Detection of Localized Retinal Nerve Fiber Layer Defects in ...

Detection of Localized Retinal Nerve Fiber
Layer Defects in Glaucoma Using Enhanced
Spectral-Domain Optical Coherence
Tomography
Masayuki Nukada, MD, Masanori Hangai, MD, Satoshi Mori, MD, Noriko Nakano, MD,
Hideo Nakanishi, MD, Hanako Ohashi-Ikeda, MD, Atsushi Nonaka, MD, Nagahisa Yoshimura, MD
Objective: To compare retinal nerve fiber layer (RNFL) defects on fundus photographs with circumpapillary
RNFL (cpRNFL) thinning or disruption on images obtained by speckle-noise–reduced spectral-domain optical
coherence tomography (enhanced SD OCT), single-scan SD OCT, and single-scan time-domain OCT (TD OCT).
Design: Retrospective, comparative case series.
Participants: Forty-four eyes of 44 patients with open-angle glaucoma with localized, wedge-shaped RNFL
defects on red-free photographs and 35 normal eyes of 35 volunteers.
Methods: Cross-sectional images of the cpRNFL and cpRNFL thinning, compared with the confidence
interval limit of the normative database where the RNFL defect was photographically identified, were compared
between the 3 types of OCT instruments: enhanced SD OCT (SD OCT with eye tracking and averaging of 16
images at the same location to reduce speckle noise; Spectralis HRAOCT; Heidelberg Engineering, Heidelberg,
Germany), single-scan SD OCT (RTVue-100; Optovue, Fremont, CA), and single-scan TD OCT (Stratus; Carl
Zeiss-Meditec, Dublin, CA).
Main Outcome Measures: Cross-sectional images of localized RNFL defects on red-free fundus photographs,
sensitivity for detecting the photographic RNFL defect, and sensitivity and specificity for detecting
glaucoma as having at least 1 abnormally thinned sector on the cpRNFL thickness map on OCT.
Results: Among the 44 eyes with glaucoma, 65 RNFL defects were identified on red-free fundus photographs.
The cpRNFL boundaries were clearer on enhanced SD OCT images than on single-scan SD OCT or TD
OCT images, particularly in regions corresponding to the RNFL defects. Enhanced SD OCT revealed various
degrees of cpRNFL thinning, and disruption of cpRNFL reflectivity was seen in the same location as the
photographic RNFL defect for 23 (35.4%) of the 65 RNFL defects. The RNFL defects were significantly less likely
to be detected by single-scan TD OCT or SD OCT (P 0.002 and P 0.006, respectively) when the RNFL was
not disrupted. Enhanced SD OCT was more sensitive in detecting the RNFL defects that were not disrupted
compared with single-scan TD OCT (P0.0001) or SD OCT (P0.0001). Enhanced SD OCT had better sensitivity
and specificity for detecting glaucoma compared with single-scan TD OCT or SD OCT (sensitivity, P 0.006 and
P 0.001; specificity, P 0.001 and P 0.004, respectively).
Conclusions: These results suggest that speckle-noise reduction can improve the detection of photographic
RNFL defects in which cpRNFL reflectivity on OCT images is not disrupted.
Financial Disclosure(s): Proprietary or commercial disclosure may be found after the references.
Ophthalmology 2011;118:1038–1048 © 2011 by the American Academy of Ophthalmology.
Glaucoma causes progressive damage to the retinal ganglion
cell axons within the optic nerve head, which diminishes
the retinal nerve fiber layer (RNFL). The RNFL defects
are detected clinically before abnormalities in the optic
nerve head appearance or visual field defects become detectable.
1–9 In 60% of reported cases, RNFL defects preceded
detection of visual field defects by approximately 6
years. 2,5,7,8 Hence, accurate early detection and monitoring
of RNFL defects has become a major focus for improved
management of glaucoma. 10
Until recently, RNFL defects could be detected clinically
only as areas of reduced optical reflectance on fundus pho-
tographs. Techniques for objective and quantitative assessment
of RNFL defects would improve glaucoma diagnosis
and management.
Optical coherence tomography (OCT) can provide highresolution,
cross-sectional images of patient eyes on which
the thickness of the circumpapillary RNFL (cpRNFL) can
be measured. Several studies have demonstrated that
cpRNFL thickness measured on OCT images has better
glaucoma detection capability than optic nerve head
changes or macula thickness. 11–15 However, some localized
RNFL defects visible on fundus photographs, particularly
narrow defects, were not detected by time domain (TS)
1038 © 2011 by the American Academy of Ophthalmology ISSN 0161-6420/11/$–see front matter
Published by Elsevier Inc. doi:10.1016/j.ophtha.2010.10.025

OCT (Stratus; Carl Zeiss-Meditec, Dublin, CA). 16–18 These
false-negative results have been attributed to differences in
ethnicity between the study subjects and those used to
determine the normative data, inadequate axial resolution of
TD OCT, poor signal strength of B-scans, and image artifacts
caused by shadows of retinal vessels. 16–20
The present study sought to overcome these limitations
to the reliability of OCT analysis of the cpRNFL as a
glaucoma diagnostic technique. First, characteristics of
RNFL defects seen on fundus photographs that could affect
OCT analysis of the cpRNFL were examined. Jeoung et al 16
and Jeoung and Park 18 showed that defects with a narrower
angular width on fundus photographs are less likely to be
detected by TD OCT cpRNFL analysis. The current authors
speculated that some cross-sectional structures of these narrow
RNFL defects may be responsible for the higher incidence
of false-negative detection.
Measurement of cpRNFL thickness is based on the automated
boundary delineation on OCT B-scan images; there are
2 critical factors in the accurate identification of the boundaries
of retinal layers on OCT B-scans. 21–23 Low axial resolution
may be one of the critical factors that can cause the falsenegative
detection of photographic RNFL defects in TD OCT.
