Standards of EEG Acquisition Procedures Including Activation

What You Will Learn in This Chapter

A qEEG is only as trustworthy as the recording it came from. You can run the most sophisticated spectral analysis in the world, compare it against the best normative database available, and still produce a report that describes jaw tension rather than cortex. That is why the International QEEG Certification Board devotes so much attention to procedure, and why this unit walks through the workflow one stage at a time.

You will follow a qEEG from the moment the electrodes go on to the moment the report reaches the referring clinician. Along the way you will study the 2025 IQCB minimum technical requirements for acquisition, the certified reviewer's visual inspection across multiple montages, artifact rejection by hand and by algorithm, the computation of spectral metrics, and source localization with LORETA. You will also examine where independent component analysis and principal component analysis help, where they hurt, and why discriminant algorithms belong in a confirmatory role rather than a diagnostic one.

IQCB Blueprint Coverage: This unit addresses Standards of EEG Acquisition Procedures Including Activation (IV.D) within EEG (IV), and supports Editing and Identifying Artifacts (IV.B).

Learning Objectives

After completing this section, you will be able to:

State the IQCB minimum technical requirements for electrode placement, impedance, recording length, and usable data selection.

Explain why activation procedures are normally excluded from a resting qEEG and when their inclusion is justified.

Describe the certified reviewer's visual inspection process and apply the rule out, throw out, and refer out classifications.

Compare independent component analysis and principal component analysis as artifact rejection tools, including the limitations of each.

Explain how the Fast Fourier transform produces absolute power, relative power, symmetry, and coherence metrics.

Describe how LORETA and its variants estimate cortical sources, and identify what determines the accuracy of those estimates.

Justify the confirmatory-only role of discriminant and classification algorithms in clinical qEEG.

Listen to the Full-Length Lecture

Why Acquisition Standards Matter

Quantitative electroencephalography (qEEG) is a clinically validated methodology that uses digitized EEG signals to quantify brain activity. Its effectiveness depends on the integrity of data acquisition, inspection, and processing, as outlined in the 2025 IQCB minimum technical requirements (Collura et al., 2025). These guidelines extend foundational standards originally established by the American Clinical Neurophysiology Society (Sinha et al., 2016), offering a comprehensive framework for acquiring and processing EEG data suitable for quantitative analysis.

This unit details each stage of the qEEG workflow, from acquisition through artifact rejection to advanced source localization. Topics include the role of spectral analysis, inverse solutions such as LORETA, and the appropriate use of discriminant algorithms. The unit also addresses procedural steps recommended by IQCB for quality assurance and board certification, and closes with an overview of process flows and critical clinical applications.

These stages are fundamental for establishing the clinical reliability of qEEG in contexts ranging from neurodiagnostic screening to forensic evaluation. The sections that follow demonstrate how each component contributes to generating accurate and clinically meaningful interpretations of brain electrical activity.

Acquisition

Electrodes must be applied using the International 10-20 system, ensuring 19 standard recording sites (Sinha et al., 2016). This standardization supports anatomical consistency across sessions and subjects, facilitating accurate data interpretation and database comparisons. At least 10 minutes of recording in eyes-open (EO) and eyes-closed (EC) conditions is required, providing enough data to select 2 to 5 minutes of artifact-free EEG for the qEEG (Collura et al., 2025).

Extended recordings are sometimes necessary when the subject exhibits high levels of movement or arousal, as in cases of hyperactivity, anxiety, or sensory sensitivity. The technician must continuously monitor electrode impedance to keep it below 5 kΩ, ensure that skin preparation minimizes drift, and regularly check signal quality. During EO recording, participants are encouraged to focus on a fixed point and limit blinking, while EC recording emphasizes relaxed alertness without drowsiness.

Coaching before starting a recording helps the subject understand what may cause artifacts and how to avoid them. Continuous recording without breaks is recommended for each condition, although it is sometimes necessary to pause the recording to provide the subject with further coaching.

Activation procedures, such as photic stimulation and hyperventilation, are generally excluded to preserve a resting baseline unless specific clinical indications require their inclusion (Collura et al., 2025). Some qEEG databases have norms for activation conditions, such as those reported by Thornton and Carmody (2009), that can be used in addition to the usual resting eyes-open and eyes-closed conditions. Real-time monitoring from acquisition software helps identify problematic channels, allowing immediate corrections. The acquisition phase thus requires a blend of technical expertise, patient coaching, and adherence to standardized protocols to ensure the resulting data are both valid and diagnostically useful.

