Graefe's Arch Clin Exp Ophthalmol (1986)224:160-164
for CNnlcaland Experimental
Retinal visual acuity with pattern VEP
normal subjects and reproducibility*
F. Simon and B. Rassow
Medical Optics Laboratory, University Eye Clinic, D-2000 Hamburg, Federal Republic of Germany
Abstract. Retinal visual acuity was determined using an
objective method by means of VEP. Sixty subjects with
subjective visual acuity ranging from 20/25 to 20/12 were
measured with a computer-controlled laser interferometer,
using interference fringes modulated in counterphase as pat-
tern-reversal stimulus. Steady-state VEPs taken at different
spatial frequencies were evaluated by Fourier analysis and
linear regression. Objective retinal acuity was calculated as
the intersection between the linear regression line and the
linear scaled spatial frequency axis. In 68% of all cases,
the difference between objective and subjective visual acuity
was less than ± 5.1 cycles per degree (c/d) ( ± 15.4%referred
to subjective acuity); in 86%, the difference was less than_+
7.7 c/d (±23.1%).
Introduction
Assessment of visual function with an objective method
has been posed an interesting question for further investiga-
tion. The application of visual evoked potentials (VEP) for
the determination of visual acuity revealed some promising
results (Regan 1978; Sherman 1979; Sokol 1978; Tyler
et al. 1979). Although simple VEP techniques are widely
used in hospital diagnosis at present, research centers and
laboratories are still the main area for more sophisticated
methods.
As Sherman (1979) reported in a general review of VEP
applications, there are different ways of determining acuity
using VEP techniques - transient as well as steady-state
potentials. Threshold criteria for the highest spatial fre-
quency giving a reproducible VEP are utilized. The thresh-
old is related to visual acuity. From the amplitudes taken
at different spatial frequencies, visual acuity can be esti-
mated in two ways: first, by identifying where the peak
response lies, e.g., if the largest response occurs between
2-3 cycles per degree (c/d), the visual acuity is assumed
to be 20/20 or 30 c/d (Sokol 1976); second, the descending
limb of the curve can be extrapolated to see where it crosses
the zero amplitude potential point. In most cases the VEP
is recorded by averaging methods. Fourier analysis, as pro-
*This work was supported by the Bundesministerium ffir For-
schung und Technologie, grant no. MMT 17
Offprint requests to." F. Simon, Abteilung fiir Medizinische Opt±k,
Universitgts-Augenklinik, Martinistr. 52, D 2000 Hamburg 20,
Federal Republic of Germany
Graefe's Archive
Ophthalmology
© Springer-Verlag1986
posed by Regan (1977) for steady-state potentials, offers
some advantages compared to averaging. The amplitude
at the stimulation frequency calculated from the power
spectrum represents one precise value from one measure-
ment. Estimation of significance is easier to perform, as
the whole averaged graph of a transient recording does
not have to be taken into account. Odom et al. (1983) com-
pared subjective and VEP acuity of adults by pattern-rever-
sal steady-state methods and Fourier analysis. Linear re-
gression gave good agreement between their subjective
(35 c/d) and VEP visual acuity average (31.3 c/d). Vrijland
and van Lith (1983) made interesting experiments for acuity
prognosis with a rather unconventional method, generally
not accepted, as a measure of retinal resolution. They used
luminance VEP for preoperative cataract estimation and
found a relationship between luminance VEP amplitudes
and acuity. They obtained 75% positive results with the
criterion that amplitudes greater than 21 ~tV could be con-
sidered as a sign of probable good acuity following extrac-
tion. Arden and Sheorey (1977) described a pattern-reversal
stimulus for the assessment of visual function with VEP,
which was built as a laser interferometer. This principle
is also a central point of our work.
Based on the Rodenstock retinometer developed by
Rassow and Wolf (1973), realizing an idea of LeGrand
(1935), later pursued by Campbell and Green (1965), a VEP
pattern-reversal stimulator was constructed in our laborato-
ry. It enables objective determination of retinal visual acuity
by means of laser interference fringes with different spatial
frequencies (Rassow et al. 1984; Kusel et al. 1985). Steady-
state VEP are the basis for calculating the VEP visual acuity
by linear regression. In this article results on normal sub-
jects and some information about reproducibility are de-
scribed.