Recently developed spectral-domain (SD) OCT instruments
have axial resolutions that are twice as high (5–7 m) compared
with TD OCT (approximately 10 m). The other, more
critical, factor is speckle noise. 23 Reducing speckle noise can
drastically improve the clarity of boundaries between inner
retinal layers and visualization of small pathologic changes. 24–27
The SD OCT instruments can acquire B-scans 45 to 130 times
faster than TD OCT; moreover, particularly when the SD OCT
system includes a 3-dimensional (3-D) eye-tracking system
(Spectralis HRAOCT; Heidelberg Engineering, Heidelberg,
Germany), multiple B-scans can be acquired at the exact same
location that, when averaged, result in a speckle-noise–reduced
image with clearly distinguishable boundaries between retinal
layers. 21
This study compared TD OCT, single-scan SD OCT, and
speckle-noise–reduced (so-called enhanced) SD OCT images
of the cpRNFL obtained in areas corresponding to
locations of RNFL defects identified on fundus photographs
and intended to estimate the sensitivity of the 3 methods for
the photographic RNFL defect detection and the sensitivity
and specificity for glaucoma detection.
Patients and Methods
Nukada et al RNFL Defect Detection Using OCT in Glaucoma
This study was approved by the Institutional Review Board and
Ethics Committee of Kyoto University Graduate School of Medicine.
All investigations adhered to the tenets of the Declaration of
Helsinki. Informed consent was obtained from all the subjects of
this study.
Patients who underwent color and red-free fundus photography,
3 types of OCT examination, and visual field testing within 2
months were enrolled consecutively and retrospectively from the
database of patients who visited the glaucoma service at Kyoto
University Hospital between July 2008 and December 2008. Patients
who were eligible for this study were those with localized
RNFL defects that met the eligibility criteria. They underwent a
comprehensive ophthalmic examination, including measurement
of uncorrected visual acuity and best-corrected visual acuity with
the use of the 5-m Landolt chart, slit-lamp examinations, intraocular
pressure measurements with a Goldmann applanation tonometer,
gonioscopy, dilated stereoscopic examination of the optic
nerve head, stereo disc photography (3-Dx simultaneous stereo
disc camera; Nidek, Gamagori, Japan), red-free fundus photography
(Heidelberg Retina Angiogram 2; Heidelberg Engineering,
Heidelberg, Germany), standard automated perimetry (SAP)
through the Humphrey Visual Field Analyzer with the 24-2 Swedish
interactive threshold algorithm (HFA 24–2 SITA; Carl
Zeiss-Meditec, Dublin, CA), and 3 types of OCT instruments
(Stratus OCT, RTVue-100 [Optovue, Fremont, CA], and Spectralis
HRAOCT).
The inclusion criteria were as follows: patient eyes had to have
normal open anterior chamber angle, presence of RNFL defects on
red-free fundus photographs that were associated with glaucomatous
appearance of the optic nerve head on stereo color fundus
photographs (diffuse or localized rim thinning and disc hemorrhage,
or vertical cup-to-disc ratio of 0.2 or more than that of the
fellow eye), and presence of glaucomatous visual field defects that
corresponded with the RNFL defects and optic nerve head abnormalities.
Exclusion criteria were best-corrected visual acuity worse
than 20/20 (Snellen equivalent); spherical equivalent refractive
error of less than –6.00 diopters; evidence of vitreoretinal disease,
uveitis, nonglaucomatous optic neuropathy, diabetes mellitus, or
any other systemic disease that might have affected the eye and
visual field results, such as a cerebrovascular event, uncontrolled
hypertension, and blood disorders. When both eyes of a patient
were eligible for this study, 1 eye was selected at random.
Control Eyes
Data for control eyes were collected retrospectively from subjects
who were determined by the department to have at least 1 normal eye
and who agreed to undergo the examinations described in this study.
These volunteers were part of a prospective study designed to evaluate
normal structures of the retina and optic nerve head by using
spectral-domain OCT. 28 Eligible normal control eyes for this study
were normal eyes (intraocular pressure of 21 mmHg or less, a normalappearing
optic disc head, no RNFL defects, normal SAP results, and
no history of chronic ocular disease) of volunteers with no systemic
diseases that might have affected the eyes and no systemic corticosteroid
use. These volunteers should have undergone the same comprehensive
ophthalmic examinations and OCT imaging as the patients.
When both of the volunteer’s eyes were normal, 1 eye was
selected at random to be included in the control group.
Visual Field Testing
Visual field defects resulting from glaucoma were defined on SAP
by using the 24–2 Swedish interactive threshold algorithm standard
as (1) abnormal range on the glaucoma hemifield test or (2)
pattern standard deviation of less than 5% of the normal reference
value confirmed on 2 consecutive tests considered reliable based
on fixation losses of 20% or less, false-positive results of 15% or
less, and false-negative results of 33% or less. For glaucoma
diagnosis, the 2 consecutive visual field tests were performed
within 1 month of each other, and when the results of these did not
agree, a third test was performed in another month.
Optic Disc Evaluation
The appearance of the optic nerve head of both eyes of patients
and volunteers on fundus photographs, including stereoscopic
photographs, was evaluated by 3 glaucoma specialists (MH,
HO-I, and AN) who were masked to all other information about
1039

the eyes. Eyes were classified as having glaucoma if the examiner
identified either diffuse or localized rim thinning, disc
hemorrhage, or a vertical cup-to-disc ratio of 0.2 or more
compared with that of the fellow eye. If all 3 examiners did not
agree on the classification of an eye, the group reviewed and
discussed the fundus color photographs and stereophotographs
until a consensus was reached.