Rosa brings her 9-year-old son in for a qEEG. He is restless, the room is warm, and by minute 4 of the eyes-closed condition he is visibly drifting toward sleep. You have two choices that both look reasonable in the moment: push through and collect the full 10 minutes, or pause, re-coach, and extend the session.

The IQCB standard is not 10 minutes of recording; it is 2 to 5 minutes of clean, representative, artifact-free data. Pause the record, let him sit up and reset, and extend the acquisition. A drowsy eyes-closed segment will slow his posterior rhythm and inflate theta, and no amount of downstream processing will undo that.

Visual Inspection

Following acquisition, a certified reviewer must inspect the EEG using multiple montages, such as linked ears, bipolar, average reference, and Laplacian, to identify usable segments and anomalies (Collura et al., 2025). This review involves not only technical quality checks but also clinical interpretation of observed rhythms, waveforms, and transient events. The inspection should assess background rhythms, such as alpha during EC and desynchronization during EO, as well as abnormal discharges like spikes, sharp waves, or slow-wave activity.

The IQCB encourages reviewers to classify findings as rule out when potentially abnormal, throw out when artifactual, and refer out when medical evaluation is required, based on clinical significance and artifact profile. A well-structured report should document the overall signal quality, the quantity and type of artifact observed, the length of clean segments available, and any clinical abnormalities such as asymmetries, paroxysmal bursts, or unusual frequency components.

The reviewer should also evaluate subject-related state changes such as drowsiness or cognitive fluctuation, often indicated by vertex events, alpha dropout, or temporal slowing. IQCB certification ensures that only qualified professionals perform this crucial step, so that important clinical information is not overlooked.

The visual inspection stage acts as a bridge between raw acquisition and analytical processing. It plays a crucial role in ensuring that the qEEG reflects true neurophysiological function rather than artifact contamination or misinterpretation.

Acquisition standards exist so that the numbers you compute later describe cortex rather than contamination. Place 19 electrodes by the 10-20 system, hold impedance below 5 kΩ, record at least 10 minutes in each of the eyes-open and eyes-closed conditions, and select 2 to 5 minutes of clean, contiguous data. Leave activation procedures out of a resting qEEG unless a specific clinical question requires them. A certified reviewer must then inspect the record across multiple montages and sort every finding into rule out, throw out, or refer out.

Check Your Understanding

  1. How much recording time does IQCB require in each condition, and how much clean data must you extract from it?
  2. What impedance threshold must the technician maintain, and why does it matter for a quantitative analysis?
  3. Why are photic stimulation and hyperventilation normally excluded from a resting qEEG?
  4. Distinguish the rule out, throw out, and refer out classifications, and give an example of each.

Selection and Artifact Rejection

The artifact rejection phase uses manual editing or algorithmic tools like principal component analysis (Buzzell et al., 2022) and independent component analysis (Delorme & Makeig, 2004; Kang et al., 2018). These tools help isolate and remove common artifacts such as eye movements, blinks, muscle tension, and electrical interference.

ICA is often preferred because it can preserve good data while isolating artifactual components across channels, allowing for their targeted removal. However, the reliability of ICA depends on proper training, dataset quality, and the number of independent components relative to the number of electrodes. Thatcher and colleagues (2020) argue that the use of ICA distorts connectivity metrics even in artifact-free segments. The total recording duration selected for analysis must be 1 to 5 minutes in length, ideally contiguous and representative of baseline brain activity (Collura et al., 2025).

It is also critical to avoid stitching together too many short segments, as this can distort spectral characteristics and reduce coherence between sites. Clinically relevant transients like FIRDA, OIRDA, and posterior slow waves of youth must be preserved, as they contribute meaningful information for diagnostic interpretation.

Automated methods must be validated by visual confirmation to prevent false rejection of valid EEG features or retention of noise. Proper rejection techniques ensure that the qEEG reflects cortical activity rather than artifactual input, supporting accurate computation of metrics and interpretation. The integration of manual and automated approaches, along with careful clinical judgment, is essential to achieving high-quality artifact-free data that meet IQCB standards for clinical and research application.

