Acta Ophthalmologica 2019

Semi-automated kinetic perimetry: Comparison of

the Octopus 900 and Humphrey visual field
analyzer 3 versus Goldmann perimetry
Claudia Bevers,1 Gauthier Blanckaert,2 Karel Van Keer,1,3 Jean-Francßois Fils,4 Evelien Vandewalle1,3
and Ingeborg Stalmans1,3
1
Department of Ophthalmology, University Hospitals Leuven, Leuven, Belgium
2
Faculty of Medicine, KU Leuven, Leuven, Belgium
3
Department of Ophthalmology Neurosciences, Laboratory of Ophthalmology, KU Leuven, Leuven, Belgium
4
Ars Statistica, Nivelles, Belgium

ABSTRACT. essential diagnostic tool for many dif-

Purpose: To evaluate the clinical usefulness and reproducibility of (semi-) ferent indications. In general, kinetic
automated kinetic perimetry of the Octopus 900 and Humphrey field analyzer 3 perimetry is used when the visual field
(HFA3) compared to Goldmann perimetry as reference technique. defects are dense and large, when the
Methods: A prospective interventional study of two study groups, divided into shape of visual field defect is important
three subgroups. The first study group consisted of 28 patients, performing one in disease diagnosis and monitoring or
visual field examination on each of the three devices. A second group of 30 when patients benefit from interaction
patients performed four examinations, one on Goldmann and three on Octopus with the examiner (Dersu et al. 2006).
900 with the following testing strategies: (1) with reaction time (RT) vector, no Kinetic perimetry can also be used to
headphone; (2) without RT vector, no headphone; and (3) without RT vector, assess patients’ fitness to drive. Despite
with headphone. Comparisons for V4e and I4e stimuli were made of the mean the enormous possibilities of the Gold-
mann perimeter, many disadvantages
isopter radius (MIR) and of the distances of the isopter to the central visual axis
are also related to the device. More in
in four directions. Statistical analysis was made with the R software version
particular, the application of the device
3.2.2. implies the need for a skilled perimetrist
Results: For V4e stimuli, the mean isopter radius showed no statistic significant experienced in performing a Goldmann
difference comparing Goldmann to HFA3 [p-value = 0.144; confidence interval kinetic visual field examination, and
(CI) 0.152 to 0.019] and comparing Goldmann to Octopus 900 without RT controlling direction and speed of the
vector, either with (p-value = 0.347; CI 0.023 to 0.081) or without headphone stimulus movement. Hence, tests are
(p-value = 0.130; CI 0.011 to 0.095). Octopus 900 with RT vector produced a difficult to standardize given the inher-
significantly larger MIR for V4e stimuli in comparison to Goldmann ent examiner bias. Along with the lack of
(p-value < 0.001). I4e stimuli produced statistically significantly larger visual quantifiable parameters, the repro-
field areas when comparing HFA3 and Octopus 900 to Goldmann perimetry. ducibility and follow-up in Goldmann
Conclusion: Humphrey field analyzer 3 and Octopus 900 without RT vector are perimetry are very difficult (Ramirez
promising successors of Goldmann perimetry. et al. 2008). And considering that the
examiner must mark the locations where
Key words: goldmann perimeter – HFA3 – kinetic perimetry – Octopus 900 – semi-automated
the patient sees the stimuli entering his
kinetic perimetry visual field, the Goldmann perimeter’s
technology does not take into account
the patient’s and examiner’s reaction
Acta Ophthalmol. 2019: 97: e499–e505
time (RT). Thus, the RT of patient and
ª 2018 Acta Ophthalmologica Scandinavica Foundation. Published by John Wiley & Sons Ltd
examiner induces a systematic shift of
doi: 10.1111/aos.13940 the isopter or scotoma border towards
the direction of the stimulus movement
Introduction perimetry. Although automated static and causes a decrease of the visual field
Since its introduction in 1945, the use of perimetry has replaced a lot of indica- area or an increase of the scotomata
a manual Goldmann perimeter has been tions for performing kinetic perimetry, (Nowomiejska et al. 2010). Finally, and
the method of choice to perform kinetic its utilization is still relevant and an being the most vital problem of

