AmJ Psychiatry 148:7, July 1991 961
APA Official Actions

Quantitative Electroencephalography: A Report
on the Present State of Computerized EEG Techniques
American Psychiatric Association Task Force
on Quantitative Electrophysiological Assessment

The American Psychiatric Association established the Task Force on Quantitative Electrophysiological Assessen1
in May I 989 to consider several questions of interest to psychiatrists. From the inception of techniques for
quantitative electroencephalography (qEEG), the possibility was raised that it would substantially assist in the diagnosis
of brain disorders. However, controversy exists over the scientific basis for such approaches and the training
necessary for interpretation of computerized records. The task force’s charge was to report 1) the present state of
scientific knowledge about qEEG, 2) the role for qEEG in clinical psychiatric practice at the present time, 3) the
training necessary for the use ofqEEG techniques, and 4) the possible future ofthe technique in the study of mental
disorders. The task force concluded that qEEG is particularly usefulfor the detection ofabnormalities in slow waves,
which are a feature of delirium, dementia, intoxication, and other syndromes involving gross CNS dysfunction. The
ability ofqEEG to help in the diagnosis ofother disorders, such as schizophrenia or depression, is not yet established.
Clinical replications and sharing ofnormative and patient data bases are necessary for the advancement ofthis field.
Proper use of this technique requires extensive training in a center experienced in its use. Standards for training and
for using the technology in psychiatry are urgently needed.
This report was approved by the Board of Trustees in December 1990.

The American Psychiatric Association (APA) established the Task
Force on Quantitative Electrophysiological Assessment in May 1989
to consider several questions of interest to psychiatrists. From the
inception of techniques for quantitative electroencephalography
(qEEG), the possibility was raised that it would substantially assist in
the diagnosis of brain disorders ( 1 , 2). However, controversy exists
over the scientific basis for such claims, the value of computerized
measurements versus traditional EEG approaches, and the training
necessary for interpretation of computerized records (3, 4). Unfortunately,
advertisements and promotional material from some manufactuners
of qEEG instruments have gone beyond the existing scientific
evidence to make claims of diagnostic utility. The task force’s
charge was to report 1 ) the present state of scientific knowledge about
qEEG, 2) the role for qEEG in clinical psychiatric practice, 3) the
training necessary for the use of qEEG techniques, and 4) the possible
future of the technique in the study of mental disorders. This report
will review the technique itself and then discuss the points addressed
by the task force.

Electrical activity originating in the brain can be recorded from the
surface of the scalp by using conventional EEG techniques to provide
I The members of the task force are Robert Freedman, M.D. (chairperson),
Daniel J. Luchins, M.D., Robert W. McCarley, M.D., and
John M. Morihisa, M.D. The task force appreciates the input provided
by Robert Cancro, M.D. The task force was staffed by Harold
Alan Pincus, M.D., and Wendy Davis, M.Ed.
1 ) analysis of the spontaneous EEG and 2) analysis of the brain activity
evoked by a stimulus. Recordings of at least three kinds of brain
electrical activity have clear clinical utility. The sudden activation of
large groups of neurons, as occurs during epilepsy, is reflected in abnormal
spike activity. Disturbances in the normal cerebral cortical
rhythms, as occurs in patients with dementia and other organic brain
syndromes, are reflected in decreased frequency and increased prominence
of slow waves in the EEG. Disturbances in specific brain pathways,
such as the optic neuritis of multiple sclerosis, are reflected in
changes in sensory evoked potentials. Electrical activity is generally
recorded from many sites over the scalp to assist in localizing changes
in underlying brain areas. The activity from each electrode is plotted
on a chart record, which can be voluminous because of the multiple
recording sites and the many minutes of recording time. Analysis by
an electroencephalognapher requires knowledge of the normal patterns
of activity of the brain in various behavioral conditions, such as
waking and sleep, as well as recognition of artifacts caused by factors
such as body movement and electrical interference.
Quantitative EEG involves the use of digital computer technology
to improve the analysis of the EEG in three ways. First, a power spectrum
is calculated for each recording site. This approach, a variant of
Fourier analysis, treats the EEG as the sum of waves of various frequencies,
i.e., slower waves and faster waves summed together. The
power spectrum quantitates the amount of activity at each frequency.
From such a spectrum, one can determine if a particular type of wave,
such as alpha, is present to a normal degree. This method is a potential
improvement over looking at the LEG waves on paper records; in that
case, various waves are mixed together, so no quantitative assessment
can be made. For example, small amounts of EEG slowing are often
difficult to detect in manual analysis. On the other hand, the computer
method has the potential drawback that only a bniefsegment is shown,