We encourage you to review qEEG Tutor's coverage of normal and abnormal waveforms, and of artifacts, for more detail about how to recognize artifactual recording segments to reject.

Computation of Metrics

Standard qEEG analysis uses the Fast Fourier transform (FFT) to convert time-domain signals into frequency-domain metrics, enabling calculation of symmetry, phase, and coherence (Collura et al., 2025). These metrics quantify the amount and distribution of neural oscillations in the delta, theta, alpha, beta, and gamma bands across cortical regions.

Metrics are mapped onto topographical representations that display activity across scalp regions and allow visual interpretation of spectral abnormalities. Frequency bands should align with published standards (Sinha et al., 2016), although some variation may be justified depending on the normative database used.

Absolute power measures the raw amplitude of activity in each band, while relative power expresses the proportion of activity in one band relative to the total power. Symmetry metrics assess interhemispheric balance, and coherence measures the degree of functional connectivity between electrode sites. Phase measures the timing relationship of wave peaks of a given frequency at two different sites. Cross-device comparisons require amplifier amplitude normalization to preserve metric integrity.

Normative databases used to assess statistical deviation from the normal healthy average for an age group must be constructed with attention to demographic diversity and clinical validity, with many seeking FDA 510(k) clearance. Proper metric computation transforms EEG from a descriptive modality into a quantitative framework capable of identifying biomarkers of dysfunction and guiding treatment interventions.

Low-Resolution Electromagnetic Tomography

LORETA and its variants, sLORETA and eLORETA, estimate the cortical sources of scalp-recorded EEG by solving the inverse problem under smoothness constraints (Thatcher et al., 2005). These algorithms compute three-dimensional current source density maps that show the probable origins of EEG activity within the brain. They assume that neighboring cortical regions activate together, which enables smooth interpolation across space.

These maps align well with neuroimaging modalities like PET and fMRI, validating their utility in identifying functional abnormalities in disorders such as ADHD, depression, and epilepsy (Lantz et al., 1997). LORETA enables clinicians to visualize not just which frequencies are elevated or reduced, but where in the cortex these alterations are occurring. This adds an anatomical dimension to spectral data and supports precision treatment planning, such as neurofeedback or targeted pharmacotherapy.

The accuracy of these estimates depends heavily on the quality of input data, the spatial fidelity of the head model, and the electrode montage. When applied rigorously, LORETA enhances the interpretability of the qEEG, deepens clinical insight, and supports integration with other diagnostic modalities.

ICA and PCA

Independent component analysis and principal component analysis are essential tools in modern EEG preprocessing, particularly for artifact removal and signal decomposition. PCA works by reducing the dimensionality of data, identifying principal components that account for the most variance, and allowing for the elimination of those that correspond to noise. This approach is helpful when broad systemic artifacts are present, but it comes at the cost of losing signal from all channels in the affected epochs.

ICA, on the other hand, assumes statistical independence among signal sources and separates mixed signals into constituent components. This allows specific artifact sources, such as eye blinks, cardiac signals, or muscle noise, to be isolated and removed without discarding entire channels.

The success of ICA depends on the amount and quality of data and on the experience of the user in recognizing artifact components from cortical sources. Moreover, ICA-processed data must be consistent with the preprocessing used in the normative database for valid comparison. Machine learning and template-based artifact recognition are sometimes integrated with ICA to increase specificity, but manual inspection remains critical to validate the decomposition results.

Improper use of ICA can result in discarding clinically relevant signals or retaining contamination, which can skew qEEG metrics. Both ICA and PCA must therefore be implemented judiciously, ideally under the supervision of certified experts, to ensure that their contribution to data quality and analytical fidelity aligns with IQCB standards.

Discriminants and Classification Algorithms

Discriminant analysis refers to statistical methods that distinguish between clinical groups based on multivariate EEG features. These algorithms generate composite scores that indicate the likelihood that a subject belongs to a diagnostic category such as mild cognitive impairment, depression, or ADHD. They are often developed using large datasets in which group membership is known, and features such as spectral power, asymmetry, or coherence are used to build predictive models.

IQCB guidelines emphasize that such models must only be used for confirmatory purposes and never as primary diagnostic tools, because of their sensitivity to false positives, particularly in diverse populations (Collura et al., 2025). These tools are best employed when there is an existing diagnostic hypothesis and sufficient clinical justification. Misapplication of these models in unscreened populations could lead to diagnostic error or unnecessary intervention.

