Abstract
purpose. To assess the accuracy, precision, repeatability, and reproducibility
of measurements made by the Humphrey optical coherence tomography (OCT)
system (Humphrey-Zeiss Medical Systems, San Leandro, CA).
methods. The performance of the system was first investigated by scanning a test
object containing an air gap of known size. Measurements were repeated
with water or glycerin in the gap. In the clinical setting, macular
thickness measurements were obtained from a control group of 20 normal
subjects. For analysis, these scans were divided into eight sections,
each containing 10 A-scans.
results. The average gap thickness was found to be close to the true value in
all cases. The overall coefficients of intersession reproducibility
were less than 1% for the test object and 1.51% for the control
group. There was no significant difference between scans acquired
during different sessions. The overall coefficients of repeatability
for the test object were between 0.2% and 1.1% and between 1% and
2% for the control group. The range of normal retinal thickness in
terms of the 5th and 95th percentiles was 222 to 248 μm in women and
234 to 257 μm in men.
conclusions. Measurements made from OCT scans were found to be accurate and precise.
Introducing water or glycerin into the test object resulted in
considerable degradation of the signal, but measurements of gap
thickness were still shown to be accurate, precise, reproducible, and
repeatable. Retinal thickness measurements in the macular area were
repeatable and reproducible. This demonstrates that OCT is a useful
tool in the monitoring of patients with conditions that affect macular
thickness, even when there is considerable degradation of the OCT
signal.
Optical coherence tomography (OCT) is becoming an
increasingly popular imaging tool in ophthalmology because of its
ability to yield cross-sectional images of the retina with a resolution
that is considerably greater than that obtained from ophthalmic
ultrasound. A number of publications have provided detailed
descriptions of the functional characteristics of OCT
1 2 3 and of the application of this technique in the clinical
environment.
3 4 5 6 7 8 9 10 11 12 13 The OCT software is designed to make
measurements of retinal thickness or nerve fiber layer (NFL) thickness
from the acquired scans, a feature that permits more effective
monitoring of patients with conditions that cause variations in retinal
or NFL thickness, such as diabetes, glaucoma, and diseases that lead to
macular edema. There have been several publications that demonstrate
the value of this measurement facility in the clinical
setting.
14 15 16 17 18 19 20 21 22 23 24 25
To evaluate changes in macular thickness it is first necessary to
determine the range of retinal thickness in the normal population and
to quantify the accuracy, reproducibility and repeatability of
measurements made by the system. At present, there is only one
commercially available OCT system, manufactured by Humphrey-Zeiss
Medical Systems (San Leandro, CA), and several groups have evaluated
the repeatability or reproducibility of both this prototype system and
the commercially available one and have published values for the range
of retinal thickness or NFL thickness in a control
population.
14 16 26 27 28 29 30 31 However, there do not seem
to be consistent definitions of the terms repeatability and
reproducibility. One purpose of this study was therefore to quantify
both the repeatability and the reproducibility of the commercially
available OCT scanner by basing our definitions of repeatability and
reproducibility on the standards set by the British Standards
Institution and the International Standards
Organization,
32 33 as recommended by Bland and
Altman.
34 In addition to assessing repeatability and
reproducibility, it is important to quantify the accuracy and precision
of the measurements made by the system, a concern that does not appear
to have been investigated so far. It is also worth assessing the
performance of the equipment in situations in which the reflected OCT
signal is somewhat degraded and boundaries are less well defined. Thus,
another purpose of this study was to analyze the performance of the
system by scanning a custom-built object of precisely known size under
three different conditions.
Materials and Methods
Equipment
The scanner used in this study was the commercially available
OCT scanner manufactured by Humphrey-Zeiss Medical Systems. Scanning is
performed using a superluminescent diode operating with a
wavelength of 850 nm and maximum power of 750 μW. Each B-scan
consists of 100 A-scans, regardless of the length of the scan line, and
images are displayed as a pseudocolor plot in which different colors
represent differences in the reflective properties of the retinal
tissue.