Materials and methods
The optical and electronic equipment of our VEP system
is presented in a block diagram (Fig. 1). The central part
is the computer-controlled optical stimulus, a modified Mi-
chelson interferometer, which projects interference fringes
in Maxwell±an view onto the retina. The setup is completed
by a Dorn HP 8911/12 biosignal amplifier with Beckman
AgC1 electrodes and a biosignal computer Plurimat PS 300
(Intertechnique) with a digital plotter for graphic output
of the results. After a description of the stimulus, a short
explanation of the other equipment follows.

stimulus generator AgCl-electrodes
contrast spatial
,, frequency
tt
Plutimat PS 300 biosignal amplifier
computer with hard disk HP 8911
highpass
I digital plotter
Fig. l. Blo~ diagram of the VEP recording system
The interferometer (Fig. 2), generating interference
fringes as pattern -reversal stimulus, uses a 0.5 mW polar-
ized HeNe laser as light source. The laser beam is split
up into two parts in the interferometer. After a precise
phase shift, the parts of coherent light are recombined and
mixed with a precisely controlled amount of incoherent red
light for contrast regulation, so that the total intensity al-
ways remains constant and is independent of the chosen
contrast. The radiant power at the eye is reduced to 0.1 gW,
a value absolutely harmless to the retina.
The mode of operation of the optical system can be
understood with help of Fig. 2. Lens L4 and L5 widen the
laser beam. The beam splitter ST1 is tilted against the opti-
cal axis (8 deg) to avoid disturbing interference fringes com-
ing from lens reflections. One part of the beam is projected
to a mirror mounted on a piezoelectric drive PZ. The mirror
has to oscillate 156.8 nm (1/4 wavelength of the HeNe laser)
in the direction of the optical axis in order to produce the
pattern reversal. Therefore, the computer-generated stimu-
lus frequency regulates the displacement of the mirror by
means of the physical properties of the piezodrive. In our
experiment, bright and dark parts of the interference fringes
change their position 6.5 times/s. The second beam can
be tilted up to 3.5 deg in two dimensions to control orienta-
tion and spatial frequency of the interference fringes. Gen-
erally, vertical fringes from 2 to 30 c/deg are used. The
following lens system (L3, Li) collimates the beam. The
diaphragm B controls the visual field (2-16 deg). Lens L1
I L6 L7 P~
PZ ! ~ L3 L2
---r ...............
LASER
Fig. 2. Optical pathway of the pattern-reversal stimulus, using laser interference fringes modulated in counterphase (Pl, Pi, P3 polarizers,
L1 to L8 lenses, STI, ST2 beam splitter, PZ piezodrive, KS mirror-tilt system, B diaphragm, E plane of the nodal point of the eye;
see text)
161
focuses the beam to the nodal point of the eye. By this
means the interference fringes are projected in Maxwellian
view onto the retina. Defects in the optical media of the
eye do not disturb the quality of the fringes; opacities only
cause laser-speckle. Crosshairs in the diaphragm help to
correct fixation.
The supplementary optical pathway serves for contrast
lowpass
control. A polarized, parallel beam of incoherent light com-
ing from a red LED (HP 4658, dominant wavelength
626 nm) is added to the coherent interferometer beam. A
system of two polarizers perpendicular to each other (Pi,
P3) performs contrast control. A certain amount of incoher-
ent light, depending on angular position of the polarizer
system, is mixed to the coherent laser beam. The total inten-
sity is always maintained at the constant value of 0.1gW
(250 trolands or td). The visual field used in our experiments
extends 12 deg. Contrast is adjustable from 0.99 to thresh-
old. After some preliminary measurements, we fixed the
contrast to 0.33 for all further VEP recordings. This value
lies below saturation level (Millodot 1977).
The electrodes are placed at CZ (reference), OZ (signal)
and right mastoid (ground), with impedance values of ap-
proximately 1 kOhm or less. The Dorn amplifier system,
especially designed for amplification of small biosignals,
has built-in high- and lowpass filters (slope 4 dB/octave),
which are set at I and 100 Hz during the experiments. An
optoisolation system guarantees galvanic separation for
safety. Amplification used is about 60,000.
The computer Plurimat PS 300 is qualified for recording
and analysis of biological signals because of hardware built-
in functions like FFT and its inverse. A Fourier transform
of 8192 data points stored during one VEP steady-state
recording takes less than 2.5 s. A 5 MByte disk provides
enough memory for 11 recording sessions. Software is writ-
ten in BASIC.
The software for our investigations consists in a measur-
ing program for steady-state VEP, using Fourier analysis
methods and an evaluation program for linear regression.