Classification algorithms based on machine learning or artificial intelligence require rigorous training, cross-validation, and external testing before clinical application. The integrity of input data, consistency with training conditions, and adherence to ethical review are essential for defensible use.

When employed properly, discriminants can streamline assessment, support diagnosis, and predict treatment response. However, they must always be accompanied by human oversight and contextual clinical knowledge to maintain their validity and reliability.

Artifact rejection combines manual editing with PCA and ICA, and every automated decision needs visual confirmation. Preserve clinically meaningful transients such as FIRDA, OIRDA, and posterior slow waves of youth, and avoid stitching many short segments together. The FFT converts your clean segments into absolute power, relative power, symmetry, coherence, and phase, which are then compared against an age-regressed normative database. LORETA adds an anatomical dimension by estimating where in the cortex the activity arises. Discriminant and classification algorithms are confirmatory tools only, appropriate when a diagnostic hypothesis already exists.

Check Your Understanding

  1. Compare PCA and ICA: what does each remove, and what does each cost you?
  2. Why must ICA preprocessing match the preprocessing used to build the normative database?
  3. Distinguish absolute power from relative power, and explain when each is more informative.
  4. What three factors most affect the accuracy of a LORETA source estimate?
  5. Why does IQCB restrict discriminant algorithms to a confirmatory role?

Recommended Procedure for EEG Recording for qEEG Evaluation

The IQCB has laid out a detailed procedural protocol for EEG recording to ensure consistency and suitability for qEEG analysis (Collura et al., 2025). The procedure begins with obtaining informed consent and clinical history, establishing a quiet recording environment, and ensuring that all electronic devices are removed to minimize interference.

Electrode application follows, with impedance checks ensuring that all sensors meet the resistance threshold. Participants are coached on behavior during EO and EC recordings, including maintaining a fixed gaze and minimizing movement. EO recording is typically conducted first to avoid drowsiness in the EC data, which are collected afterward.

Continuous observation during both sessions allows the technician to note EMG activity, artifacts, and behaviors of interest. Upon completion, visual inspection should be carried out across at least three montages to verify data quality and identify segments suitable for analysis. The goal is to collect 2 to 5 minutes of artifact-free EO and EC data for further processing.

This rigorous procedure supports the collection of high-quality, interpretable EEG data, facilitating accurate and standardized qEEG assessment in clinical and research settings.

Process and Data Flow for qEEG Evaluation

The qEEG evaluation process is systematic, beginning with raw EEG acquisition under standardized conditions and progressing through data cleaning, generation of metrics, and analysis (Collura et al., 2025). Once EEG data are collected, the technician or clinician performs artifact rejection using visual inspection and computational methods. Clean data, ideally 2 to 5 minutes in length, are then processed using FFT to derive spectral metrics.

These metrics are compared against age-regressed normative databases to identify deviations, often visualized in tables and color-coded topographic maps. Additional analyses may include asymmetry, phase coherence, and source localization using LORETA.

The results are compiled into a report that includes raw data review, metric analysis, and clinical interpretation. This report can guide treatment planning, monitor progress, or support diagnostic decisions.

The process flow ensures that every phase, from acquisition through inspection, selection, computation, and interpretation, is executed with precision and consistency. Flow diagrams reinforce this structure, making procedures transparent and replicable across settings. Adhering to this standardized workflow enhances the scientific reliability and practical utility of qEEG in clinical practice.

Conclusion

The 2025 IQCB guidelines mark a major advancement in standardizing qEEG practice, ensuring data quality and clinical reliability (Collura et al., 2025). The integration of ACNS standards (Sinha et al., 2016), spectral methods, and advanced source localization expands the utility of qEEG across neuropsychiatric contexts.

Tools like ICA, PCA, and LORETA enhance analysis when used properly, while procedural protocols ensure data integrity. The emphasis on artifact management, certification, and validated metrics supports accurate interpretation and reinforces the role of qEEG as a complement to traditional clinical EEG.

Standardized data flow models and process recommendations increase reproducibility and support training and certification. Collectively, these advances strengthen the scientific and clinical foundation of qEEG and improve its capacity to aid diagnosis, inform treatment, and contribute meaningfully to neurobehavioral healthcare.