The custom-built test object consisted of two 1-cm-thick plates of
glass separated by four 200-μm-thick spacers. The thickness of the
spacers was known to an accuracy of 0.5 μm. A technique called
optical contacting made it possible to attach the spacers to the glass
plates without the use of any sort of adhesive. This technique is a
process by which two surfaces adhere to one another through molecular
attraction. The surfaces of the plates to be contacted are parallel to
within 0.5 arc-second, extremely flat, and cleaned to an exceptionally
high degree. When brought together, the surfaces then adhere with no
adhesive. This technique ensures that the thickness of the gap
corresponds exactly to the thickness of the spacers. Thus, this object
basically provided us with a gap of precisely known thickness. This gap
could also be filled with liquids. Scans of the test object resulted in
two reflecting bands representing the glass–air or glass–liquid
boundaries. The plates of the test object were made of fused silica
with a refractive index of 1.452 at 840 nm. Thus, imaging of the
air-filled gap resulted in very strong reflections due to the large
change in refractive index at the boundary. In the eye, reflections
tend to be less pronounced; and to model the in vivo situation more
effectively the experiment was therefore repeated with the gap filled
with water (refractive index, 1.333) and glycerin (refractive index,
1.473). This had the effect of degrading the intensity of the reflected
signal and making the glass–liquid boundary far less well defined.
The OCT software assumes a refractive index of 1.38 for retinal tissue
(this index value was provided by the manufacturers of the system);
thus, measurements made from the A-scan were multiplied by this index
to convert them back into measurements in air. These values were then
divided by the refractive index of the material within the gap to
arrive at the true gap thickness, as measured from the OCT scan. The
results were then compared with the known thickness of the gap.
Subjects
Twenty volunteers (10 men and 10 women), ranging in age from 21
to 57 years (average age, 31.9 years) participated in this study. The
study was conducted according to the tenets of the Declaration of
Helsinki, and volunteers gave informed consent after the nature and
intent of the study had been fully explained to them. The exclusion
criterion was history of known retinal disease. All scanning was
performed in the right eye, which was dilated with 1% tropicamide.
Scanning
The known distance between the plates of the test object assumes
an incident beam normal to the sample. Thus tilting the sample would
increase the distance traveled by the OCT beam, leading to a higher
measurement. Each pixel within the A-scan represents a distance of 4μ
m in retinal tissue that corresponds to 5.5 μm in air, 4.1 μm in
water and 3.7 μm in glycerin; thus, OCT measurements can distinguish
only between measurements that vary by more than these amounts. From
geometric calculations and experimentation, it was found that tilts
within 10° from normal caused inaccuracies that were less than the
intrinsic thickness resolution of the system. Thus, the positioning of
the test object was not particularly critical for making accurate
measurements. It was observed that the intensity of the reflections
from the interfaces varied with different focusing and polarization
settings. In the case of the air-filled gap, these were adjusted to
give the strongest possible signal. Several scans across the surface of
the object were then acquired.
For the cases in which the gap was filled with liquid, we wanted to
quantify the degradation of the signal. The object was first set up
with an air-filled gap, as just explained, and then the liquid was
introduced carefully without altering the position of the object or the
polarization and focusing settings. Scans were acquired before and
after the liquid was inserted, so that the reduction in intensity of
the reflections could be assessed.
In our study of normal subjects we initially took a series of
horizontal single-line scans across the fovea in each subject. Scans
were repositioned, using the repeat-scan feature that provides a
landmark cursor to facilitate the repeat positioning of subsequent
scans. We then discarded any scans in which the landmark cursor was not
in the correct position, thus ensuring that all the saved scans were of
exactly the same portion of retina. However, after careful analysis of
the fundus pictures it was discovered that there was a degree of
inaccuracy in the positioning of this landmark cursor. We found that
there was some displacement, even between scans in which the landmark
cursor appeared to be in exactly the same position. This displacement
was of the order of 0.2 to 0.3 mm. In the region of the fovea where the
thickness of the retina is varying, a shift of this amount could cause
considerable variations in the measured retinal thickness and would
lead to inaccuracies in the coefficients of repeatability and
reproducibility.
In an attempt to minimize the effects of landmark positioning errors,
we therefore decided to scan across a band rather than across a single
line. This was achieved by having a number of very closely spaced scan
lines using the raster six-lines option, which allows six tomographic
scans to be acquired in succession. In this scanning mode, an aiming
rectangle of adjustable dimensions is displayed on the fundus-viewing
unit. The width of this rectangle determines the length of the scan
lines, whereas its height determines the spacing between the scans. In
this case, the width of the aiming rectangle was set at 4 mm and its
height was 0.5 mm. Thus, the spacing between successive scans was 0.1
mm. The aiming rectangle was positioned such that at least four of the
scans traversed and were centered on the foveal pit. These four scans
therefore covered a vertical length of 0.3 mm, which corresponds to the
maximum error in positioning found from our initial investigation.
These four scans from each group were then used in subsequent
calculations. The focusing and polarization settings were adjusted so
that the best-quality image was obtained.