e499
Acta Ophthalmologica 2019
Goldmann perimetry, the Goldmann
technology is no longer produced since
2007 and will therefore be disappearing
in the coming years. Hence, there is a
need for alternative technologies which
can replace the Goldmann perimeter as
reference technique in kinetic perimetry.
This led to the introduction of (semi-)
automated kinetic perimetry (SKP),
that is computer-assisted kinetic perime-
try. Theoretically, SKP helps to stan-
dardize the examination procedure,
assists in inter-test comparison and
requires less technical expertise. Haag-
Streit (Koenig, Switzerland) developed
a computer-assisted kinetic examina-
tion mode using test stimuli and a
bowl-shaped design similar to the Gold-
mann perimeter: Octopus 101 and its
successor Octopus 900. These two
devices differ mainly in the applicable
software programs. Octopus 900 offers
the ability to assess a reaction time
vector, which makes it possible to cor-
rect the given responses for the RT that
the patient exhibits. In parallel, Carl
Zeiss Meditec (Jena, Germany) also
developed computer-assisted kinetic
perimetry, the Humphrey visual field
analyzer was introduced as another
potential successor of Goldmann
perimetry as it offers a kinetic testing
mode too.
Unfortunately, little information is
yet available on the reproducibility of
SKP (using Octopus 900 and HFA3) in
comparison to Goldmann perimetry,
with only few studies yet published. In
this study, we compared these two new
technical devices (Octopus 900 and
HFA3) to the reference technique.
Preliminary results suggested that for
Octopus 900, two important parame-
ters have an influence on the results,
namely the reaction time vector and an
auditory factor, given that the Octopus
900 produces a notable sound when
projecting the stimuli. Hence, we intro-
duced two separate study groups, one
comparing Goldmann to both the
Octopus 900 and the HFA3 and
another one comparing Goldmann to
the Octopus 900 with or without RT
vector and with or without headphone
to mask the auditory stimulus.
Materials and Methods
We conducted a prospective interven-
tional study in which we compared
three different kinetic perimeters:
Goldmann manual perimeter, HFA3
e500
and Octopus 900 perimeter. This sur-
vey was conducted with local ethical
approval and in accordance with the
Tenets of the Declaration of Helsinki.
Patients
Twenty-eight patients were enrolled in
the first study group, the second group
consisted of thirty patients. Each study
group was divided into three subgroups:
(1) patients diagnosed with glaucoma,
who were followed-up in the glaucoma
clinic with the Goldmann perimeter due
to extensive visual field defects; (2)
patients suffering from neurological dis-
eases, who have an abnormal peripheral
visual field; and (3) healthy control
patients who do not have any diseases.
Healthy individuals were recruited
during the general ophthalmological
consultation, where they had an app-
ointment for fitness to drive consulta-
tion or general screening. Detailed
medical history was taken to exclude
relevant systemic diseases which could
influence the visual field examination.
An extensive ophthalmological exami-
nation was performed to exclude dis-
eases. All patients were recruited at the
University Hospitals UZ Leuven.
Patients in study group 1 were asked
to undergo three examinations: one on
a Goldmann perimeter and the second
and third examination using the HFA3
and the Octopus 900. Study group 2
underwent four examinations, one on a
Goldmann perimeter and three test
strategies on the Octopus 900: (1) with
RT vector, no headphone; (2) without
RT vector, no headphone; and (3)
without RT vector, with headphone.
Only one eye per subject was tested.
If the two eyes were evaluated with
Goldmann perimetry, the eye with the
most affected visual field performed the
additional investigations. An informed
consent was given to the patients prior
to any study examination and only
those who agreed were included in the
study. Exclusion criteria were as fol-
lows: (1) inability to perform a periph-
eral visual field examination; (2)
Table 1. General characteristics of the enrolled patients.
Number of Number of
patients examinations
Study group 1 28 3 52
Study group 2 30 4 60
inability to provide informed consent;
and (3) aged under 18 years. Table 1
illustrates the general characteristics of
the investigated patients.
Examination procedure
The study protocol consisted of per-
forming three (group 1) or four (group
2) assessments with Goldmann, Octo-
pus 900 and/or HFA3 perimetry on the
same day. An online available random-
ization program was used to program
the testing order (www.random
ization.com). Between the examina-
tions, every patient had a minimum
5-min break. The same testing strategy
was applied for each device, whereby a
direct comparison could be done. The
peripheral visual field boundaries were
assessed using a I4e and V4e target. If
the I4e target could not be seen by the
patient, we performed a test using the
III4e target. This means that two
isopters were obtained for each device.
A minimum of twelve vectors was used
to evaluate the subject’s peripheral
visual field. The mean stimulus speed
on Octopus 900 and HFA3 was 5°/
second. On Goldmann perimetry, the
perimetrist kept a constant stimulus
presentation speed with an average of
5°/second. Every patient always had
the same perimetrist who guided him/
her during the examinations. Only two
different perimetrists were involved in
this study, one perimetrist for each
study group.
The Goldmann perimeter was cali-
brated on a weekly basis, and no
aberrations were detected throughout
the study period. The automated
perimeters perform an auto-calibration
each time they are restarted (which was
done on a daily basis).
Data analysis
The obtained isopters for the different
perimeters were compared quantita-
tively by assessing the visual field
(VF) area inside the different isopters.
We used the software of ‘Image J’ to
Mean Left eye Right eye
age tested tested
14 14
15 15