Measurement of Retinal Nerve Fiber Layer
Defects on Red-Free Fundus Photographs
The width of each RNFL defect identified on red-free fundus
photographs was measured in degrees through a previously reported
technique, with modification. 16 First, a circle with a diameter
of 3.46 mm centered on the optic nerve head was drawn on the
red-free image. Next, a line was drawn from the center of the optic
nerve head to each point where the borders of RNFL defects met
the circle. Then, the angle between each pair of lines, representing
the angular width of the RNFL defect, was measured.
Cross-sectional Imaging of Circumpapillary Retinal
Nerve Fiber Layer with Speckle-Noise–Reduced
(Enhanced) Spectral-Domain Optical Coherence
Tomography
Speckle-noise–reduced (enhanced) SD OCT B-scan images of
cpRNFL were obtained through Spectralis HRAOCT (software
version 4.1). 21 This instrument combines confocal laser scanning
ophthalmoscopy, which enables real-time 3-D tracking of eye
movements, with real-time averaging of multiple B-scans (870-nm
axial resolution) acquired at an identical location of interest on the
retina, to reduce speckle noise. 21 For this study, each enhanced SD
OCT image was obtained by averaging 16 circular B-scans, each
with a diameter of 3.46 mm centered on the optic disc, obtained by
acquiring 1536 A-scans at a rate of 40 000 per second, providing
a digital transverse sampling resolution of 5 m per pixel.
Assessment of the Cross-sectional Circumpapillary
Retinal Nerve Fiber Layer Features Corresponding
to Retinal Nerve Fiber Layer Defects on Spectral-
Domain Optical Coherence Tomography Images
Two glaucoma specialists (HN and SM) who were masked to
clinical information about the eyes evaluated the appearance of
cpRNFL in regions corresponding to the locations of RNFL defects
on enhanced SD OCT images. Each specialist independently
categorized the RNFL defect according to whether the enhanced
SD OCT image showed complete loss of cpRNFL reflectivity
(representing disrupted reflectivity) or only decreased thickness of
highly reflective cpRNFL (representing not disrupted reflectivity)
in regions corresponding to the locations of RNFL defects seen on
red-free fundus photography (Fig 1). Before the assessment, the
specialists received training, by observing sample cpRNFL images
on speckle-noise–reduced SD OCT of normal eyes, and eyes with
disrupted and not disrupted cpRNFL reflectivity. The interrater
reliability was calculated for the independent classifications of the
2 glaucoma specialists. When the evaluations of the 2 specialists
did not agree, a third glaucoma specialist (MN) examined the
images and results were discussed by the 3 examiners until a
consensus was reached.
1040
Ophthalmology Volume 118, Number 6, June 2011
Measurement of Circumpapillary Retinal Nerve
Fiber Layer Thickness by Optical Coherence
Tomography
Three types of OCT instruments were used to assess cpRNFL: TD
OCT (Stratus OCT), SD OCT (RTVue-100), and speckle-noise–
reduced (enhanced; each image is obtained with eye tracking and
by averaging 16 B-scans at the same location) SD OCT (Spectralis
HRAOCT).
The built-in software of each of the 3 OCT imaging systems
was used to measure the thickness of the cpRNFL and to calculate
the mean thickness over a defined area (sector). On TD OCT
(Stratus), the fast cpRNFL program (256 A-scans 3 B-scans)
was used to scan a circle with a diameter of 3.43 mm, centered on
the optic nerve head; the software calculated the mean cpRNFL
thickness in each of the 12 clock-hour sectors. For single-scan
cpRNFL analysis with SD OCT (RTVue-100, software version
4.0), the RNFL3.45 scan program was used to obtain a circular
scan with a diameter of 3.45 mm (1024 A-scans 4 B-scans); the
software calculated the mean RNFL thickness in 16 circumferential
(circumpapillary) sectors. For enhanced SD OCT (Spectralis with
software version 4.0C), the cpRNFL scan feature was used to scan a
circle with a diameter of 3.46 mm (1536 A-scans 1 B-scan)
centered on the optic nerve head; mean RNFL thickness was calculated
in each of 6 circumferential (circumpapillary) sectors.
The software for each system also automatically compares the
mean values in each sector with means for an age-matched normative
database. The result is a color-coded map in which sectors
with mean thicknesses between the 95% confidence interval (CI)
values in normal eyes are in green; sectors with mean thicknesses
between the lower 95% CI and lower 99% CI (P0.05 and
P0.01) are in yellow, indicating borderline results; and mean
thicknesses that are below the normal range (less than the lower
99% CI; P0.01) are in red, indicating outside normal results.
To compare the sensitivity of the 3 OCT systems for detecting
the RNFL defect, the RNFL defect detected was defined as abnormal
thinning (outside normal) of the cpRNFL in the sector that
corresponded with the location of the RNFL defect on red-free
fundus photography. To compare the sensitivity and specificity of
the 3 OCT systems for detecting glaucoma, glaucoma detected was
defined as abnormal thinning of the cpRNFL (at least 1 red outside
normal sector on the cpRNFL thickness map).
Statistical Analysis
All statistical analyses were performed with SPSS software version
11.01 J (SPSS, Inc., Chicago, IL). The statistical significance
of differences between values for glaucoma versus normal eyes
was evaluated with the unpaired t test and chi-square test. The
chi-square test also was used to study the relationships in individual
eyes between mean deviation (MD) on visual field testing and
identification of RNFL reflectivity disruption on enhanced SD
OCT images and between the angle of the RNFL defect on red-free
fundus photography and identification of RNFL reflectivity disruption
on enhanced SD OCT images. The Fisher exact test was
used to compare the sensitivity for RNFL defect detection on OCT
and identification of RNFL reflectivity disruption on enhanced SD
OCT images. McNemar’s test was used to compare the sensitivity
for RNFL defect detection on OCT and to compare the sensitivity
and specificity for glaucoma detection between the different OCT
systems. Cohen’s coefficient was used to estimate interrater
reliability of the 2 glaucoma specialists in identifying RNFL
reflectivity disruption on enhanced SD OCT images. The level of
statistical significance was set at P0.05.