The result calculated from every measurement is the ampli-
tude at the stimulus frequency, with its associated phase de-
rived from the power spectrum. Another programming fea-
ture is that 8192 data points are sampled with a frequency
L8 ', I
I2 LI E
[...... /
162

[c/d]
50.0
/ / /
ompl i t u d e [}Jr] / / /

10.0 /: / /
40. 0
./ >, ¢,
3 5 . 182 /
VEP
8.0
/ ,l /
8
30. 0 i/ /! / / ,

6.0
ob j e c t i v ( / /"
/L
20. 0
/ /
/
4.0 / /
/ / /
10.0 / / /
2.8
/ /
/ /
0.0 i i i l 0.0 " Z I : i i i
O.O 6.0 12.0 18.0 24.0 30.0 o.o 10. o 2o.o ao.o 40.0 so.o [c/d3
spotiol Frequency [c/d]
visual ocuity subjectiv~
Fig. 3. Determination of VEP acuity by linear regression with linear Fig. 4. Objective retinal visual acuity of 53 healthy normal subjects
scaled spatial frequency in the domain 12-30 c/d. The result of as function of subjective visual acuity. The hatched lines show the
35.2 c/d corresponds well to the subjective value of 37.5 c/d confidence area of _+6 c/d

of 104 Hz at each run. Measuring time for one steady-state
VEP is 512 sweeps in 77.8 s. The high number of sweeps
is made to obtain a higher signal-to-noise ratio even at
high spatial frequencies. Noise is defined as the mean of
the amplitudes in the interval of 1 Hz around the stimula-
tion frequency. The stimulus response amplitude does not
enter into this mean. The stimulation rate of 6.5 reversals/s,
known to be well suited for pattern VEP, is established
as a multiple of the sampling period to avoid leakage.
One measuring session takes about 45 min; 18 steady-
state VEPs at 6 different spatial frequencies are recorded
in the sequence 2, 30, 12, 24, 6, 18 c/d. Therefore, adapta-
tion effects are avoided (Meccaci and Spinelli 1976). Fur-
ther analysis of the calculated VEP amplitudes given by
the Fourier transform is done with linear regression meth-
ods. As we are interested in retinal visual acuity, we look
at the declining part of the spatial tuning function towards
high spatial frequencies. We use linear extrapolation tech-
niques described by Tyler et al. (1979) and Campbell and
Maffei (1970). The bandpass structure of the VEP shows
large interindividual variations. Tyler et al. (1978) found
various minima and maxima in the low spatial frequency
region. The only typical feature is the nearly linear decrease
in the high frequency region, which is the base for linear
regression. The regression line is fitted from the peak of
the amplitude-spatial frequency function towards high fre-
quencies. The intersection point with the spatial frequency
axis is defined as the objective or VEP retinal visual acuity
(see Fig. 3). Advantages of this definition are discussed
later. To compare subjective and objective results, the dif-
ference D between objective or VEP and subjective acuity
is calculated (Dun with linear scaled spatial frequency, D]og
with logarithmic scaled spatial frequency). The relative dif-
ference R is obtained dividing D by the corresponding sub-
jective visual acuity.
Linear regressions are calculated for amplitudes, normal-
ized amplitudes or signal-to-noise ratios as a function of
spatial frequency. The spatial frequency interval, contribut-
ing to the regression line, is chosen, so that the amplitudes
within fit the regression line well. Amplitudes with signal-
to-noise ratios below 1, are rejected. If the ratio lies between
1 and 2, phase information is considered. When phase dif-
ferences greater than 30 deg belonging to amplitudes at one
spatial frequency are found, these critical amplitudes are
eliminated from the regression. Amplitudes are taken into
consideration when the phase differences of critical values
are smaller or equal to this value.
Results
VEP retinal visual acuity was measured in 60 cases. Subjec-
tive acuity of the subjects, aged from 17 to 35 years, ranged
from 24 (20/25) to 48 c/d (20/12). It was measured with
Snellen optotypes under standardized conditions. Figure 4
presents objective or VEP retinal acuity as a function of
subjective acuity. Results were calculated by linear regres-
sion, using the declining part of the VEP amplitude curve
as a function of the linear scaled spatial frequency. The
mean difference between objective or VEP retinal acuity
and subjective acuity DI~, is --2.01 c/d. More important
than the mean value is the fact that 68% of all cases lie
within a standard deviation of 4-5.1 c/d. The corresponding
relative confidence range leads to 4-15.4% (/~,, = - 4 . 6 % ) .