assure size equivalency between the
different devices and to measure the
VF area inside the obtained isopters.
These data (in square degrees) were
translated into the mean radius of the
isopter (MIR, in degrees) which gives a
more comprehensive way to compare
the different outcomes (M€ onter et al.
2017). Besides quantitative comparison
of the isopter areas, the position of the
isopters was also tested. We measured
the distance of the isopter (in degrees)
in relation to the central visual axis.
The distances were measured in four
directions: superior, inferior, nasal and
temporal, each one 90° perpendicular
to each other.
Sample size calculation
Sample size calculations were based on
previous reports on differences between
Octopus and Goldmann kinetic
perimetry (Nowomiejska et al. 2012).
For a cross-over design with an alpha
equal of 0.025 in the first study group,
0.017 in the second study group, a
power of 80%, and an expected effect
size between the Goldmann and Octo-
pus 900 V4e area of 3000 square
degrees, a sample size of 28 patients is
necessary to achieve significance. To
accommodate for an expected exclu-
sion rate of 10%, we included 31
patients. The TrialSize R package was
used to calculate the sample size (Chow
et al. 2008).
Statistical analysis
Statistical analysis was made with the
R software version 3.2.2. A mixed
model for continuous data was utilized
to test differences in paired data (Ver-
beke & Molenberghs 2000). The Sha-
piro-Wilks test was used to test
whether the mixed model had normally
distributed data. Some data were not
normally distributed, in this case we
performed Dunnett’s multiple compar-
isons tests for paired nonparametric
data using the nparcomp R package,
taking Goldmann perimetry as refer-
ence (Konietschke et al. 2015). The
statistical proposition developed by
Konietschke and implemented in the
nparcomp R Package calculates 95%
confidence intervals (CI) for rank data
which comprise the zero value when no
difference is observed between the
groups and, when a difference is found,
the zero is not comprised in the
confidence interval (Konietschke et al.
2010). This mathematical solution then
mimics the behaviour of confidence
intervals for normally distributed data.
When the Shapiro-Wilks test indicated
nonsignificant deviation from the nor-
mality, we performed multiple compar-
ison tests for paired normally
distributed data using the multcomp
R package and taking the Goldmann
group as reference (Bretz et al. 2011).
Sample size calculation was made with
TrialSize R package.
Results
Comparison of MIR in study group 1
The data of V4e stimuli were not
normally distributed. Median MIR of
Goldmann perimetry was 50.1° and of
HFA3 was 50.3°. Statistical analysis
for V4e stimuli showed no statistically
significant difference between Gold-
mann and HFA3 (p-value = 0.144; CI
0.152 to 0.019). The Octopus 900
(with reaction time vector) on the
other hand had a median MIR of
59.9°, which showed a statistically sig-
nificantly wider (19.5%) mean isopter
radius than Goldmann perimetry
(p-value < 0.001; CI 0.112–0.290)
(Table 2).
I4e stimuli data were normally dis-
tributed. The mean difference in MIR
for I4e isopters was significantly smal-
ler for the Goldmann perimeter in
comparison to the Octopus 900 (16.3°;
Table 2. Median mean isopter radius (MIR) for V4e stimuli (in degrees) in study group 1.
Mean isopter radius Median (Q25–Q75)
Goldmann (I) 50.1 (41.2–60.1)
HFA (J) 50.3 (43.1–54.9)
Octopus 900 (J) 59.9 (50.9–59.9)
Q25 = lower quartile; Q75 = upper quartile.
Table 3. Mean isopter radius for I4e stimuli (in degrees) in study group 1.
Mean isopter radius Mean (SD) Mean difference (J–I)
Goldmann (I) 28.1 (19.6)
HFA (J) 37.1 (15.9) 9.0
Octopus 900 (J) 44.3 (17.0) 16.3
SD = Standard deviation.
Acta Ophthalmologica 2019
p-value < 0.001) and the HFA3 (9.02°;
p-value<0.001). (Table 3).
Comparison of the position of the isopters
in study group 1
For V4e and I4e stimuli, all data were
not normally distributed. A compar-
ison was made with respect to the
distances of the isopters in relation to
the central visual axis. HFA3 produced
larger distances for V4e stimuli in
inferior, temporal and nasal quadrants
(p-value < 0.001), but did not result in
statistically significant differences for
superior limits (p-value = 0.046; CI
0.268 to 0.002). The Octopus 900
demonstrated higher limits (12.7% on
average) in all four directions (p-
value < 0.001).
I4e stimuli also produced longer
distances in the four directions with
the Octopus 900 and in temporal and
nasal direction on HFA3 compared to
the Goldmann. Distances for superior
(p-value 0.676; CI 0.083 to 0.614) and
inferior (p-value 0.224; CI 0.027 to
0.128) quadrants on HFA3 were not
significantly different from Goldmann
perimetry.
Figure 1 shows the visual field of a
patient in study group 1, performed on
the three different devices.
Comparison of MIR in study group 2
Data for V4e and I4e stimuli were not
normally distributed. Median mean
95% Confidence
interval (J–I)
Lower Upper p-Value
0.15 0.017 0.144
0.112 0.290 <0.001
95% Confidence
interval (J–I)
Lower Upper p-Value
5.7 12.3 <0.001
13.0 19.6 <0.001
e501
Acta Ophthalmologica 2019

