Downloaded from http://jnnp.bmj.com/ on June 17, 2017 - Published by group.bmj.com
527
J7ournal of Neurology, Neurosurgery, and Psychiatry 1 992;55:527-529
7ou/r,hl of
NEUROLOGY
NEUROSURGERY
& PSYCHIATRY
Editorial
Brain mapping
a
useful tool
or a
Spatial distribution is an important consideration in the
interpretation of the electroencephalogram (EEG) and
cerebral evoked potentials (EPs). It is, however, difficult to
determine topography from a conventional chart
recording. Each channel provides a two-dimensional
representation of potential difference between two
electrodes as a function of time, and the interpreter has to
visualise the topography by integrating information across
channels. The task can be assisted by the use of brain
mapping, which generates coloured or shaded contour
maps derived from cerebral electrical activity. The maps
may display various features, either instantaneous or
averaged over time, such as amplitude and polarity or
frequency and power, or statistical derivatives, relating for
instance to the probability of the activity at each site being
found in a normal reference population.
Most of us feel intuitively familiar with cartography, and
the presentations are visually attractive, with regions of
high power or electrical negativity shown in red or white,
and areas of positive potential or low power shaded blue or
black. A range of well-designed brain mappers are now
commercially available. Apply the electrodes and plug
them into the head box, and a few keystrokes, guided by
user-friendly menus, suffice to capture a segment of EEG
or to elicit EPs and present the result as a map, the
interpretation of which offers no apparent difficulty.
Surely, this signals the end of the arcane tradition of
subjective interpretation of multichannel paper charts by
clinical neurophysiologists who lay claim to special visual
skills.
Numerical analysis of cerebral electrical activity has a
long and respectable tradition. As early as the 1930s,
Deitsch,' followed by Grass and Gibbs2 and Drohocki3
were applying Fourier analysis to the EEG. Walter4
described an analogue frequency analyser in 1943 and
later Walter and Shipton"6 developed the forerunner of
automated topographic displays in the "toposcope". Since
the early work of Brazier7 and the elegant
chrono-topograms of R&mond,89 the appearance of
microcomputers with colour graphics has accelerated the
development by Duffy' " and many others'2 of
techniques for quantification and topographic display, and
made these available as affordable commercial products.
In its simplest form, brain mapping presents an
isopotential display of the electrical field over the head.
This may be used to study the topography of a particular
feature at an instant in time, at the culmination of a spike
for instance, or the average of a recurring event, such as an
evoked potential. Such a map can provide no information
not available from careful inspection of the original
dangerous toy?
recording. However, isopotential mapping may highlight
characteristics not obvious to the observer, for instance by
drawing attention to important but inconspicuous features
of a transient event. Clinically significant examples include
the mid-frontal positivity of the typical Rolandic spike in
benign childhood epilepsy'3 and a diffuse contralateral
central positivity accompanying temporal spikes,
specifically when these arise from mesial temporal
lesions.'4 If the event occurs asynchronously at different
sites, which is often the case with EP components,
mapping is again a powerful tool, as is well illustrated by its
use to demonstrate the distinct origins of the N20 and P22
components of the somatosensory EP. '5
Various data manipulations may be used to enhance
isopotential maps. Multiple examples of EEG transients
may be manually selected and averaged (after the manner
of EPs) to reduce noise, and mapping of the average may
then show subtle features, patterns of propagation for
instance, not recognisable in the primary tracing. '3 In
conventional EEG recordings, source reference derivation
provides a display with properties intermediate between
those of common reference and bipolar recording. The
display is in a sense reference-free in that there is no single
reference common to all electrodes, and local potential
gradients are emphasised relative to more widespread
fields. Mapping has been applied to source reference
recordings, with interesting results.'6 '7 Consistent, fine
details may be shown up which are not readily detected
with other references; whether these are biologically
meaningful, or simply represent an artifact of the source
reference transform, is perhaps open to dispute. More
direct reference-free measurements of local activity (that
is, potential gradients or currents) can be obtained by
calculating spatial variance or global field power, 1 or
current density." This too appears a promising approach
which highlights the details of focal transients.
One important use of EEG quantification, for instance
by spectral analysis, is data reduction, which allows a long
record to be described by a few numerical values. These
may, of course, be subjected to statistical analysis for a
variety of purposes including clinical decision-making, but
can also be useful as an aid to visual EEG interpretation.