Fifty-three subjects contributed to this result. For seven
subjects, the percentage of nonsignificant amplitudes was
greater than 0.3; they were therefore excluded. There exists
a small tendency to underestimate VEP acuity with linear
spatial frequency. The logarithmic scaled spatial frequency
leads to considerable overestimation of objective retinal acu-
ity. Here the 68% confidence interval of the mean
difference of VEP and subjective acuity lies within ± 27.7 c/d
around the mean of /3log= 13 c/d, the percentage range
(R]og=35.1%) is _+64%. The results demonstrate that a
linear scaled spatial frequency axis gives better results than
a logarithmic one.
Results concerning reproducibility were taken from a
subgroup of seven selected subjects. A significant amplitude
at high spatial frequencies ( > 18 c/d) in the first VEP ses-
sion was criterion for acceptance. With one exception,
where only two VEP sessions were possible, three measure-
 163

Table 1. Reproducibility results in seven Subjects Besides this, low contrast is guaranteed to be below satura-
tion level (Regan 1983).
Subject Subjective VEP retinal acuity Mean VEP Linear regression as an analysis method has advantages
acuity (3 measurements) acuity compared to threshold methods. First, the result is deter-
mined not only by the amplitude at one spatial frequency;
E.W. 36.0 29.9 30.0 41.4 33.8 instead the VEP acuity depends on amplitudes at several
C.G. 37.5 32.1 33.3 31.6 32.3
A.O. 30.0 31.1 33.1 33.1 32.4 spatial frequencies. The influence of individual, isolated, in-
G.M. 37.5 43.8 35.4 35.5 38.2 accurate amplitude values at high spatial frequencies is
F.O. 37.5 30.2 29.4 30.3 30.0 therefore limited and not the only basis for the acuity esti-
M.A. 30.0 29.9 31.6 33.4 31.6 mate. Second, no linear transformation is necessary to con-
J.F. 30.0 33.2 35.6 34.4 vert the VEP threshold to subjective acuity (Teping 1981).
(c/d) (c/d) (c/d) Towle and Harter (1977) report that some authors believe
the VEP pattern threshold underestimates acuity - without
linear transformation of the threshold.
The first measurements of subjective acuity values less
ments of each subject at intervals of about 8 days were than 12/20 indicate that, in this region, the precision of
taken. The results are presented in Table 1. The mean values our method decreases, and scattering increases in cases
for each subject are given as well as all single results. In where only low spatial frequencies contribute to VEP acu-
all cases VEP retinal acuity could be found. The mean dif- ity.
ference between VEP retinal acuity and subjective acuity Applications of our apparatus are for basic research
/Slln is - 1 . 0 7 c/d (/~11.= - 2 . 2 % ) , with a standard deviation and perhaps for clinical cases where retinal acuity is difficult
of 4.9 c/d (13.9%). The small underestimation of VEP acu- to obtain and an objective method is required.
ity shows scattering that is no greater than the grading
of optotypes.
With a logarithmic scaled spatial frequency axis, the References
results show significant overestimation of VEP acuity: the Arden GB, Sheorey UB (1977) The assessment of visual functions
mean difference of subjective from VEP acuity Dlog is 10 c/d in patients with opacities : a new evoked potential method using
(/~1og=29.6%), with 8.6 c/d (24.6%) standard deviation. a laser interferometer. VEPs in man In: Desmedt JE (ed) New
The precision of our piezoelectric drive (resonance fre- developments Clarendon Press, Oxford, pp 381-394
quency > 20 kHz), which is responsible for pattern reversal Campbell FW, Green DG (1965) Optical and retinal factors affect-
(reversing time < 0.5 ms), contributes to high phase stabili- ing visual resolution. J Physiol 197:576
ty. Strong stimuli (2 and 6 c/d) give phase differences of Campbell FW, Maffei L (1970) Electrophysiological evidence for
less than 10 deg at repeated measurements; higher spatial the existence of orientation and size detectors in the human
visual system. J Physiol 207:635-52
frequencies reach 20 deg, sometimes even more. By this Grall Y, Regaudiere F, Delthil S, Legargasson JF, Sourdille J
means the decision as to whether or not amplitudes with (1976) Potentiels evoques et acuite visuelle. Vision Res