Our definitions of repeatability and reproducibility were based on the
definitions adopted by the British Standards
Institution.
32 33 Under repeatability conditions,
independent test results are obtained with the same method on the same
subject by the same operator and on the same set of equipment, with the
shortest time lapse possible between successive sets of readings. We
investigated repeatability initially on the test object by acquiring 10
scans in rapid succession. Repeatability on the control group was
investigated by obtaining 10 sets of six tomographic scans from the
same subject. All scanning was performed by the same operator. The time
elapsed between successive sets of scans corresponded to the time taken
to set up and position the aiming rectangle for a new set of
tomographic scans and was always less than 1 minute. Repeatability was
investigated for three different subjects.
Under reproducibility conditions sets of readings are obtained using
the same method but on different occasions. Intersession
reproducibility for the test object was investigated by acquiring
readings in the morning and afternoon on five consecutive days. The
time separation between the morning and afternoon sessions was at least
5 hours, and the OCT scanner was not switched off during this period.
The temperature of the room varied by approximately 2° during that
time. We then analyzed intersession reproducibility in the clinical
setting for each of the 20 subjects by obtaining two sets of six
tomographic scans with a minimum time separation of 30 minutes.
Analysis
A-scans of the test object showed two peaks corresponding to the
edges of the air gap. We decided to measure the thickness of the gap by
considering the distance between the two maxima. Computer programs that
identified the peaks and calculated the distance between them were
developed, because the software provided with the OCT scanner could not
perform these functions. When the gap was filled with air, the gap
edges were very well defined, and the two maxima corresponding to the
glass–air interfaces were easily identifiable on each of the A-scans.
Filling the gap with either water or glycerin caused a reduction in the
overall intensity of the reflections from the interfaces. At some
positions along the scan line, the returning reflection was so weak
that it fell below the noise threshold, and this meant that the gap
edges no longer appeared as continuous lines on the B-scan but had a
more patchy appearance. At these positions, it was impossible to
identify the two maxima corresponding to the gap edges from the A-scan.
Thus, thickness measurements were made only from the A-scans in which
the two maxima could clearly be identified.
To quantify the degradation of the signal caused by introducing a
liquid into the air gap we calculated the percentage reduction in
reflectance. For each B-scan acquired as part of the repeatability
study, we selected the A-scans in which the two maxima were clearly
identifiable and quantified the intensity of the reflections from the
interfaces These values were then averaged over the entire B-scan. This
value was then divided by the average intensity calculated from the
scans acquired just before the liquid was introduced. In the
water-filled condition, the intensity of the reflection from anterior
edge of the gap (closest to the machine head) was found to be 48.5% of
the air-filled condition; for the posterior edge it was 57.0%. With
glycerin in the gap, these values were 42.9% for the anterior edge and
47.7% for the posterior.
For the water-filled gap, an average of 97 A-scans per B-scan showed
two easily identifiable maxima, whereas in the glycerin-filled
condition this value was 45 A-scans per B-scan.
In the normal control study, four scans that traversed the foveal pit
were selected from each set of six tomographic scans, and only these
were used for determining the coefficients of repeatability and
reproducibility. These scans were labeled level 1 to level 4, with
level 1 being the most inferior scan of the set. The retinal thickness
along each point of each scan was found by using the retinal thickness
tool provided with the OCT software. This software assumes that the
first highly reflecting band corresponds to vitreoretinal interface and
that the second corresponds to the retinal pigment epithelium. Thus,
retinal thickness measurements are made by evaluating the displacement
between the anterior surfaces of these two interfaces. The results from
this tool cannot be exported directly from the system; however, the
manufacturer provides a separate program that exports this data as a
text file. Thickness values were thus exported for each scan, and any
obvious errors in boundary detection were corrected manually.
The center of each scan was taken to be the thinnest point of the
retina, which was assumed to correspond to the deepest portion of the
foveal pit. The A-scan at the center of the scan was labeled
A
0. Scans to the left of this were labeled
A
−1 to A
−49, and scans to
the right were labeled A
1 to
A
50. Each scan was then divided into eight
sections, each containing 10 A-scans. Sections to the left of the
center were labeled S
−1 to
S
−4 and those to the right were labeled
S
1 to S
4 as shown in
Figure 1 . Thus, for each B-scan, only the 80 A-scans from
A
−39 to A
40 were used in
the calculations. The retinal thickness obtained from each of these 80
scans was averaged across the four levels in each set in an attempt to
minimize the effects of errors in scan positioning. Thus, we were left
with a single set of 80 thickness values from each group of tomographic
scans. The overall average retinal thickness (the average of these 80
thickness values) and the average retinal thickness per section were
calculated for each group of data.