radius of the isopter in Goldmann

perimetry for V4e stimuli was 50.2°.
When comparing the MIR for V4e
isopters, analysis showed no statisti-
cally significant differences between
Goldmann perimetry and the group
of the Octopus 900 without RT vector
with headphone (p-value = 0.347; CI
0.023 to 0.081) and without RT
vector without headphone (p-
value = 0.130; CI 0.011 to 0.095).
When comparing V4e stimuli on Gold-
mann perimetry with the Octopus 900
with reaction time vector, Octopus had
a median MIR of 57.0°, being statisti-
cally significantly wider (13.5%) than
the Goldmann (p-value < 0.001; CI
0.09–0.248) (Table 4).
As seen in Table 5, I4e stimuli
demonstrated a statistically significant
bigger MIR with the Octopus 900 with
or without reaction time vector, and
with or without headphone compared
to Goldmann perimetry.

Comparison of the position of the isopters

in study group 2
Comparing the distances in the superior
direction for V4e stimuli (normally
distributed data), Goldmann perimetry
had smaller distances compared to the
Octopus 900 with RT vector (mean
difference 7.4°; CI 5.0–9.9; p-value <
0.001), the Octopus 900 without RT
vector and without headphone (mean
difference 4.2°; CI 1.7–6.7; p-value <
0.001) and the Octopus 900 without RT
vector and with headphone (mean dif-
ference 2.7; CI 0.2–5.1; p-value < 0.001).
Data of V4e stimuli in the other
directions (inferior, nasal and tempo-
ral) were not normally distributed.
They produced no difference in dis-
tances when comparing Goldmann
perimetry to the Octopus 900 without
RT vector, with or without headphone.
But the Octopus 900 with RT vector
resulted in wider lengths than Gold-
mann perimetry (17% on average).
For I4e stimuli, there were statisti-
cally significant differences for all direc-
tions: Octopus 900 (with or without RT
vector and with or without headphone)
produced bigger distances than Gold-
mann perimetry did (p-value < 0.001).
Octopus 900 with RT vector produced
distances that were on average 56%
wider in comparison to Goldmann,
Fig. 1. Visual field examination of a patient from study group 1 with left superior quadran-
Octopus without RT vector without
tanopia, right eye tested. The isopter with the widest range was created with a V4e stimulus,
whereas the smaller isopter was created with a I4e stimulus. (A) Goldmann perimetry. (B) HFA3.
headphone was 32% wider and with
(C) Octopus 900, the double headed arrows show the RT vector. headphone was 29% larger on average.