Results
Nukada et al RNFL Defect Detection Using OCT in Glaucoma
Figure 1. Cross-sectional enhanced (speckle-noise–reduced) spectral-domain optical coherence tomography images of 14 eyes with retinal nerve fiber
layer (RNFL) defects. A, These 7 eyes had varying degrees of thinning, but no disruption in reflectivity of the RNFL (top layer in each image) in the
circumpapillary area. B, These 7 eyes had disruption in RNFL reflectivity (top layer in each image) in the circumpapillary area. Images on the right are
magnified (4) views of the area outlined by red dashed lines in images on the left.
A total of 158 patients underwent color and red-free fundus
photography, 3 OCT examinations (Stratus OCT, RTvue-100,
and Spectralis), and SAP 24–2 tests within a 2-month period at
the authors’ clinic. Among these, 125 eyes of 80 patients were
determined to meet the inclusion criteria, including having
localized RNFL defects visible on red-free fundus photography.
Of these, 62 eyes of 36 patients were excluded because of
spherical equivalent refraction errors of less than –6 diopters
(D; 57 eyes of 31 patients) and evidence or history of vitreoretinal
disease (5 eyes of 5 patients). Finally, 63 eyes of 44
patients (17 men and 27 women) were eligible for this study.
After 1 eye was selected at random from the patients with both
eyes deemed eligible, 44 eyes of the 44 patients were used for
analysis. Subjects ranged in age from 29 to 76 years
(meanstandard deviation, 55.212.1 years). Visual field
(MD) ranged from 2.30 to –22.6 (meanstandard deviation,
–8.57.0 dB). Refractive error ranged from 1.75 to –6 D
(meanstandard deviation, –3.32.2 D). Of the 44 eyes, 23
had a single RNFL defect identified and 21 had 2 localized
RNFL defects identified on red-free photographs.
Thirty-five eyes of 35 normal volunteers (15 men and 20
women), who ranged in age from 22 to 70 years (meanstandard
deviation, 5413.7 years), were included. The spherical equivalent
of the refractive errors of the 35 normal eyes ranged from 1.5
to –5.75 D (meanstandard deviation, –3.72.9 D). No statistically
significant differences in gender, age, or refractive error were
observed between the patient and control groups. The 3 glaucoma
specialists agreed that a total of 65 RNFL defects were evident in
eyes with glaucoma.
Evaluation of Retinal Nerve Fiber Layer Thinning
on Spectral-Domain Optical Coherence
Tomography Images
The SD OCT analysis with speckle-noise reduction provided improved
clarity of RNFL boundaries, particularly the boundary
between the RNFL and ganglion cell layer. Speckle-noise–reduced
(enhanced) SD OCT imaging revealed varying degrees of thinning
of the cpRNFL in all areas where the 65 RNFL defects were seen
on red-free fundus photographs (Fig 1). For 23 (35.4%) of the 65
RNFL defects, the highly reflective RNFL was disrupted within
the region corresponding to the RNFL defect on red-free fundus
1041

Figure 2. Red-free fundus photographs, single-scan time-domain optical coherence tomography (TD OCT) images, and single-scan and specklenoise–reduced
spectral-domain (SD) OCT images of 2 eyes with retinal nerve fiber layer (RNFL) defects. AG, Right eye of a 44-year-old woman with
open-angle glaucoma and a mead deviation (MD) of –2.47 dB. HM, Left eye of a 68-year-old man with open-angle glaucoma and an MD of –7.09 dB.
A, H, Red-free fundus photographs showing wedge-shaped RNFL defects in the supertemporal sector in (A) and the inferotemporal sector in (H). B, E,
H, K, Speckle-noise–reduced (Spectralis HRAOCT; Heidelberg Engineering, Heidelberg, Germany) images of the circumpapillary (cp) RNFL. C, F, I,
L, Single-scan SD OCT cpRNFL images (RTvue-100; Optovue, Fremont, CA). D, G, J, M, Single-scan TD OCT (Stratus OCT; Carl Zeiss-Meditec,
Dublin, CA) cpRNFL images. B2G2, H2M2, Magnified (4) views of the area outlined by red dashed lines in the corresponding scan (B1G1,
H1M1) where RNFL defects are visible on red-free photographs. Automatically generated lines defining the internal limiting membrane and border
between the RNFL and ganglion cell layer are shown in red on enhanced SD OCT (Spectralis HRAOCT) images (E, K) and in white on single-scan
SD OCT (RTvue-100) images (F, L) and TD OCT images (G, M). Multiple small dots (representing speckle noise) are visible on single-scan SD OCT
and TD OCT cpRNFL images (C, D, F, G, I, J, L, M), but not in on speckle-noise–reduced SD OCT (B, E, H, K) cpRNFL images. The boundaries
between the RNFL and GCL are clearer in speckle-noise–reduced SD OCT images compared with single-scan SD OCT and TD OCT images. The areas
of cpRNFL thinning on OCT images (red dashed lines) depict abrupt cpRNFL corresponding to the areas where RNFL defects appear on red-free fundus
photographs. The cpRNFL images on speckle-noise–reduced SD OCT clearly depict abrupt thinning. The boundary line on Spectralis appears to draw
a line along the abruptly diminished cpRNFL, whereas the boundary lines on RTVue and Stratus do not.
photography (Fig 1B). For the other 42 (64.6%) of the 65 defects,
the highly reflective layer on enhanced SD OCT appeared thinned
but not disrupted (Fig 1A). There was good agreement between the
2 glaucoma specialists (Cohen , 0.840) regarding whether the
RNFL defects on red-free fundus photographs appeared as disruption
of the cpRNFL reflectivity on enhanced SD OCT images.