The median and the 5th and 95th percentiles of the overall retinal
thickness and retinal thickness per section were calculated for the
sample as a whole, as well as for the women and men alone, because Hee
et al.
14 found that the foveal thickness is significantly
different between men and women. This group used the Student’s
t-test for their analysis, which implies that the data are
normally distributed. However, we opted for nonparametric tests,
because we could not be sure that the retinal thickness in the macular
region follows a normal distribution.
As suggested by Bland and Altman
34 who based their
definitions on the recommendations of the British Standards
Institution, the coefficient of intersession reproducibility was
defined as the SD of the differences between pairs of measurements
obtained during different sessions divided by the average of the means
of each pair of readings. Intersession repeatability was evaluated for
the test object from the average air gap thickness computed from each
session. The overall coefficient of intersession reproducibility for
the control group was calculated from the 20 overall average retinal
thickness values. Coefficients of reproducibility were also calculated
for each of the eight retinal sections. A graph of differences against
means was plotted both for the overall average retinal thickness and
for each section. In both cases, the Wilcoxon matched-pairs test (5%
significance level) was also used to establish whether there was any
statistically significant difference between measurements obtained
during different sessions.
The coefficient of repeatability obtained from the repeated
administration of the test under identical conditions was defined as
the SD of the difference from the mean of these repeat measurements
divided by the average response. Coefficients of repeatability were
calculated from the 10 consecutive scans of the test object, as well as
in each of the three subjects participating in the repeatability study.
To establish whether scanning across a 0.3-mm band of retina, rather
than across a single line, actually improved the repeatability, we also
computed the coefficient of repeatability for each level and compared
that with the value obtained from the average over four levels.
Results
Gap Thickness Measured from Test Object
The average gap thickness was calculated from the individual
A-scan thicknesses for each of the 10 consecutive scans in the
repeatability study in all three subjects. These values were then
averaged over the 10 scans, and the final values for average gap
thickness were found to be 197.7 ± 0.6 (air-filled), 195.1 ± 0.7 (water-filled), and 196.1 ± 1.7 μm (glycerin-filled).
Intersession Reproducibility
Test Object.
The average gap thickness was calculated from the individual A-scan
thicknesses for each morning and afternoon session. These values were
then used to compute the overall coefficients of intersession
reproducibility, which were found to be 0.67% (air-filled), 1.05%
(water-filled), and 0.45% (glycerin-filled). The Wilcoxon
paired-measurement test (5% significance level) showed that there were
no statistically significant differences between the measurements
obtained in the morning and afternoon scanning sessions.
Control Group.
For each subject the overall average retinal thickness was calculated
for sessions 1 and 2. The overall coefficient of reproducibility was
then computed from these values and was found to be 1.51%. In
addition, we found the average retinal thickness for each of the eight
sections and computed the coefficient of intersession
reproducibility for each section. The results obtained are shown in
Table 1 . The intraclass correlation coefficient (ICC) for intersession
reproducibility was found to be 0.96. Graphs of differences against
means were plotted for the overall average retinal thickness values, as
well as for each section. We have shown only the graph of differences
against means for the overall retinal thickness
(Fig. 2) ; the graphs for the individual sections were very similar in
appearance. In all cases it was found that 95% (19 of 20) of
differences fell within 2 SDs of the mean. According to the definitions
of the British Standards Institution
33 this indicates
reproducibility in overall and sectional retinal thickness
measurements. The Wilcoxon paired measurement test (5% significance
level) was also performed on the overall retinal thickness values and
on the values for each retinal section. No statistically significant
differences were found between the two sets of data. There is therefore
no evidence to suggest that slight variations in room temperature (on
the order of 2°C) have any effect on the performance of the OCT
scanner. With a study involving 20 patients, the probability is 80%
that the study will detect a difference in morning and afternoon
sessions if the true difference between sessions is 2.18 μm
(two-tailed 5% significance level).
Repeatability
Test Object.
The coefficients of repeatability, calculated from the average gap
thickness computed for each of the 10 consecutive scans, were found to
be 0.29% (air-filled), 0.34% (water-filled), and 0.43%
(glycerin-filled).
Control Group.
For each of the three subjects in the repeatability study we computed
the overall coefficient of repeatability by analyzing the overall
average retinal thickness from each set of readings. We then computed
the coefficient of repeatability for each of the eight retinal
sections. The results obtained are shown in
Table 2 .