e502

Table 4. Median mean isopter radius for V4e stimuli (in degrees) in study group 2.
95% Confidence
interval (J–I)
Mean isopter radius Median (Q25–Q75) Lower Upper p-Value
Goldmann (I) 50.2 (43.1–59.2)
Octopus 900 + RTv (J) 57.0 (51.9–62.7) 0.09 0.25 <0.001
Octopus 900 – RTv – H (J) 54.2 (46.0–59.9) 0.011 0.095 0.130
Octopus 900 – RTv + H (J) 52.4 (44.9–58.1) 0.023 0.081 0.347
RTv = reaction time vector, H = headphone, Q25 = lower quartile, Q75 = upper quartile.
Table 5. Median mean isopter radius for I4e stimuli (in degrees) in study group 2.
95% Confidence
interval (J–I)
Mean isopter radius (MIR) Median (Q25–Q75) Lower Upper p-Value
Goldmann (I) 30.0 (22.3–40.8)
Octopus 900 + RTv (J) 47.6 (37.6–52.6) 0.16 0.36 <0.001
Octopus 900 – RTv – H (J) 42.6 (28.3–49.8) 0.086 0.24 <0.001
Octopus 900 – RTv + H (J) 42.0 (29.0–49.1) 0.084 0.22 <0.001
RTv = reaction time vector; H = headphone; Q25 = lower quartile; Q75 = upper quartile.
Figure 2 shows the visual field of a lot of opportunities, a new vector was
patient in study group 2, evaluated in introduced to the Octopus 900: the RT
four different examination modes. vector. This enables to register individ-
ual reaction times for each patient by
presenting a stimulus in the intact
Discussion seeing area of the visual field. The time
In this study, we compared two devices period between the presentation of the
for automated kinetic perimetry stimulus and the reaction of the patient
(Octopus 900 and HFA3) to the refer- is then recorded and represents the
ence technique, (manual) Goldmann patient’s reaction time. The measured
perimetry, thereby using the mean reaction time can then be corrected for
isopter radius and the distance in four each given response (Ramirez et al.
directions. In general, the above results 2008). Earlier studies showed added
indicate that semi-automated kinetic value in measuring the patient’s RT.
perimetry can offer a more standard- For example, an increase in RT can be
ized method to evaluate the peripheral an indicator of increased variability
visual field if compared to Goldmann and fatigue (Nowomiejska et al. 2012).
perimetry. Given that the stimulus In another study, RT appeared to
moves at a constant speed and in the increase along with the age of the
same direction, a more standardized patient, with the shortest RT at the
and reproducible examination can be age of 20–30 years. RT was also longer
achieved. SKP also eliminates the reac- for stimulus combinations producing
tion time of the perimetrist who regis- small isopters compared with those
ters the results, thereby producing producing large isopters (Vonthein
more stable and reliable results. Addi- et al. 2007).
tionally, the Octopus 900 automatically These findings suggest that, taking
measures the VF area of isopters (in into account a patient’s reaction time,
square degrees), hence enabling the use using a RT vector with the Octopus
of quantifiable parameters in the fol- 900, produces larger visual fields. This
low-up. suggestion was confirmed in our study:
Since the Octopus 900 has an orig- Octopus 900 with RT vector, produces