For this purpose, a set of maps of total power in the
standard frequency bands is a convenient form of
presentation, assisting visual assessment of overall EEG
abnormality, and highlighting asymmetries or more
localised anomalies.
Some workers have taken mapping of derived features a
step further and display statistical measures of the
differences between the values obtained in a particular
Downloaded from http://jnnp.bmj.com/ on June 17, 2017 - Published by group.bmj.com
528
record and those of a reference population. The claims for
the diagnostic power of "significance probability
mapping"" are impressive but so also are the counter
arguments.8 These techniques form part of the
armamentarium of the as yet controversial approach to the
quantitative study of cerebral electrical activity termed
"neurometrics".
Major problems, both biological and technical, belie the
apparent simplicity of brain maps. The similarity of brain
mapping to CT, MRI and PET scans is illusory. In
neuroimaging, whether anatomical or functional, there is a
close and direct correspondence between the affected
structure and the image. However, topographic changes in
electrical activity have a more complex and poorly
understood relationship to cerebral pathology and
function. To take but one example, when a cerebral lesion
produces local hypofunction and more widespread
dysfunction, the greatest amount of abnormal EEG
activity may be seen over that region of the brain which is
least involved in the pathological process. There are clues
by which this situation may be identified from a
conventional chart recording'9 but they will be recognised
only with difficulty if at all in a brain map.
In no field of biomedical investigation is interpretation
so sensitive to the quality of data acquisition as in EEG.
Clinical neurophysiologists spend years coming to terms
with the effects of montages, references, control settings,
and biological factors such as level of awareness and
medication, which must be considered before attempting
to interpret the data in relation to the clinical problem.
With brain mapping, all can seemingly be bypassed (even
the primitive artifact rejection algorithms are switched off
by inexperienced users if the electrodes malfunction), and
behold, there is the cerebral abnormality printed in
glorious colour and statistically validated for all to see. The
pictures can be entirely spurious. As Duffy himself has
written20 "the brain map without the EEG is blind".
Brain maps are produced from 16 to 32 electrodes
arranged in a grid pattern on the scalp, giving a spatial
resolution of about 6 cm, which is at best one quarter of
the resolution required for accurate representation of
cerebral electrical fields.2' The maps are produced by a
process of interpolation (for which there are several
methods, themselves the subject of active controversy)
between this inadequate number of electrode sites.
Expressed more succinctly, 'most of the information in a
brain map is fabricated to produce nice contours. The
number of contours and their separation is under user
control: you can make things as "hot" or "cold" as you
like. As with the device of zero suppression in graphs
commonly used to support a particular political
interpretation of economic indices, differences can be
exaggerated at will. At any instant one part of the head
must be most positive and another most negative; with
automatic or manual scaling these maxima and minima
appear as red and blue regions, to persuade the naive that
a focal abnormality has been detected. By the same token,
any focal feature will be balanced by an extrema of
opposite polarity, usually over a distant region of the scalp,
so that, for instance, brain mapping may "reveal" that a
frontotemporal spike is accompanied by a focus in the
contralateral posterior temporal region.
Isopotential maps offer ample scope for misuse and
misinterpretation. The selection of what to map is at the
discretion of the user; the rules are flexible and subjective
and there is no standard. The map provides no means of
distinguishing between cerebral potentials and artifact, or
between the feature of interest and superimposed activities
with a different topography. It is essential that the primary
trace be recorded, and assessed with no less skill and
attention than in traditional EEG and EP work, as the user
Editorial
must be in no doubt as to what feature is being plotted.
Asynchronous events present another pitfall, particularly
for EPs. If a wave peaks earlier at electrode A than at B, a
single map may suggest that it is larger at A, or at B, or
equal at the two, depending on the time point selected for
mapping. Thus a cavalier approach of plotting a single map
at a point in time which roughly corresponds to the peak of
an EP component, confounds amplitude distribution with
topographic differences of latency, and invites
misinterpretation, the more so if the measurements are
given spurious objectivity by statistical comparison with a
reference database of normative values.