962 Am J Psychiatry 148:7, July 1991
whereas conventional EEG records display many minutes of activity,
which provide an opportunity to detect rare abnormal events.
The second technique provides a visual image in the form of a map
over the surface of the head. These colorful maps attract attention in
research papers, clinical reports, and advertisements. The computer
constructs these maps by selecting a particular frequency and color
coding the amplitude in the power spectrum for each EEG lead. Hot
colors, such as red, represent high amplitude, and cool colors, such
as blue, represent low amplitude. Between the recording sites, the
computer interpolates values to provide a continuous map over the
entire head. Thus, if the alpha activity of a normal subject is plotted,
one might see a red (hot) area over the posterior head near the occipital
lobes. In the initial reports of this technique it was called “brain dcctnical
activity mapping” (BEAM) ( 1 ). The potential advantage of this
approach is that patterns such as hemispheric asymmetry can be more
easily spotted because one sees the data displayed over the surface of
the head in color. An abnormally high frequency of slow waves over
the frontal lobes, found in some dementias, might be difficult to see
in paper records but would be obvious on a color-coded map.
The third strategy is to compare the maps, or the electrical activity
on which they are based, by various statistical techniques. An individual
patient’s values at each point on the scalp are compared with those
of a known group, either normal subjects or subjects with a specific
illness, such as schizophrenia. The values are often analyzed statistically
by calculating the z score, the difference between the patient and
the comparison group, divided by the group’s standard deviation.
These z scores are then color coded to form a new map, which emphasizes
areas in which the patient differs from the comparison group.
These “significance probability maps” are also frequently displayed
in advertisements. A problem is that these z statistic maps involve
multiple comparisons and thus can suggest statistical significance
when it is not present.
A more sophisticated approach is the use of multiple discriminant
analyses. Quantitative EEG data for groups of normal and mentally
disordered subjects are recorded to form qEEG data bases, in which
each subject is represented by many recording sites and each site has
different amplitude values for many frequencies. Given the complexities
of brain function, the disturbances that are the pathophysiological
bases of mental disorders may not be apparent in a single power spectnum
or even in maps of particular frequencies. The discriminant
analysis approach allows the construction of complex functions that
would discriminate between groups, by summing very small distinctions
into larger, more statistically significant differences. An analogy
could be drawn to psychiatric diagnosis itself, in which single symptoms
themselves carry little weight but a combination - e.g., depressed
mood, weight loss, sleep disturbance, decreased energy, and ideas of
worthlessness - form a complex with much higher predictive value.
The potential advantage of this approach is that, given an adequate
data base, the computer might be able to determine if an individual
had a qEEG that was typical ofa particular clinically defined disorder,
if it were assumed that each illness has a unique pattern of brain activity
as determined by scalp recording.
One potential disadvantage is that multiple discriminant analysis
cannot determine if the features it extracts are of fundamental biological
or psychological importance. This limitation is particularly
troublesome because EEG measures themselves are only indirect
measures of neuronal activity and 1 ) are subject to subtle artifacts
from nonneuronal sources and 2) may represent trivial correlates of
the disorder in the population being examined. An example ofa trivial
correlation would be a high prevalence of higher EEG frequencies in
a group of acutely psychotic schizophrenic patients, which might simply
be due to their high level of stress. From the data on these patients,
the computer might categorize all patients with high-frequency activity
as schizophrenic. However, this feature is not typical of schizophrenic
patients in remission and can be found in anxious nonschizophrenic
patients as well. Only a very large body of data, carefully
collected from many centers and analyzed with rigorous clinical and
statistical methods and proper control for artifacts, can prevent such
adventitious, but ultimately spurious findings.
Critical features of allstatistical comparisons are the quality of the
recordings and the actual normality of the normal comparison group.
Many instrument manufacturers offer a normative data base for cornparisons
with the patient data. However, these data bases are of vanable
quality and, because of their proprietary origin, have generally
not been subjected to the usual scientific scrutiny. Details about diagnostic
procedure, recording technique and artifact control, and the
distribution of individual scores are inadequate. Many of these deficiencies
could be lessened if the missing information and the raw data
from the individuals constituting the data base were provided. If these
features were available and if the data bases were in the public domain,
normal peer review would highlight any biases and deficiencies
in the data base. The task force is currently unable to endorse any
available data base because this information is lacking.

