.
Multimodal research
Concurrent EEG, eye tracking and other biosignals, time alignment and joint analysis planning.
Research projects
Build your project’s analysis plan with us.
We can scope data analysis consulting and hands-on training for research projects, including TÜBİTAK, TÜSEB and university BAP projects.
Analysis plan
Research question, data structure, outcomes and analysis methods.
Data quality and preprocessing
Channel/beat/artifact checks, exclusion criteria and processing log.
Analysis and reporting
Feature tables, figures and methods within the agreed scope.
Hands-on training
Step-by-step sample-data analysis, parameter selection and interpretation.
Charging services to a project depends on the call conditions, approved budget and institutional procurement procedures. The scope and quotation should be assessed with your principal investigator and institution.
Start with your project.
Share your research question, signals, devices and software, recording setup and the analysis or training you need. There is no need to send raw participant data with the initial request; we will agree on the scope of data sharing together.
03 / Analysis and training request
Tell us about your research.
Describe your needs and project stage. Do not include raw participant data or identifying/health information in the initial request. The form prepares your email; complete sending in your email application.
Consulting and data analysis
Request project consultingThe right system and experiment design.
We also offer advice on devices and accessories, technical compatibility and synchronization for multimodal experiments combining EEG–fNIRS, eye tracking and other signals.