Automated Boundary Line Delineation on Optical
Coherence Tomography
Figures 2 and 3 show images with boundary lines for the
cpRNFL drawn automatically on single-scan SD OCT (RTVue)
and TD OCT (Stratus OCT) images (lines drawn in white) and
on enhanced SD OCT (Spectralis) images (lines drawn in red).
There was an abrupt thinning of the cpRNFL seen on enhanced
SD OCT (Figs 2B,H and 3E) in areas that corresponded to the
locations of RNFL defects on red-free photographs (Figs 2A,H
and 3E). The depiction of this abrupt thinning of cpRNFL was
less clear in single-scan SD OCT (Figs 2C,I and 3C) or TD OCT
(Figs 2D,J and 3D). The boundary lines accurately delineated
the abrupt thinning of the cpRNFL in enhanced SD OCT images
(Figs 2E,K and 3F). On single-scan SD OCT (Figs 2F,L and
3G) and TD OCT (Figs 2G,M and 3H) images, however, the
1042
Ophthalmology Volume 118, Number 6, June 2011
boundary lines drawn for the cpRNFL did not show abrupt
thinning. The case in Figure 3 indicates that the accurate
delineation of the abrupt thinning of cpRNFL likely causes the
false-negative results when RNFL defects are defined as outside
normal thinning (lower than the lower 99% confidence interval
of normal mean).
Circumpapillary Retinal Nerve Fiber Layer
Disruption and Visual Field Changes
Table 1 shows the number of eyes that include RNFL defects for
which enhanced SD OCT images of the cpRNFL showed no
disruption or disruption as a function of the SAP results (MD
value). No cpRNFL disruption was associated significantly with
lower MD values (P 0.007).
Sensitivity for Detecting Photographic Retinal
Nerve Fiber Layer Defects by Optical Coherence
Tomography in Eyes with Glaucoma
The sensitivity of the 3 OCT systems was compared for detecting
the RNFL defect as abnormal thinning (outside normal) of the
cpRNFL in the sector that correspond with the location of the

Nukada et al RNFL Defect Detection Using OCT in Glaucoma
Figure 3. Fundus photographs, optical coherence tomography (OCT) images, thickness maps, and profiles of thickness of the circumpapillary retinal nerve
fiber layer (cpRNFL) in the right eye of a 60-year-old woman with open-angle glaucoma and a mead deviation (MD) of –2.33 dB. A, Color fundus
photograph showing a neuroretinal rim thinning and disc hemorrhage in the inferior temporal area. BD, FH, OCT images obtained without (BD)
and with (FH) automatically generated lines (red in F, white in G and H) indicating the internal limiting membrane and the boundary between the
RNFL and ganglion cell layer. E, Red-free fundus photograph showing a wedge-shaped RNFL defect in the inferotemporal area. B, F, Speckle-noise–
reduced SD OCT (Spectralis HRAOCT; Heidelberg Engineering, Heidelberg, Germany) images. C, G, Single-scan SD OCT (RTvue-100; Optovue,
Fremont, CA) images. D, H, Single-scan time-domain OCT (Stratus OCT; Carl Zeiss-Meditec, Dublin, CA) images. B2D2, F2H2, Magnified (4)
views of the area outlined by red dashed lines in the corresponding scan (B1D1, F1H1) where an RNFL defect was visible on red-free photographs.
IK, Color-coded sector maps (I-1, J-1, K-1) and thickness profile graphs (I-2, J-2, K-2) of cpRNFL thickness in this patient eye compared with the
mean of the normative data base. Red abnormal thinning (mean thickness less than the lower 99% confidence interval [CI] value in normal eyes).
Yellow borderline thinning (mean thickness between the lower 95% and lower 99% CI values in normal eyes). Green within normal limits (mean
thickness within the 95% CI in normal eyes). At the location of the RNFL defect on the red-free fundus photograph (E), measurements of cpRNFL
thickness on Spectralis SD OCT identified an area of abnormal thinning (red sector on I-1) and abrupt RNFL thinning (red arrow on I-2), but thickness
maps obtained by RTVue (J-1) and Stratus OCT (K-1) show only areas of borderline thinning, and thickness profiles obtained by RTVue (J-2) and Stratus
OCT (K-2) show less abrupt thinning (red arrows). IN inferior nasal; INF inferior; IT inferior temporal; NAS nasal; NL nasal lower;
NU nasal upper; SN superior nasal; ST superior temporal; SUP superior; TEMP temporal; TL temporal lower; TU temporal upper.
RNFL defect on red-free fundus photography (Table 2). All 3
types of OCT showed higher proportions of RNFL defects detected
as cpRNFL thinning (either borderline thinned [P0.05 and
0.01] or abnormally thinned [P0.01]) when the cpRNFL was
disrupted versus not disrupted in the location of RNFL defects on
red-free fundus photographs. However, differences were statistically
significant only for single-scan images (P 0.002 for ab-
normal thinning on Stratus [TD OCT], P 0.006 for borderline
thinning on Stratus, and P 0.0002 for abnormal thinning on
RTVue [SD OCT]; Table 2).
Abnormal thinning of the cpRNFL (P0.01) in the OCT sector
was matched significantly more often to localized RNFL defects
on red-free fundus photographs when the images were obtained by
enhanced SD OCT (Spectralis) compared with single-scan TD
1043

Table 1. Number of Eyes with Glaucoma That Include Retinal
Nerve Fiber Layer Defects for Which Enhanced Spectral-
Domain Optical Coherence Tomography Images of the
Circumpapillary Retinal Nerve Fiber Layer Showed
Circumpapillary Retinal Nerve Fiber Layer Reflectivity
Disruption as a Function of the Perimetry Results
(44 Eyes of 44 Patients)
Mean Deviation (dB)
Retinal Nerve Fiber Layer Reflectivity on
Enhanced Spectral-Domain Optical
Coherence Tomography
Not Disrupted (n 21) Disrupted (n 23)
–6 17 (81.0%) 8 (34.8%)
–6 to –12 2 (9.5%) 5 (21.7%)
–12 2 (9.5%) 10 (43.5%)
P 0.007, chi-square test.