The possibilities for misinterpreting maps of derived
measures are yet greater than for isopotential displays. A
longstanding problem of quantitative EEG analysis is to
determine the best method of deriving the signals to be
analysed. Common reference derivation might appear the
obvious choice. There are difficulties in interpreting
reference montages in EEG tracings, but these are
overcome by an experienced observer who can see both the
amplitude and the phase of the waveforms. For instance, it
is not difficult to recognise that apparent frontal alpha
activity on average reference recording is due to the same
parieto-occipital generator as that at the back of the head,
but spectral mapping, with loss of phase information,
indicates only that the frontal leads are apparently
recording alpha activity. Asymmetrical activity involving
the oieoned ears reference is particularlv open to
misrepresentation so that alpha asymmetries may even be
shown as reversed.22 Paradoxically, when a focal activity
occurs at or near the reference, the deflections produced
are greatest on channels recording from the most distant
electrodes.'6 This effect is seen in isopotential plots but is
particularly liable to misinterpretation in spectral maps, as
information about polarity is lost: thus focal temporal delta
activity may even be mislocalised to the contralateral
central region.23
The development in recent years of low cost colour
computer graphics has increased both the appeal and the
availability of brain mapping, and has contributed to a
renewed interest in quantitative investigation of EEG and
EPs.20 In principle, this is a welcome development; the
problem of mapping lies not so much in the method itself,
as in the spurious validity which may be attached to the
maps, particularly by uninformed users (encouraged by
irresponsible advertising by some manufacturers) who see
them as a shortcut to providing a clinical EEG or EP
service, or even as an inexpensive neuroimaging technique.
The serious investigators developing these technologies to
advance the objective measurement of cerebral electrical
activity are outnumbered by less sophisticated users. Only
one third of mapping systems in the USA are operated by
EEG certified neurologists.24 Against the background of
technical and theoretical pitfalls-some obvious and
avoidable, others subtle and easily overlooked-the
dangers of clinical misuse of mapping are clear. For
research purposes too, interest in graphic presentation of
derived EEG features may be counterproductive if it
promotes the generation of pictures above the rigorous
application of statistical methods.
Digital "paperless" EEG machines are likely to replace
conventional recorders within a few years and mapping
will be a standard feature of such systems. It is important
that the need for expertise in the use of brain mapping be
recognised now, before misuse of this promising
technology brings harm to patients and disrepute to
quantitative clinical neurophysiology.
CD BINNIE,
BB MACGILLIVRAY
The Maudsley Hospital and
the Royal Free Hospital, London
Downloaded from http://jnnp.bmj.com/ on June 17, 2017 - Published by group.bmj.com
Editorial
529
1 Dietsch G. Fourier-analyser von Elektencephalogrammen des Menschen.
Arch ges Physiol 1932;230:106-12.
2 Grass AM, Gibbs FA. Fourier bandform of the electroencephalogram. J
Neurophysiol 1937;1:521-6.
3 Drohocki Z. L'electrospectrographie du cerveau. CR Soc Biol
1938;129:889-93.
4 Walter WG. An automatic low frequency analyzer. Electronic Eng
1943;11:237.
5 Walter WG, Shipton HW. A toposcopic display system applied to
neurophysiology. J Br Inst Radio Eng 195 1;3:260-73.
6 Walter WG, Shipton HW A new toposcopic display system.
Electroencephalogr Clin Neurophysiol 1951 ;3:281-92.
7 Brazier MAB. The electrical fields at the surface of the head during sleep.
Electroencephalogr Clin Neurophysiol 1949;1: 195-204.
8 Remond A. Orientation et tendances des methodes topographiques dans
l'etoile de l'activite electrique du cerveau. Rev Neurol (Paris)
1955;93:399-432.
9 Lesevre N, Remond A. Selected applications of a topographic approach to
event-related potentials. In: Duffy FH ed, Topographic mapping of brain
electrical activity. Boston: Butterworths, 1986:143-74.
10 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;5:309-21.
11 Duffy FH, Bartels PH, Burchfiel JL. Significance probability mapping: an
aid in the topographic analysis of brain electrical activity.
Electroencephalogr Clin Neurophysiol 1981 ;51 :455-62.
12 Lehmann D. Human scalp EEG fields: evoked, alpha, sleep and spike-wave
patterns. In: Petsche H, Brazier MAB, ed. Synchronization ofEEG activity
in the epilepsies, New York: Springer, 1972:307-26.
13 Gregory DL, Wong PK. Topographical analysis of the centrotemporal
discharges in benign Rolandic epilepsy of childhood. Epilepsia
1984;25:705-1 1.
14 Ebersole JS. Equivalent dipole modelling-a new EEG method for
localisation of epileptogenic foci. In: Pedley TA, Meldrum BS, eds.