inal Goldmann bowl with a radius of significantly larger isopters in compar-
30 cm in all directions, it aims to retain ison to Goldmann perimetry. The fact
all the original characteristics and spec- that considering a patient’s reaction
ifications of a manual Goldmann time produces larger visual fields is a
perimeter. As the automation brings a logic consequence since a response time
Acta Ophthalmologica 2019
cannot be negative. A more interesting
issue is whether eliminating a patient’s
reaction time reduces visual field vari-
ability. Further research is needed to
investigate this assumption.
To eliminate the influence of the
reaction time vector on Octopus 900, a
second study group was introduced,
comparing Goldmann to the Octopus
without RT vector. These findings did
not show any difference in mean
isopter radius for V4e stimuli on either
Goldmann perimetry or Octopus 900
without RT vector. Overall, the shape
of the visual field was also comparable.
When comparing the mean VF area,
findings for the Goldmann were 16%
smaller then for the Octopus without
RT vector. These findings are compa-
rable to other studies conducted earlier,
including the study of Nowomiejska
(et al.) in which the visual fields from
patients with advanced visual loss were
obtained by using two devices, that is
the Octopus 101 and Goldmann
perimetry, both being quantitatively
compared. The latter research con-
cluded that in general, the Gold-
mann’s VF was 20% smaller than
those for the Octopus 101 (Nowo-
miejska et al. 2005). Rowe & Rowlands
(2014) compared the diagnostic accu-
racy between the Octopus 900 and the
Goldmann kinetic visual fields. This
research again showed that the Gold-
mann results were qualitatively smaller
in the isopter area when compared to
those measured by the Octopus.
A notable issue while conducting
research on the Octopus 900 was the
mechanical sound which can be heard
each time a stimulus is presented by the
movement of the projector. This clearly
audible sound can possibly bias
the patient’s responses, given that the
patient might then anticipate to the
stimulus by the mere sound. Some
patients also reported that the sound
leads to loss of fixation and concentra-
tion. When comparing Goldmann and
Octopus 900 without RT vector and
with the use of a headphone to mask
the sound, no statistically significant
difference was found for V4e mean
isopter radius. Nonetheless, there was
also no difference when the research
was conducted without the applica-
tion of a headphone. This suggests
that the mechanical sound did not
have an influence on the test results.
One final point to note, however, is
that relatively elderly patients (aged
e503
Acta Ophthalmologica 2019

Fig. 2. Kinetic visual field of a patient with glaucoma in study group 2. The isopter with the widest range was created with a V4e stimulus, whereas
the smaller isopter was created with a I4e stimulus. (A) Goldmann perimeter. (B) Octopus 900 with RT vector without headphone, the double headed
arrows show the RT vector. (C) Octopus 900 without RT vector, without headphone. (D) Octopus 900 without RT vector, with headphone.