OCT (Stratus) or SD OCT (RTVue) images (P0.0001 for both;
Table 2).
Circumpapillary Retinal Nerve Fiber Layer Disruption
and Retinal Nerve Fiber Layer Defect Angle
Table 3 shows the number of photographic RNFL defects for
which OCT images showed abnormal thinning of the cpRNFL
(P0.01) as a function of angular width of the RNFL defect on
fundus photography and whether the cpRNFL was disrupted on
enhanced SD OCT images. All of the 24 RNFL defects identified
on red-free fundus photographs that had angular widths of 30° or
less did not show disrupted cpRNFL reflectivity on enhanced SD
OCT images (P 0.002), whereas 23 (56.1%) of the 41 RNFL
defects with angular widths greater than 30° showed disruption of
the cpRNFL reflectivity on enhanced SD OCT images, whereas the
other 18 (43.9%) RNFL defects had no disruption.
Sensitivity and Specificity for Glaucoma Diagnosis
as More Than 1 Sector Abnormality in
Circumpapillary Retinal Nerve Fiber Layer
Analysis
Table 4 shows the sensitivity and specificity of the 3 types of OCT
for detecting glaucoma as having abnormal thinning (less than the
lower 99% CI value in normal eyes) or borderline thinning (less
than the lower 95% CI) of the cpRNFL in at least 1 sector on
cpRNFL thickness maps. In the normal control eyes, false RNFL
defects as borderline thinning (6 eyes) and abnormal thinning (5
eyes) of cpRNFL at least in 1 sector were detected by Stratus OCT
in 6 eyes and 3 eyes, respectively, by RTVue-100 and in 1 eye and
no eyes, respectively, by Spectralis. When abnormal thinning was
the criterion, enhanced SD OCT (Spectralis) was more sensitive
and specific than single-scan TD OCT (Stratus OCT; P 0.006
and P 0.001) or single-scan SD OCT (RTVue-100; P 0.001 and
P 0.004). Single-scan SD OCT was more specific than singlescan
TD OCT (P 0.006).
Discussion
The current study showed that reducing speckle noise on SD
OCT images by using eye tracking and averaging multiple
B-scans (enhanced SD OCT) made the boundary between
the RNFL and the ganglion cell layer much clearer on
circumpapillary scans in eyes with glaucoma, particularly at
locations corresponding to those of RNFL defects on fundus
photographs. Consequently, the enhanced SD OCT images
clearly visualized the severity of cpRNFL thinning, allow-
Table 2. Detection of Photographic Retinal Nerve Fiber Layer Defects by 3 Types of Optical Coherence Tomography as Abnormal
or Borderline Thinning of Circumpapillary Retinal Nerve Fiber Layer in the Optical Coherence Tomography Sector That
Corresponds with the Location of the Retinal Nerve Fiber Layer Defect on Red-Free Fundus Photography by Retinal Nerve Fiber
Layer Reflectivity Disruption in Eyes with Glaucoma (44 Eyes of 44 Patients)
Type of Optical Coherence
Tomography
Confidential Interval Limit
Levels of Normal Eyes All (n 65)
Retinal Nerve Fiber Layer Reflectivity on
Enhanced Spectral-Domain Optical Coherence
Tomography
Not Disrupted (n 42) Disrupted (n 23)
Time-domain (Stratus OCT; Abnormal thinning (P0.01)* 43 (66.2%) 22 (52.4%) 21 (91.3%)
single scan)
Borderline thinning (P0.05) †
54 (83.1%) 31 (73.8%) 23 (100%)
Spectral-domain (RTVue-100; Abnormal thinning (P0.01)
single scan)
‡
44 (67.7%) 22 (52.4%) 22 (95.7%)
Borderline thinning (P0.05) §
60 (92.3%) 37 (88.1%) 23 (100%)
Enhanced spectral-domain
Abnormal thinning (P0.01)
(Spectralis)
60 (92.3%) 37 (88.1%) 23 (100%)
Borderline thinning (P0.05)
62 (95.4%) 39 (92.9%) 23 (100%)
Stratus OCT vs. RTVue-100 (at the 0.01 level) #
P 1.000 P 1.000 P 0.804
Stratus OCT vs. Spectralis (at the 0.01 level) #
P0.0001 P0.0001 P 0.752
RTVue-100 vs. Spectralis (at the 0.01 level) #
P0.0001 P0.0001 P 0.634
OCT optical coherence tomography.
*P 0.002.
† P 0.006.
‡ P 0.0002.
§ P 0.152.
P 0.152
P 0.547 (Fisher exact test).
# MacNemar test.
1044
Ophthalmology Volume 118, Number 6, June 2011

Table 3. Detection of Photographic Retinal Nerve Fiber Layer Defects by 3 Types of Optical Coherence Tomography as Abnormal
Thinning of Circumpapillary Retinal Nerve Fiber Layer by Angular Width of Retinal Nerve Fiber Layer Defect on Red-Free Fundus
Photography in Eyes with Glaucoma (44 eyes of 44 patients)
Angular Width of
Retinal Nerve Fiber
Layer Defect (°)
No. (%) of Retinal Nerve Fiber Layer Defects on Red-Free Fundus Photography Detected as Abnormal Circumpapillary
Retinal Nerve Fiber Layer Thinning (P

localized RNFL defects. In this study, having no disruption
of RNFL reflectivity was significantly associated
with the smaller angular width of the photographic RNFL
defects and was not found in localized RNFL defects
with angular width of 30° or less. Single-scan TD OCT
(Stratus) and SD OCT (RTVue-100) failed to detect more
photographic RNFL defects because the angular width of
the photographic RNFL defects were smaller, consistent
with the previous studies. 16,18 In addition, enhanced SD
OCT (Spectralis) showed high sensitivity for detecting
photographic localized RNFL defects with angular
widths of more than 10°, but failed to detect nearly half
of the RNFL defects with angular widths smaller than
10°. Thus, it seems that the smaller angular width of
RNFL defects together with no disruption of cpRNFL
reflectivity, both of which appeared completely inseparable,
make the RNFL defects on red-free fundus photography
less likely to be detected as abnormal cpRNFL
thinning on OCT, leading to a relatively high falsenegative
rate for detection of these defects by OCT.