The qEEG technique has been extensively used to study physiological
changes in the brains of patients with neurological and mental
illnesses. Examples of the more than I 00 studies that have used these
techniques are comparisons of patients with normal subjects that
indicated more delta waves in schizophrenic patients (5), differences
in beta activity in alcoholics (6), and more delta and theta waves in
patients with Alzheimer’s dementia (7). Research with the mapping
technique for evoked potentials has also yielded useful findings. For
example, the maps and multivaniate feature characterization demonstrate
focal abnormalities in left temporal scalp regional activity in
schizophrenia that are associated with temporal lobe structural abnormalities
(8). Frontal scalp regional abnormalities are present in
obsessive-compulsive disorder ( 9). The mapping technique has also
been used to demonstrate actions of psychotropic drugs by measuring
their effect on brain activity (10).
Most of these studies have used qEEG to help define and localize
a particular abnormality in brain activity as part of the study of the
pathophysiology of a specific illness. However, only a few studies have
suggested that qEEG findings correlate with clinical diagnoses. Such
studies have not concentrated on particular pathophysiological features
but, rather, have used discniminant analysis. Shagass et al. ( I 1)
found that the variables derived from EEG recordings could discnirninate
between some illnesses, diagnosed initially with DSM-II criteria.
For example, latent schizophrenia, which they equated with DSM-III
borderline and schizotypal personality disorders, was distinguished
from neuroses with a sensitivity of 60% and a specificity of 92%.ýIf
the two disorders were equally frequent, the computer could match
the clinical diagnosis in 88% ofthe cases. Shagass et al. suggested that
this technique might have particular value for distinctions between
pairs of conditions that are difficult to resolve clinically. John et al.
(2, 12) described a similar strategy for discniminant analysis with the
qEEG technique, which they termed “neurometrics.” Using discriminant
analysis, they were able to construct a formula that classified 212
subjects as normal or as having primary depression, alcoholism, or
dementia with concordances to clinical diagnosis ranging from 72%
to 80%. A replication with 166 subjects achieved similar results with
the same formula. The technique was also used to separate unipolar
and bipolar depressed subjects; 85% of each group were assigned to
the appropriate clinical categories. Whether or not these distinctions
can be generalized to nonresearch conditions has not been established.

Various instruments are now commercially available for qEEG determinations.
These instruments, with appropriately trained operatons,
are capable of producing high-quality EEG records, accurate
power spectra, and color-coded maps. These maps are useful for determination
of abnormal slow wave activity. They are less useful for
detection of seizure activity. The instruments used for different functions
have similar hardware but different software. Also, the manufacturers
differ in their advertised claims ofclinical utility. At this time,
the ability of any qEEG procedure to make psychiatric diagnoses or
to discriminate between various groups of psychiatric patients and
normal subjects is not well established.
The most valuable clinical role seems to be the evaluation of conditions
likely to involve slow wave abnormalities, i.e., stroke, dementia,
delirium, and intoxication. EEG recordings have traditionally
been part of the evaluation of these conditions, although one survey
of psychiatric practice suggests that the results are rarely relied on for
differential diagnosis (13). In only one study were readings of paper
EEG records by a clinical electroencephalographer compared with
interpretation of a qEEG map (14). In that study the qEEG was
slightly more sensitive than routine EEG for detection of abnormalities
in I 00 consecutive patients referred for recording.