Check Your Understanding

  1. Why is the eyes-open condition normally recorded before the eyes-closed condition?
  2. Across how many montages should visual inspection be performed before you select segments?
  3. Trace the qEEG data flow from raw acquisition to clinical report, naming each stage.
  4. What does amplifier amplitude normalization accomplish, and when is it required?

Cutting-Edge Topics in qEEG Research

The 2025 IQCB Requirements Raise the Floor

Collura and colleagues (2025) published minimum technical requirements that give the field something it has long lacked: a single, citable procedural standard for clinical qEEG. The practical consequence is that a qEEG report can now be evaluated against a published benchmark rather than against a reviewer's individual habits, which matters most in forensic and insurance contexts. Expect referral sources and courts to begin asking whether a given recording met these requirements, and expect reports to start documenting compliance explicitly.

Does ICA Distort What It Cleans?

ICA is the field's default artifact tool, but Thatcher and colleagues (2020) reported that ICA artifact correction distorts EEG phase even in segments that contained no artifact to begin with. If that finding holds, connectivity and coherence metrics computed after ICA may carry a systematic bias that no amount of visual confirmation would reveal. The conservative response is to match your preprocessing to the normative database exactly, and to treat post-ICA connectivity findings with more caution than post-ICA power findings.

Machine Learning Meets the Confirmatory-Only Rule

Classification algorithms built on machine learning can achieve impressive accuracy on the datasets that trained them, and much less on the clients in your office. IQCB's insistence that these models remain confirmatory reflects a base-rate problem rather than a technical one: applied to an unscreened population, even a highly accurate classifier produces mostly false positives. The research frontier here is not better accuracy but better calibration, external validation, and honest reporting of performance in the populations where the tool will actually be used.

Activation Norms as an Alternative to Resting Baselines

Resting eyes-open and eyes-closed baselines are the standard because they are reproducible, not because they are the most informative. Thornton and Carmody (2009) developed activation databases that compare a client's EEG during cognitive tasks against normative task data, which can surface deficits that a resting record misses entirely. As more databases add task norms, the resting-only qEEG may come to look like a starting point rather than a complete assessment.

Assignment

Now that you have completed this section, create a flowchart of the process by which you acquire the qEEG. Mark on your flowchart the point at which each IQCB requirement is satisfied, and note where in your own workflow a failure would be hardest to detect after the fact.

Glossary

absolute power: the raw amplitude of EEG activity within a given frequency band, expressed independently of activity in other bands.

acquisition: the process of recording the EEG using standardized electrode placement and controlled conditions to ensure consistent and reliable data suitable for quantitative analysis.

artifact: any non-neural signal contaminating EEG data, including movements, eye blinks, muscle activity, and electrical interference.

coherence: a metric that quantifies the degree of synchronization between EEG signals from two electrode sites, often used to assess functional connectivity.

discriminant analysis: a statistical method that uses multiple EEG features to predict group membership, such as diagnostic categories.

eyes-closed (EC): a condition in EEG acquisition where subjects close their eyes, enhancing alpha rhythm visibility and reducing visual input.

eyes-open (EO): a condition in EEG acquisition where subjects keep their eyes open, typically used to contrast with EC states and monitor attentional modulation.

Fast Fourier transform (FFT): an algorithm that converts time-domain EEG signals into frequency-domain data, allowing computation of spectral power within predefined bands.

frequency band: ranges of EEG signal frequency associated with different brain states, typically including delta, theta, alpha, beta, and gamma bands.

impedance: the resistance to electrical current at the electrode-scalp interface, which must be minimized for accurate EEG recording.

independent component analysis (ICA): a computational technique that separates mixed signals in EEG data into statistically independent components, facilitating artifact removal.

International 10-20 system: a standardized method for electrode placement on the scalp used to ensure consistency in EEG recording across individuals and studies.

IQCB: the International QEEG Certification Board, which establishes standards and certification procedures for clinical qEEG practice.