To determine whether averaging over four levels had any significant
effect on the repeatability of the measurements, we also computed the
overall and per-section single-level coefficient of repeatability for
each of the four individual levels. We then averaged these values for
each section and compared the results with the values obtained for the
multiple-level case. The results for the single-level coefficients of
repeatability are shown in
Table 2 .
Retinal Thickness
The median and the 5th and 95th percentiles of the overall and
sectional retinal thickness values are shown in
Table 3 .
For comparison with the results of Koozekanani et al.,
27 we calculated the average retinal thickness for sections
S
3 and S
4 together, which
was found to be 271 ± 16 μm. We also calculated the average
retinal thickness for sections S
1 and
S
−1 and found that this was 178 ± 18 μm
in the women and 190 ± 24 μm in the men. Wilcoxon analysis (5%
significance level) showed that there were statistically significant
differences in the overall average retinal thickness in the men and
women (values were higher for the male subjects in the study).
Discussion
Before using measurements made from OCT scans as part of a
clinical diagnosis of a condition, it is necessary to ensure that the
repeated scans give consistent results. Our study on the air gap in the
test object showed a high degree of repeatability (0.29%). Even with
considerable degradation of the OCT signal and poorly defined
interfaces, the coefficients of repeatability were still under 1%. The
measurements made from the control group showed that the overall
coefficients of repeatability were between 1% and 2% and sectional
coefficients of repeatability were all less than 5%. As expected,
these values were slightly lower than for the test object, due to the
introduction of additional errors, such as inaccuracies in positioning
of scans, but nevertheless show a high degree of repeatability in the
clinical setting. This indicates that within the same scanning session,
measurements made by the system are repeatable, and hence for clinical
applications, there is no need to take a large quantity of readings for
reliable measures of retinal thickness. Our results from the
liquid-filled test object indicate that even in cases in which retinal
disease causes degradation of the OCT signal, provided that the
position of vitreoretinal interface and the retinal pigment epithelium
can be identified at a least a few positions along the scan line,
measurements of retinal thickness should still show a high degree of
repeatability.
The overall coefficients of intersession reproducibility were found to
be between 0.67% and 1.05% for the test object and 1.5% for the
control group. Sectional coefficients of reproducibility were all less
than 5%. This indicates that any significant variation in retinal
thickness measurements from different scanning sessions is likely to be
due to changes in the patient’s retinal thickness rather than to
inconsistencies in the values given by the OCT system. Thus, OCT may be
used to monitor patients with conditions that affect the thickness of
the retina in the macular region, even in situations in which the
retinal interfaces are poorly defined.
The average test object gap thicknesses computed under the three
different conditions agree very closely with the known thickness of the
air gap, which was 200 μm. This indicates a high level of accuracy
and precision in the measurements made by the system, even in
situations in which the OCT signal is relatively weak.
From
Table 2 it is clear to see that the repeatability of the retinal
measurements made over a band of 0.3 mm is consistently better than
when repeated scans are made across a single line. This confirms that
slight errors in scan positioning occur and consequently, that our
method of acquiring a series of scans across a 0.3-mm band yields more
reliable measures of repeatability and reproducibility of the system
than simply scanning repeatedly across a single line. These errors in
positioning may be partly because the quality of the fundus picture
displayed on the fundus-viewing monitor is somewhat poor. Moreover,
whereas in the slow-scanning mode, the landmark cursor tends to lose
its definition, thus making it very difficult to ensure that it remains
in the correct position. It is hoped that future versions of the
hardware will include a better quality fundus-viewing unit that would
enable more precise repeat scanning.
Our methods of quantifying repeatability and reproducibility of retinal
thickness measurements in the foveal region differ slightly from those
used in other publications; however, the results obtained are quite
similar. Koozekanani et al.
27 analyzed sets of scans
obtained during independent measuring sessions. They found that there
were no significant differences between different sessions or between
different scans within the same session; however, they do not specify
which statistical test was used. We performed the Wilcoxon
matched-pairs test on the data obtained under reproducibility
conditions and found that there were no significant differences between
the sets of data acquired during different scanning sessions. This is
true both for the overall retinal thickness as well as for the
thickness in each retinal section.
Our method of subdividing each scan into sections containing 10 A-scans
is very similar to the system adopted by Baumann et al.
26 They divided their images into seven regions, each containing 10
A-scans, and computed the coefficient of variation for retinal
thickness measurements made in each of these sections. They found that
the greatest coefficient of variation occurred in the central section,
which covered a retinal length of 320 μm centered on the foveal pit.