There are no generally agreed on standards for the training of psychiatrists
who perform qEEG or interpret qEEG records. Artifact-free
recording requires experience in electrode application and elimination
of common artifacts, such as artifact from ECG, muscle tension, or
eye movements. This technical ability is likely to require performance
of SO or more recordings under the supervision of experienced personnel.
Interpretation of the qEEG record requires substantial clinical
experience. Extensive experience in a center devoted to such recordings
and clinical interpretation of qEEG records is necessary. The user
of these data must understand that the increased power of these techniques
also increases their potential for misinterpretation (4, 15, 16).
Training should include comparison of qEEG records with other data
and with clinical outcome. Because qEEG itself contributes only limited
information of direct clinical significance, persons otherwise not
qualified to perform differential diagnoses of mental disorders are not
qualified to make diagnoses with qEEG. Because the current uses of
qEEG and the current uses of conventional EEG have somewhat different
purposes and advantages, it is not clear that all users of qEEG
need to be certified electroencephalographers ( 1 7), although some
electroencephalographers have suggested such a requirement ( 18).
However, the qEEG user must know the strengths and limitations of
qEEG compared to conventional EEG. Although weekend training
courses are offered by several of the instrument manufacturers, these
brief sessions are not sufficient for learning how to perform qEEG and
interpret qEEG records. Instead, appropriate training programs and
certification criteria for qEEG are needed for psychiatrists who wish
to use the technique. Additional efforts should be made to provide
general education about the technique in residency training and in
postgraduate courses.
There is also no certification by the U.S. Food and Drug Administration
or other governmental agency of the instruments themselves
or of the indications for their use in clinical practice. Their proper use
in an individual clinical practice is therefore the ethical responsibility
of the physician. In addition to this report, there are similar reports
from other professional societies concerned with the use ofqEEG (19,
20). Although they differ in specific recommendations, these reports
express similar concerns about establishment of scientific and clinical
bases for the use of qEEG and about proper standards for the training
of those who use the techniques.

Quantitative EEG is a technique of great promise that should have
a place along with other brain imaging techniques in research on brain
illnesses. In conjunction with magnetoencephalographic recording
(MEG), it has potential usefulness in the localization of the sources of
abnormalities in brain electrical activity, which are part of the
pathophysiology of many brain disorders. Findings obtained with
qEEG have greater power when they are correlated with other biological
features, such as underlying structural changes revealed by
other techniques, such as computed tomography (21) and magnetic
resonance imaging. Such investigations will probably increase our understanding
of the pathogenesis of mental disorders and thereby perhaps
lead to better diagnoses. The technique may also have utility in
the monitoring of response to psychotropic agents and other treatments.
Psychiatrists hope that the successful use of qEEG for detection
of slow wave abnormalities in organic brain disorders may be cxtended
some day to the detection of abnormal features in such illnesses
as schizophrenia and affective disorder. Such investigations may lead
to better diagnostic systems and more objective criteria for classification.
These investigations are already being conducted in several re-
AmJ Psychiatry 148:7,July 1991 963

search centers with support from the National Institute of Mental
Health, the Medical Research Service of the U.S. Department of Veterans
Affairs, and other sources. As with most other research efforts,
their quality is monitored by scientific peer review, both at the time
of funding and at the time of publication of the results in scientific
Quantitative EEG has been proposed as a diagnostic aid for psychiatnic
practice. It has particular utility for the detection of abnormalities
in slow waves, which are a feature of many organic brain
The ability of qEEG to assist in the diagnosis of other disorders,
such as schizophrenia or depression, is not yet established. Clinical
replications and sharing of normative and patient data bases are necessary
for the advancement of this field. Proper use of this technique
requires extensive training in a center experienced with its use. Establishment
of standards for training and for the use of the technique in
psychiatry are urgently needed.