Current problems in epilepsy. Edinburgh: Churchill Livingstone,
1992:51-71.
15 Deiber MP, Giard MH, Mauguiere F. Separate generators with distinct
orientations for N20 and P22 somatosensory evoked potentials to finger
stimulation. Electroencephalogr Clin Neurophysiol 1986;65:321-34.
16 Macgillivray BB, Sawyers FJP. A comparison of common reference, average
and source derivations in mapping. In: Sampson-Dollfus D, ed. Statistics
and topography in quantitative EEG. Amsterdam: Elsevier, 1988:72-87.
17 Rodin E, Cornellier D. Source deviation recordings of generalized
spike-wave complexes. Electroencephalogr Clin Neurophysiol 1989;
73:20-9.
18 Lehmann D, Skrandies W. Spatial analysis of evoked potentials in man-a
review. Prog Neurobiol 1984;23:227-50.
17 Pernier J, Perrin F, Bertrand 0. Scalp current density: concept and
properties. Electroencephalogr Clin Neurophysiol 1988;69:385-9.
18 Fisch BJ, PedleyTA. The role of quantitative EEG topographic mapping or
'neurometrics' in the diagnosis of psychiatric and neurological disorders:
the cons. Electroencephalogr Clin Neurophysiol 1989;73:5-9.
19 Binnie CD. Pathological EEG phenomena. In: Osselton JW, Binnie, CD,
Cooper R, Fowler CJ, Maugiere F, Prior PM, eds. A manual of clinical
neurophysiology. London: Butterworths (in press).
20 Duffy FH. Brain electrical activity mapping: ideas and answers. Topographic
mapping of brain electrical activity. Boston: Butterworths, 1986:401-98.
21 Spitzer AR, Cohen LG, Fabrikant J, Hallett M. A method for determining
optimal interelectrode spacing for cerebral topographic mapping.
Electroencephalogr Clin Neurophysiol 1989;72:355-61.
22 Fisch BJ, Hauser WA, Keller DL, Pedley TA. EEG topography: a
comparison of bipolar and linked ear reference recordings.
Electroencephalogr Clin Neurophysiol 1989;73: 13P.
23 Binnie CD. Spatial analysis. In: Osselton JW, Binnie CD, Cooper R, Fowler
CJ, Maugiere F, Prior PM, eds. A manual of clinical neurophysiology.
London: Butterworths (in press).
24 John ER. The role of quantitative EEG topographic mapping or
"neurometrics" in the diagnosis of psychiatric and neurological disorders:
the pros. Electroencephalogr Clin Neurophysiol 1989;73:2-4.
Neurological stamp
~~~s
P .5....'..
..;^m., .....
L75
Johannes Kepler 1571-1630
In his first work on optics (a sequel toVitello's in 1604) the
astronomer Kepler gave a clear description of the eyeball.
He was the first to demonstrate that cones of light given off
by luminous bodies were refracted and focused by the lens
on to the retina and that the retina was essential for sight.
He discovered that if the point of light fell in front or
behind the retina the vision would be blurred, and was
thus able to explain why spectacles were useful for those
with poor eyesight. Kepler was portrayed in 1971 by the
German Democratic Republic on a stamp honouring
prominent Germans. (Stanley Gibbons No E1372, Scott
no 1275).
t ti si
X .~~~~:5Q
~~~~~~~~~~~~~~~~~~~~~~~~~~~~'
S
L F HAAS
da
SL ^, ^ ^,3y ^ ^ ^ ^ ^ _
~~4* i;g£
N
AiL
kAi.A
-AME-M-Am
kA
m
hA
4
mA
MLALlb- Am -m'L--M6A I
Downloaded from http://jnnp.bmj.com/ on June 17, 2017 - Published by group.bmj.com
Brain mapping--a useful tool or a dangerous
toy?
C D Binnie and B B MacGillivray
J Neurol Neurosurg Psychiatry 1992 55: 527-529
doi: 10.1136/jnnp.55.7.527
Updated information and services can be found at:
http://jnnp.bmj.com/content/55/7/527.citation
These include:
Email alerting
service
Receive free email alerts when new articles cite this article. Sign up in the
box at the top right corner of the online article.
Notes
To request permissions go to:
http://group.bmj.com/group/rights-licensing/permissions
To order reprints go to:
http://journals.bmj.com/cgi/reprintform
To subscribe to BMJ go to:
http://group.bmj.com/subscribe/