LORETA: low-resolution electromagnetic tomography; a technique that estimates the cortical sources of scalp-recorded EEG activity by solving the inverse problem with spatial smoothing constraints.

metric: a quantitative measure derived from EEG data, such as absolute power, relative power, symmetry, or coherence.

montage: a configuration of EEG electrode pairings used for data display and analysis, such as linked ears, average reference, or bipolar setups.

normative database: a reference dataset of EEG metrics collected from healthy individuals, used for statistical comparison in qEEG analysis.

phase: a metric expressing the timing relationship between wave peaks of a given frequency recorded at two different sites.

photic stimulation: a procedure using rhythmic light flashes during EEG recording to provoke specific brain responses, often excluded from a resting-state qEEG.

principal component analysis (PCA): a method for dimensionality reduction in EEG data that identifies uncorrelated variables, or components, explaining the most variance.

qEEG: quantitative electroencephalography; the mathematical and statistical analysis of digitized EEG signals to derive metrics indicative of brain function or dysfunction.

relative power: the proportion of spectral power in a given frequency band relative to the total power across all bands.

rule out, throw out, refer out: categories used in visual inspection of EEG data to classify findings as potentially significant, artifactual, or requiring clinical referral.

source localization: the process of estimating the origin of EEG activity within the brain, often using algorithms like LORETA.

symmetry: a qEEG metric assessing the balance of power or activity between homologous regions of the two hemispheres.

transients: brief EEG events that deviate from background activity, which may be physiological or pathological, such as vertex waves or epileptiform discharges.

Test Yourself on ClassMarker

Click the button below to take a 10-question exam over this entire unit. There is no password.

Test Yourself on ClassMarker

Review Flash Cards on Quizlet

Click the button below to review our chapter flash cards.

Review Flash Cards on Quizlet

References

Buzzell, G. A., Niu, Y., Aviyente, S., & Bernat, E. (2022). A practical introduction to EEG time-frequency principal components analysis (TF-PCA). Developmental Cognitive Neuroscience, 55, 101114. https://doi.org/10.1016/j.dcn.2022.101114

Collura, T., Cantor, D., Chartier, D., Crago, R., Hartzoge, A., Hurd, M., Kerson, C., Lubar, J., Nash, J., Prichep, L. S., Surmeli, T., Thompson, T., Tracy, M., & Turner, R. (2025). Guideline minimum technical requirements for performing clinical quantitative electroencephalography. Clinical EEG and Neuroscience, 1–9. https://doi.org/10.1177/15500594241308654

Delorme, A., & Makeig, S. (2004). EEGLAB: An open source toolbox for analysis of single-trial EEG dynamics including independent component analysis. Journal of Neuroscience Methods, 134(1), 9–21. https://doi.org/10.1016/j.jneumeth.2003.10.009

Kang, G., Jin, S.-H., Kim, D. K., & Kang, S. W. (2018). EEG artifacts removal using machine learning algorithms and independent component analysis. Clinical Neurophysiology, 129, e24. https://doi.org/10.1016/j.clinph.2018.04.060

Lantz, G., Michel, C. M., Pascual-Marqui, R. D., & Spinelli, L. (1997). Extracranial localization of intracranial interictal epileptiform activity using LORETA. Electroencephalography and Clinical Neurophysiology, 102(5), 414–422. https://doi.org/10.1016/S0921-884X(96)96551-0

Sinha, S. R., Sullivan, L., Sabau, D., Dombrowski, K. E., Halford, J. J., Gloss, D., Herman, S. T., Kanner, A. M., LaFrance, W. C., & Lambrakis, C. F. (2016). American Clinical Neurophysiology Society Guideline 1: Minimum technical requirements for performing clinical electroencephalography. Journal of Clinical Neurophysiology, 33(4), 303–307. https://doi.org/10.1097/WNP.0000000000000308

Thatcher, R. W., North, D., & Biver, C. (2005). Evaluation and validity of a LORETA normative EEG database. Clinical EEG and Neuroscience, 36(2), 116–122. https://doi.org/10.1177/155005940503600211

Thatcher, R. W., Palmero-Soler, E., & Otte, G. (2020). Independent components analysis "artifact correction" distorts EEG phase in artifact free segments. Journal of Neurology and Neurobiology, 6, 1–14. https://doi.org/10.16966/2379-7150.172

Thornton, K. E., & Carmody, D. P. (2009). Eyes-closed and activation QEEG databases in predicting cognitive effectiveness and the inefficiency hypothesis. Journal of Neurotherapy, 13, 1–21. https://doi.org/10.1080/10874200802429850

Return to Top