Even the enhanced SD OCT imaging had poor sensitivity
(55.6%) for detecting photographic RNFL defects
with angular widths smaller than 10° as abnormal thinning
(outside normal limits) in sector analysis, although
automatic drawing of the outer boundary of the cpRNFL
appeared to delineate exactly the boundaries of thinned
RNFL in areas corresponding to the locations of RNFL
defects. This false-negative result in sector analysis may
be attributable to the much wider angular width of each
sector of Spectralis (45° or 95°) compared with the
narrow angular width (10°) of the RNFL defects. In
addition, thinning of the cpRNFL in these narrow RNFL
defects was abrupt and mild (not disrupted). These considerations
suggest the possibility that the thicker
cpRNFL thickness outside the narrow RNFL defects
within each sector compensated for the abnormal
cpRNFL thickness within the narrow RNFL defects, and
consequently the mean cpRNFL thickness within the
sector remained within normal limits. This compensation
may be enhanced when the RNFL defects cross the
border of 2 sectors. Thus, the inability of detecting photographic
RNFL defects of less than 10° seems to result
from the limitation of the cpRNFL sector analysis.
According to the definition of glaucoma detected as
identification of at least 1 sector that had abnormal thinning
(less than the lower 99% CI value for the mean in
normal eyes) on the cpRNFL thickness, the 3 OCT systems
varied in sensitivity and specificity (Table 4). Sensitivity
for abnormal cpRNFL thinning was similar for
the single-scan systems (RTVue-100 and Stratus OCT),
but significantly higher than with enhanced SD OCT
(Spectralis). Specificity for abnormal thinning was also
higher with Spectralis than either of the single-scan systems,
but it also was significantly higher for single-scan
SD OCT (RTVue-100) than for TD OCT (Stratus OCT;
Table 4). The main difference between these 2 singlescan
OCT systems in the analysis of cpRNFL thickness is
their axial resolution (10 m for Stratus OCT vs. 5 m
for RTVue-100). Thus, the sensitivity of OCT for detecting
abnormal cpRNFL thinning in locations with RNFL
1046
Ophthalmology Volume 118, Number 6, June 2011
defects on red-free fundus photographs was improved
only by speckle noise reduction, whereas the specificity
of OCT for detecting abnormal cpRNFL thinning was
improved by both higher axial resolution and speckle
noise reduction.
A limitation of this study was that the possible influences
of instrumental differences on cpRNFL results
cannot be excluded, because different SD OCT instruments
were compared. For cpRNFL thickness analysis,
commercially available SD OCT instruments use circumpapillary
circle scans (e.g., RTVue-100, Spectralis, and
3D OCT-1000/2000 [Topcon, Tokyo, Japan]), or 3-D
raster scan (e.g., Cirrus HD-OCT [Carl Zeiss Meditec,
Dublin, CA] and 3D OCT-1000/2000). The differences in
scan protocol can affect the results. For example, as an
advantage of cpRNFL analysis based on 3D imaging, the
center of the circle for sampling can be determined
exactly after the scan is completed, which probably decreases
measurement variability resulting from the different
centers of the circles that can occur during image
acquisition in circle scans. 30 A disadvantage of 3D imaging
is that it requires much more time (a couple of
seconds) than circle scans, such that the results of 3D
scans are more susceptible than circle scans to influence
by involuntary ocular oscillations. This study used nearly
the same circumpapillary circle scans that Stratus OCT,
RTVue-100, and Spectralis use to eliminate the influence
of different scan protocols. However, there were still
possible influences from the differences among OCT
instruments in single B-scan quality, sampling density,
algorithms used for detection and marking of the boundaries
of cpRNFL, and patterns for dividing the cpRNFL
into sectors to measure mean regional cpRNFL thicknesses.
31 Prospective studies will be required to compare
single-scan images and multiple B-scan averaged images
using the same SD OCT instrument.
High-speed imaging of SD OCT technologies enables
speckle-noise reduction and 3-D imaging. However, the
imaging speed of current SD OCT instruments is too slow
to allow speckle-noise reduction with a 3-D scan protocol;
it takes at least approximately 1.4 to 2.4 seconds to
perform a 3-D raster scan, and when 16 B-scans are
averaged, as in this study, it takes theoretically at least
22.4 to 38.4 seconds to reduce speckle noise in a 3-D
raster scan. Although both speckle-noise reduction and
3-D raster scans seem to provide an advantage in detecting
RNFL defects, a choice must be made between 3-D
imaging and speckle-noise reduction, and it is not clear
which one offers better detection of RNFL defects.
In summary, the authors firmly suggest that the severity
of cpRNFL thinning identified as disrupted versus not disrupted
RNFL reflectivity taken together with angular width
of photographic RNFL defects are responsible for the falsenegative
detection by single-scan cpRNFL systems in OCT.
They also suggest that speckle-noise reduction can improve
the detection of photographic RNFL defects that have a
small angular width on red-free photography, no RNFL
disruption on OCT images, or both.

References
Nukada et al RNFL Defect Detection Using OCT in Glaucoma
1. Hoyt WF, Newman NM. The earliest observable defect in
glaucoma? Lancet 1972;1:692–3.