60 on average) were included in the
study, who could thus already suffer
from presbyacusis. For younger
patients, a headphone might prove
its use not to overestimate the visual
field boundaries.
Another worthy competitor for
Goldmann perimetry, is the Humphrey
field analyzer 3. Besides static perime-
try, the HFA3 also offers a kinetic
mode. HFA3 offers less options in its
kinetic mode, there is for example no
RT vector nor is there a possibility to
automatically measure VF areas.
Nonetheless, the HFA3 is easy to use
and does not produce a specific sound
when presenting the stimuli. Previous
studies never compared the HFA3 to
Goldmann perimetry. In our study, the
mean isopter radius of V4e stimuli is
comparable to Goldmann perimetry.
Distances in temporal, nasal and infe-
rior direction were, however, statisti-
cally significantly wider (in general
25% bigger lengths for temporal, nasal
and inferior direction). In superior
direction, distances on HFA3 were
shorter in comparison to Goldmann
perimetry, but it is important to keep in
mind that the HFA3 only measures the
extent of 40° compared to the Gold-
mann perimeter which has the extent of
70° superiorly.
Unlike the results with the V4e
stimulus, I4e stimuli resulted in overall
statistically significant differences with
SKP in comparison to Goldmann
perimetry. Both Octopus 900 and
HFA3 showed larger MIR and dis-
tances when compared to the reference
technique. Most probably, this relates
to a difference in light intensity and
contrast sensitivity. We noticed a
striking difference in colour tempera-
ture: the Goldmann device, which was
used during the research, produces a
‘warm’ light, whereas the two SKP
devices show a ‘cold’ white LED light.
In small stimuli, patients will probably
have more difficulty in noticing the
dimmer Goldmann stimuli than the
stronger SKP stimuli. Many patients
also reported to our staff that the I4e
stimuli in SKP devices were more