Sections closest to the fixation point showed less reproducibility than
those farther away. We calculated the coefficients of repeatability and
reproducibility for each of our retinal sections and similarly found
that these coefficients tended to be highest for regions
S
1 and S
−1, which
correspond to the sections closest to the center of the fovea. In this
region the retinal thickness varies, and hence any errors in scan
positioning will causes variations in the measured thickness of this
region. We have attempted to compensate for this by scanning across a
0.3-mm band. However, although we have shown that this reduces the
effects of errors in positioning, the higher coefficients of
repeatability and reproducibility in these regions relative to other
regions indicate that there is still a degree of inaccuracy in the
positioning of scans.
It is also important to establish confidence intervals for retinal
thickness in the normal population. We therefore computed the median
and the 5th and 95th percentiles for retinal thickness in the macular
region. Our control group was not ideal, because all our subjects were
relatively young. Nevertheless, our results compare very well with
those obtained by other investigators.
The size of our sample and the mean age are similar to those
investigated by Baumann et al.,
26 and our sectional
results are comparable to those quoted in their publication. The
average retinal thickness for sections S
3 and
S
4 together was 271 ± 16 μm. This
represents an average over an 0.8-mm section of retina at a distance of
0.8 mm from the foveal pit, and the result is almost identical with the
average retinal thickness of 274 ± 17 μm of Koozekanani et
al.
27 for a 1-mm section at a distance of 0.75 mm from the
foveal pit. The average retinal thickness for sections
S
1 and S
−1 together was
178 ± 18 μm in the women and 190 ± 24 μm in the men.
These values are higher than the foveal thicknesses of Hee et
al.
14 of 169 ± 4 μm for the female and 181 ±
4 μm for the male subjects. This difference is probably because Hee
et al. analyzed a circular region of 500-μm diameter, whereas our
analysis was performed on a larger rectangular region of 800 ×
300 μm.
In this study we concentrated on total retinal thickness within the
macular area and showed that measurements made in this area are
accurate and precise and that they demonstrate a high level of
repeatability and reproducibility. This implies that OCT can reliably
be used to monitor patients with conditions that affect macular
thickness. A simple model has shown that in patients in which
interfaces are not very well defined, the measurements made
nevertheless agree very closely with the known value of the distance
being measured and that these measurements still show a high degree of
repeatability and reproducibility. This has important implications for
assessing changes in macular thickness in patients affected by
conditions, such as macular edema, which may degrade the quality of the
OCT image. In the assessment of conditions such as glaucoma and macular
edema, it is becoming increasingly common to make use of six radial
scans to create a retinal map. A possible extension of this study would
be to assess the repeatability and reproducibility of the OCT scanner
in this situation.
Supported by Scottish Office Grant K/MRS/50/C2712.
Submitted for publication March 21, 2001; revised August 24, 2001;
accepted September 10, 2001.
Commercial relationships policy: N.
The publication costs of this article were defrayed in part by page
charge payment. This article must therefore be marked“
advertisement” in accordance with 18 U.S.C. §1734
solely to indicate this fact.
Corresponding author: David Keating, Electrodiagnostic Imaging Unit,
Tennent Institute of Ophthalmology, Gartnavel General Hospital, 1053
Great Western Road, Glasgow G12 0YN, United Kingdom;
d.keating@clinmed.gla.ac.uk.