low signal-to-noise ratio are significant can be made. 16:1007-1012
Kusel R, Wesemann W, Rassow B (1985) A new laser interfer-
ometer for the stimulation of pattern-reversal visual evoked
Discussion potentials. Clin Phys Physiol Meas 6:239-246
Le Grand Y (1935) Sur la mesure de l'acuite oculaire au moyen
Application of VEP for determination of visual acuity is de frange d'interference. C R Acad Sci (Paris) 200:490
Meccaci L, Spinelli D (1976) The effects of spatial frequency adap-
not a routine clinical method, but is used in some clinical
tation on human evoked potentials. Vision Res 16:477-79
research laboratories. The results of our study show that Millodot M (1977) The use of VEPs in optometry. VEPs in man
VEP techniques can be of value to acquire objective infor- In: Desmedt JE (ed) New developments. Clarendon Press, Ox-
mation on retinal acuity in the domain of 20/25 to 20/12. ford, pp401-409
In 86% of all evaluable cases, the precision of our method Odom JV, Maida TM, Dawson WW, Romano PE (1983) Retinal
amounts to + 2 3 . 1 % (_+7.7 c/d). Such deviations can be and cortical pattern responses: a comparison of infants and
tolerated in most cases. Deviation of reproducibility data adults. Am J Optom Physiol Opt 60:369-375
for the same confidence interval yields _+20.9% (_+ 7.4 c/d). Rassow B, Wolf D (1973) Erfahrungen mit dem Laserinterferenz-
Comparison of subjective acuity values, influenced by streifentest bei der Messung des retinalen Aufl6sungsverm6-
gens. Graefe's Arch Clin Exp Ophthalmal 187:61
such factors as brightness of projected test targets, back-
Rassow B, Kusel R, Simon F, Wesemann W (1984) Visuell evo-
ground, and even psychological factors, often reach the zierte Potentiale. In: Schfitz J (ed) Medizinische Physik. Hfithig,
same order of magnitude. Grall et al. (1976) obtained good Heidelberg, pp 181-190
results in a large number of cases, although one has to Regan D (1977) Fourier analysis of evoked potentials: some meth-
keep in mind that his method was not like ours (no Maxwel- ods based on Fourier analysis. VEPs in man In: Desmedt JE
lian view, no constant visual field, no constant luminous (ed) New developments. Clarendea Press Oxford, pp 110-117
target density). He found 80% agreement of subjective and Regan D (1978) Assessment of visual acuity by EP-recording: am-
objective visual acuity, results that were later confirmed biguity caused by temporal dependance of spatial frequency
by Towle and Harter (1977). selectivity. Vision Res 18:439-434
Although the wavelength used (632.8 nm) is not located Regan D (1983) Spatial frequency mechanisms in human vision
investigated by evoked potential recording. Vision Res
at the eys's maximum sensitivity (Siegfried 1970), we do 23 : 1401 07
not need high pattern intensities to evoke responses. Sub- Sherman J (1979) Visual evoked potential: basic concepts and clini-
jects regard the applied contrast of 0.3 as comfortable. They cal application. J Am Optom Assoc 50:19-30
do not become as tired as with pattern of high contrast. Siegfried JB (1970) The relationship between stimulus wavelength
164
and the waveform of averaged visual evoked cortical potentials.
Am J Optom 47:282-287
Sokol S (1976) Visual evoked potentials: theory, techniques and
clinical applications. Surv Ophthalmol 21:18-44
Sokol S (1978) Measurement of infant visual acuity from pattern
reversal evoked potentials. Vision Res 18:33-39
Teping C (1981) Visusbestimmung mit Hilfe des visuell evozierten
cortikalen Potentials (VECP). Klin Monatsbl Augenheilkd
179:169-172
Towle VL, Harter MR (1977) ~Objective determination of human
visual acuity: pattern evoked potentials. Invest Ophthalmol Vis
Sci 16:1073-1076
Tyler CW, Apkarian P, Nakayama K (1978) Multiple spatial-fre-
quency tuning of electrical responses from human visual cortex.
Exp Brain Res 33:535-50
Tyler CW, Apkarian P, Levi DM, Nakayama K (1979) Rapid
assessment of visual function: an electronic sweep technique
for the pattern visual evoked potential. Invest Ophthalmol Vis
Sci 18:703-13
Vrijland HRT, Lith GHM van (1983) The value of preoperative
electro-ophthalmological examination before cataract extrac-
tion. Doc Ophthalmol 55:153-156
Received April 10, 1985 / Accepted September 20, 1985