1. Duffy FH, Burchfiel JL, Lombroso CT: Brain electrical activity
mapping (BEAM): a method for extending the clinical utility of
EEG and evoked potential data. Ann Neurol 1979; 5:309-321
2. John ER, Pnichep IS, Easton P: Neurometnics: computer assisted
differential diagnosis of brain dysfunctions. Science 1988; 293:
3. Duffy FH: Clinical value of topographic mapping and quantified
neurophysiology. Arch Neurol 1989; 46:1133-1136
4. Nuwer MR: Uses and abuses of brain mapping. Arch Neurol
1989; 46:1134-1136
S. Monihisa JM, Duffy PH, Wyatt RJ: Brain electrical activity mapping
(BEAM) in schizophrenic patients. Arch Gen Psychiatry
1983; 40:719-728
6. Pollock VE, Volavka J, Goodwin DW, et al: Pattern reversal visual
evoked potentials after alcohol administration among men at
risk for alcoholism. Psychiatry Res 1988; 26:191-202
7. Duffy FH, Albert MS, McAnulty G: Brain electrical activity in
patients with presenile and senile dementia of the Alzheimer type.
Ann Neurol 1984; 16:439-448
8. McCarley RW, Faux SF, Shenton ME, et al: CT abnormalities in
schizophrenia: a preliminary study of their correlations with
P300/P200 electrophysiological features and positive/negative
symptoms. Arch Gen Psychiatry 1989; 46:689-708
9. Malloy P, Rasmussen 5, Braden W, et al: Topographic evoked
potential mapping in obsessive-compulsive disorder: evidence of
frontal lobe dysfunction. Psychiatry Res 1989; 28:63-71
10. Itil FM: The significance of quantitative pharmaco-EEG in the
discovery and classification of psychotropic drugs, in Electroencephalography
in Drug Research. Edited by Herrrnann WM.
New York, Gustav Fischer Stuttgart, 1982
1 1 . Shagass C, Roemer RA, Straumanis JJ, et al: Psychiatric diagnostic
discriminations with combinations of quantitative EEG vanables.
Br J Psychiatry 1984; 144:581-592
12. John ER, Karmel BZ, Corning WC, et al: Neurometnics: numenical
taxonomy identifies different profiles of brain functions
within groups of behaviorally similar people. Science 1 977; 196:
13. Lam RW, Hurwitz TA, Wada JA: The clinical use of EEG in a
general psychiatric setting. Hosp Community Psychiatry 1 988;
14. Jernett SA, Corsak J: Clinical utility of topographic EEG brain
mapping. Clin Electroencephal 1988; 19:134-143
is. Kahn EM, Weiner RD, Brenner RP, et al: Topographic maps of
brain electrical activity - pitfalls and precautions. Biol Psychiatry
1988; 23:628-636
16. MonihisaJM: Computerized EEG and evoked potential mapping,
in BrainlmagingApplications in Psychiatry. Edited byAndneasen
NC. Washington, DC, American Psychiatric Press, 1989
17. Cancro R: The art of turf creation (letter). Arch Gen Psychiatry 20. American Academy of Neurology: Assessment: EEG Brain Map-
1989; 46:191-194 ping. Minneapolis, American Academy of Neurology, 1989
18. Nuwer MR: Quantitative electroencephalography (letter). Arch 21. Morihisa JM, McAnulty GB: Structure and function: brain
Gen Psychiatry 1987; 44:840 electrical activity mapping (BEAM) and computerized tomo-
19. American Electroencephalographic Society: Statement on clinical graphy (CT scan) in schizophrenia. Biol Psychiatry 1985; 20:3-
use of quantitative EEG. J Clin Neurophysiol 1987; 4:75 19
964 Am J Psychiatry 148:7, July 1991