Table 1. Coefficient of Reproducibility, Overall and for Each Retinal
Section
Table 1. Coefficient of Reproducibility, Overall and for Each Retinal
Section
| Coefficient of Reproducibility (%) |
| Overall | 1.51 |
| S−4 | 1.99 |
| S−3 | 2.24 |
| S−2 | 3.49 |
| S−1 | 4.20 |
| S1 | 4.04 |
| S2 | 2.98 |
| S3 | 1.24 |
| S4 | 1.62 |
Table 2. Coefficients of Repeatability, Overall and for Each Retinal Section and
for the Multiple-Level and Single-Level Case
Table 2. Coefficients of Repeatability, Overall and for Each Retinal Section and
for the Multiple-Level and Single-Level Case
| Subject 1 | | Subject 2 | | Subject 3 | |
| ML | SL | ML | SL | ML | SL |
| Overall | 1.07 | 1.62 | 1.27 | 1.75 | 1.98 | 2.88 |
| S−4 | 1.93 | 2.36 | 0.75 | 1.36 | 1.58 | 2.41 |
| S−3 | 1.28 | 1.87 | 1.08 | 1.74 | 1.58 | 2.39 |
| S−2 | 1.70 | 3.07 | 3.19 | 4.30 | 2.15 | 3.18 |
| S−1 | 3.56 | 5.69 | 4.03 | 5.60 | 4.14 | 7.01 |
| S1 | 2.49 | 3.85 | 3.92 | 5.73 | 4.78 | 7.44 |
| S2 | 1.24 | 2.16 | 4.35 | 5.26 | 1.87 | 3.13 |
| S3 | 0.92 | 1.59 | 1.68 | 2.37 | 1.54 | 2.68 |
| S4 | 1.07 | 1.85 | 1.06 | 1.56 | 1.55 | 2.89 |
Table 3. Median, 5th, and 95th Percentile Values for Retinal Thickness
Table 3. Median, 5th, and 95th Percentile Values for Retinal Thickness
| Whole Sample | | | Females | | | Males | | |
| Median | 5th | 95th | Median | 5th | 95th | Median | 5th | 95th |
| Overall | 238 | 225 | 257 | 233 | 222 | 248 | 248 | 234 | 257 |
| S−4 | 263 | 240 | 297 | 257 | 240 | 280 | 278 | 262 | 307 |
| S−3 | 263 | 244 | 292 | 259 | 243 | 278 | 280 | 259 | 295 |
| S−2 | 232 | 213 | 266 | 229 | 215 | 252 | 243 | 218 | 268 |
| S−1 | 178 | 154 | 213 | 174 | 152 | 202 | 184 | 160 | 213 |
| S1 | 187 | 161 | 218 | 177 | 160 | 205 | 197 | 163 | 222 |
| S2 | 245 | 221 | 267 | 239 | 220 | 251 | 252 | 229 | 269 |
| S3 | 273 | 251 | 289 | 272 | 247 | 288 | 276 | 262 | 289 |
| S4 | 272 | 242 | 291 | 272 | 240 | 286 | 283 | 249 | 292 |
Huang D, Swanson EA, Lin CP, et al. Optical coherence tomography. Science. 1991;244:1178–1181.
Hee MR, Izatt JA, Swanson EA, et al. Optical coherence tomography for ophthalmic imaging.
IEEE Eng Med Biol. 1995;14:67–76.
[CrossRef] Puliafito C, Hee MR, Schuman JS, Fujimoto JG. Optical Coherence Tomography of Ocular Diseases. 1996; Slack, Inc Thorofare, NJ.
Baumal CR. Clinical applications of optical coherence tomography.
Curr Opin Ophthalmol. 1999;10:182–188.
[CrossRef] [PubMed]Puliafito CA, Hee MR, Lin CP, et al. Imaging of macular diseases with optical coherence tomography.
Ophthalmology. 1995;102:217–229.
[CrossRef] [PubMed]Hee MR, Izatt JA, Swanson EA, et al. Optical coherence tomography of the human retina.
Arch Ophthalmol. 1995;113:325–332.
[CrossRef] [PubMed]Gaudric A, Haouchine B, Massin P, et al. Macular hole formation: new data provided by optical coherence tomography.
Arch Ophthalmol. 1999;117:744–751.
[CrossRef] [PubMed]Hee MR, Puliafito CA, Wong C, et al. Optical coherence tomography of macular holes. Ophthalmology. 1999;102:748–756.
Chauhan DS, Antcliff RJ, Rai PA, Williamsom TH, Marshall J. Papillofoveal traction in macular hole formation: the role of optical coherence tomography.
Arch Ophthalmol. 2000;118:32–38.
[CrossRef] [PubMed]Jacobson SG, Cideciyan AV, Iannaccone A, et al. Disease expression of RP1 mutations causing autosomal dominant retinitis pigmentosa.
Invest Ophthalmol Vis Sci. 2000;41:1898–1908.
[PubMed]Ip M, Garza-Karren C, Duker JS, et al. Differentiation of degenerative retinoschisis from retinal detachment using optical coherence tomography.
Ophthalmology. 1999;106:600–605.
[CrossRef] [PubMed]Hee MR, Baumal CR, Puliafito CA, et al. Optical coherence tomography of age-related macular degeneration and choroidal neovascularization.
Ophthalmology. 1996;103:1260–1270.
[CrossRef] [PubMed]Schaudig U, Hassenstein A, Bernd A, et al. Limitations of imaging choroidal tumours in vivo by optical coherence tomography.
Graefes Arch Clin Exp Ophthalmol. 1998;236:588–592.