e504

noticeable than those in the Gold-
mann perimeter.
A limitation to this study is the
rather small size of the study group.
Hence no distinction was made among
the different subgroups (i.e. glaucoma,
neurological diseases and control
patients).
In this study, the reliability and
repeatability of the three perimeters
was not assessed. The recently pub-
lished studies of optimal perimetric
testing in children (OPTIC) tested reli-
ability and repeatability of Goldmann
and Octopus 900 kinetic perimetry in
children (Patel et al. 2015). They devel-
oped an examiner based assessment of
reliability (EBAR) scoring system and
a kinetic perimetry reliability measure
(KPRM) that can be used in quantify-
ing within-test variability (Patel et al.
2017). Future studies can use these
parameters to quantitatively measure
reliability of kinetic perimetry in
adults.
Summarizing our experience with
the devices, with hindsight, semi-auto-
mated kinetic perimetry offers a lot of
advantages, not the least its ease in
use. Less involvement of staff is
needed, given the introduction of
computed auto-calibration and the
ability to run pre-defined templates
which can be very helpful for consis-
tency in the follow-up of patients, in
which the pattern of their visual field
defects is already known. Neverthe-
less, we believe that the role of the
examiners should not be underesti-
mated, as they remain to play an
important role in explaining and sup-
porting the patient during the execu-
tion of the test. Furthermore, when a
new patient has to perform kinetic
perimetry, little is known about the
expected borders of his visual field. In
this case, pre-defined templates can
prove to be insufficient in evaluating
the borders of his visual field. Hence,
despite the advanced technology, we
believe that a need for skilled exam-
iners remains crucial.
Conclusion
In this study, both Octopus 900 and
HFA3 prove to be worthy successors
for the reference technique, Goldmann
perimetry. V4e stimuli gave similar
results for the HFA3 and the Octopus
900 (without reaction time vector) when
compared to Goldmann perimetry. It is
however important to take into account
the audible sound which the Octopus
900 produces when presenting a stimu-
lus. No added value was found in using
headphones during the examination,
but some patients did nonetheless expe-
rience more comfort in performing the
exam with headphones. Further studies
are warranted to evaluate the influence
of age on the use of headphones during
Octopus 900 perimetry.
References
Bretz F, Hothorn T & Westfall P (2011):
Multiple Comparisons Using R. Boca
Raton: CRC Press.
Chow S-C, Shao J & Wang H. (2008): Sample
Size Calculations in Clinical Research, 2nd
Edn. Boca Raton: Chapman & Hall/CRC
Press.
Dersu I, Wiggins M, Luther A, Harper R &
Chacko J (2006): Understanding visual
fields, part I; Goldmann perimetry. J Oph-
thalmic Med Technol 2: 1–10.
Konietschke F, Bathke AC, Hothorn LA &
Brunner E (2010): Testing and estimation of
purely non-parametric effects in repeated
measures designs. Comput Stat Data Anal
54: 1895–1905.
Konietschke F, Placzek M, Schaarschmidt F &
Hothorn LA (2015): nparcomp: An R soft-
ware package for nonparametric multiple
comparisons and simultaneous confidence
intervals. J Stat Softw 64: 1–17.
M€onter VM, Crabb DP & Artes PH (2017):
Reclaiming the periphery: automated kinetic
perimetry for measuring peripheral visual
fields in patients with glaucoma. Invest
Ophthalmol Vis Sci 58(2): 868–875.
Nowomiejska K, Vonthein R, Paetzold J,
Zagorski Z, Kardon R & Schiefer U
(2005): Comparison between semiautomated
kinetic perimetry and conventional Gold-
mann manual kinetic perimetry in advanced
visual field loss. Ophthalmology 112: 1343–
1354.
Acta Ophthalmologica 2019
Nowomiejska K, Vonthein R, Paetzold J,
Zagorski Z, Kardon R & Schiefer U
(2010): Reaction time during semi-auto-
mated kinetic perimetry (SKP) in patients
with advanced visual field loss. Acta Oph-
thalmol 88: 65–69.
Nowomiejska K, Brzozowska A, Zarnowski T,
Rejdak R, Weleber RG & Schiefer U (2012):
Variability in isopter position and fatigue
during semi-automated kinetic perimetry.
Ophtalmologica 227: 166–172.
Patel DE, Cumberland PM, Walters BC,
Russell-Eggitt I & Rahi JS (2015): Study of
optimal perimetric testing in children
(OPTIC): feasibility, reliability and repeata-
bility of perimetry in children. Ophthalmol-
ogy 114: 1065–1072.
Patel DE, Viswanathan AC, Garway-Heath D
et al. (2017): Study of optimal perimetric
testing in children (OPTIC): development
and feasibility of the kinetic perimetry reli-
ability measure (KPRM). Br J Ophthalmol
101: 94–96.
Ramirez M, Chaya C, Gordon L & Giaconi J
(2008): A comparison of semiautomated
versus manual Goldmann kinetic perimetry
in patients with visually significant glau-
coma. J Glaucoma 17: 111–117.
Rowe F & Rowlands A (2014): Comparison of
diagnostic accuracy between Octopus 900
and Goldmann kinetic visual fields. Biomed
Res Int 2014: 214829.
Verbeke G & Molenberghs G (2000): Linear
Mixed Models for Longitudinal Data. New
York: Springer-Verlag.
Vonthein R, Rauscher S, Paetzold J, Now-
omiejska K, Krapp E et al. (2007): The
normal age-corrected and reaction time
corrected isopter derived by semi-automated
kinetic perimetry. Ophthalmology 114:
1065–1072.
Received on April 7th, 2018.
Accepted on September 14th, 2018.
Correspondence:
Ingeborg Stalmans
Department of Ophthalmology
Kapucijnenvoer 33
3000 Leuven
Belgium
Tel: +32 16 332372
Fax: +32 16 330467
Email: ingeborg.stalmans@mac.com
e505