[CrossRef] [PubMed]Hee MR, Puliafito CA, Duker JS, et al. Topography of diabetic macular edema with optical coherence tomography. Ophthalmology. 1998;15:360–369.
Hee MR, Puliafito CA, Wong C, et al. Quantitative assessment of macular edema with optical coherence tomography.
Arch Ophthalmol. 1995;113:1019–1029.
[CrossRef] [PubMed]Otani T, Kishi S, Maruyama Y. Patterns of diabetic macular edema with optical coherence tomography.
Am J Ophthalmol. 1999;127:688–693.
[CrossRef] [PubMed]Otani T, Kishi S. Tomographic assessment of vitreous surgery for diabetic macular edema.
Am J Ophthalmol. 2000;129:487–494.
[CrossRef] [PubMed]Mikajiri K, Okada AA, Ohji M, et al. Analysis of vitrectomy for idiopathic macular hole by optical coherence tomography.
Am J Ophthalmol. 1999;128:655–657.
[CrossRef] [PubMed]Watanabe M, Oshima Y, Emi K. Optical cross-sectional observation of resolved diabetic macular edema associated with vitreomacular separation.
Am J Ophthalmol. 2000;129:264–267.
[CrossRef] [PubMed]Iida T, Hagimura N, Sato T, Kishi S. Evaluation of central serous chorioretinopathy with optical coherence tomography.
Am J Ophthalmol. 2000;129:16–20.
[CrossRef] [PubMed]Parisi V, Manni G, Gandolfi SA, Centofanti M, Colacino G, Bucci MG. Visual function correlates with nerve fibre layer thickness in eyes affected by ocular hypertension.
Invest Ophthalmol Vis Sci. 1999;40:1828–1833.
[PubMed]Zangwill LM, Williams J, Berry CC, Knauer S, Weinreb RN. A comparison of optical coherence tomography and retinal nerve fiber layer photography for detection of nerve fiber layer damage in glaucoma.
Ophthalmology. 2000;107:1309–1315.
[CrossRef] [PubMed]Bowd C, Weinreb RN, Williams JM, Zangwill LM. The retinal nerve fiber layer thickness in ocular hypertensive, normal, and glaucomatous eyes with optical coherence tomography.
Arch Ophthalmol. 2000;118:22–26.
[CrossRef] [PubMed]Hoh ST, Greenfield DS, Mistlberger A, Liebmann JM, Ishikawa H, Ritch R. Optical coherence tomography and scanning laser polarimetry in normal, ocular hypertensive, and glaucomatous eyes.
Am J Ophthalmol. 2000;129:129–135.
[CrossRef] [PubMed]Parisi V, Manni G, Spadara M, et al. Correlation between morphological and functional retinal impairment in multiple sclerosis patients.
Invest Ophthalmol Vis Sci. 1999;40:2520–2527.
[PubMed]Baumann M, Gentile RC, Liebmann JM, et al. Reproducibility of retinal thickness measurements in normal eyes using optical coherence tomography.
Ophthalmic Surg Lasers. 1998;29:280–285.
[PubMed]Koozekanani D, Roberts C, Katz SE, et al. Intersession repeatability of macular thickness measurements with the Humphrey 2000 OCT.
Invest Ophthalmol Vis Sci. 2000;41:1486–1491.
[PubMed]Chauhan DS, Marshall J. The interpretation of optical coherence tomography images of the retina.
Invest Ophthalmol Vis Sci. 1999;40:2332–2342.
[PubMed]Gurses-Ozden R, Ishikawa H, Hoh ST, et al. Increasing sampling density improves reproducibility of optical coherence tomography measurements.
J Glaucoma. 1999;8:238–241.
[PubMed]Blumenthal EZ, Williams JM, Weinreb RN, Girkin CA, Berry CC, Zangwill LM. Reproducibility of nerve fiber layer thickness measurements by use of optical coherence tomography.
Ophthalmology. 2000;107:2278–2282.
[CrossRef] [PubMed]Schuman JS, Pedut-Kloizman T, Hertzmark E, et al. Reproducibility of nerve fiber layer thickness measurements using optical coherence tomography.
Ophthalmology. 1996;103:1889–1898.
[CrossRef] [PubMed]British Standards Institution. Accuracy (trueness and precision) of measurement methods and results. General principles and definitions. 1994; British Standards Institution London. BS ISO 5725 part 1
British Standards Institution. Accuracy (trueness and precision) of measurement methods and results: basic methods for the determination of repeatability and reproducibility of a standard measurement method. 1994; British Standards Institution London: . BS ISO 5725 part 2
Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet. 1986;2:307–310.