2. Sommer A, Miller NR, Pollack I, et al. The nerve fiber layer
in the diagnosis of glaucoma. Arch Ophthalmol 1977;95:
2149–56.
3. Sommer A, Pollack I, Maumenee AE. Optic disc parameters
and onset of glaucomatous field loss. I. Methods and progressive
changes in disc morphology. Arch Ophthalmol 1979;97:
1444–8.
4. Funk J. Early detection of glaucoma by longitudinal monitoring
of the optic disc structure. Graefes Arch Clin Exp Ophthalmol
1991;229:57–61.
5. Sommer A, Katz J, Quigley HA, et al. Clinically detectable
nerve fiber atrophy precedes the onset of glaucomatous field
loss. Arch Ophthalmol 1991;109:77–83.
6. Motolko M, Drance SM. Features of the optic disc in preglaucomatous
eyes. Arch Ophthalmol 1981;99:1992–4.
7. Tuulonen A, Airaksinen PJ. Initial glaucomatous optic disk
and retinal nerve fiber layer abnormalities and their progression.
Am J Ophthalmol 1991;111:485–90.
8. Tuulonen A, Lehtola J, Airaksinen PJ. Nerve fiber layer defects
with normal visual fields: do normal optic disc and
normal visual field indicate absence of glaucomatous abnormality?
Ophthalmology 1993;100:587–97.
9. Kass MA, Heuer DK, Higginbotham EJ, et al, Ocular Hypertension
Treatment Study Group. The Ocular Hypertension
Treatment Study: a randomized trial determines that topical
ocular hypotensive medication delays or prevents the onset of
primary open-angle glaucoma. Arch Ophthalmol 2002;120:
701–13.
10. Quigley HA, Miller NR, George T. Clinical evaluation of
nerve fiber layer atrophy as an indicator of glaucomatous optic
nerve damage. Arch Ophthalmol 1980;98:1564–71.
11. Guedes V, Schuman JS, Hertzmark E, et al. Optical coherence
tomography measurement of macular and nerve fiber layer
thickness in normal and glaucomatous human eyes. Ophthalmology
2003;110:177–89.
12. Wollstein G, Schuman JS, Price LL, et al. Optical coherence
tomography (OCT) macular and peripapillary retinal nerve
fiber layer measurements and automated visual fields. Am J
Ophthalmol 2004;138:218–25.
13. Wollstein G, Ishikawa H, Wang J, et al. Comparison of
three optical coherence tomography scanning areas for detection
of glaucomatous damage. Am J Ophthalmol 2005;
139:39–43.
14. Medeiros FA, Zangwill LM, Bowd C, et al. Evaluation of
retinal nerve fiber layer, optic nerve head, and macular
thickness measurements for glaucoma detection using optical
coherence tomography. Am J Ophthalmol 2005;
139:44–55.
15. Ojima T, Tanabe T, Hangai M, et al. Measurement of retinal
nerve fiber layer thickness and macular volume for glaucoma
detection using optical coherence tomography. Jpn J Ophthalmol
2007;51:197–203.
16. Jeoung JW, Park KH, Kim TW, et al. Diagnostic ability of
optical coherence tomography with a normative database to
detect localized retinal nerve fiber layer defects. Ophthalmology
2005;112:2157–63.
17. Kim TW, Park UC, Park KH, Kim DM. Ability of Stratus
OCT to identify localized retinal nerve fiber layer defects in
patients with normal standard automated perimetry results.
Invest Ophthalmol Vis Sci 2007;48:1635–41.
18. Jeoung JW, Park KH. Comparison of Cirrus OCT and Stratus
OCT on the ability to detect localized retinal nerve fiber layer
defects in preperimetric glaucoma. Invest Ophthalmol Vis Sci
2010;51:938–45.
19. Wu Z, Vazeen M, Varma R, et al. Factors associated with
variability in retinal nerve fiber layer thickness measurements
obtained by optical coherence tomography. Ophthalmology
2007;114:1505–12.
20. Wu Z, Huang J, Dustin L, Sadda SR. Signal strength is an
important determinant of accuracy of nerve fiber layer thickness
measurement by optical coherence tomography. J Glaucoma
2009;18:213–6.
21. Hangai M, Yamamoto M, Sakamoto A, Yoshimura N. Ultrahighresolution
versus speckle noise-reduction in spectral-domain
optical coherence tomography. Opt Express 2009;17:
4221–35.
22. Wollstein G, Paunescu LA, Ko TH, et al. Ultrahigh-resolution
optical coherence tomography in glaucoma. Ophthalmology
2005;112:229–37.
23. Mumcuoglu T, Wollstein G, Wojtkowski M, et al. Improved
visualization of glaucomatous retinal damage using highspeed
ultrahigh-resolution optical coherence tomography.
Ophthalmology 2008;115:782–9.
24. Schmitt JM, Xiang SH, Yung KM. Speckle in optical coherence
tomography. J Biomed Opt 1999;4:95–105.
25. Sander B, Larsen M, Thrane L, et al. Enhanced optical coherence
tomography imaging by multiple scan averaging. Br J
Ophthalmol 2005;89:207–12.
26. Sakamoto A, Hangai M, Yoshimura N. Spectral-domain optical
coherence tomography with multiple B-scan averaging
for enhanced imaging of retinal diseases. Ophthalmology
2008;115:1071–8.
27. Byeon SH, Chu YK, Lee H, et al. Foveal ganglion cell layer
damage in ischemic diabetic maculopathy: correlation of optical
coherence tomographic and anatomic changes. Ophthalmology
2009;116:1949–59.
28. Ooto S, Hangai M, Sakamoto A, et al. Three-dimensional
profile of macular retinal thickness in normal Japanese eyes.
Invest Ophthalmol Vis Sci 2010;51:465–73.
29. Pieroth L, Schuman JS, Hertzmark E, et al. Evaluation of focal
defects of the nerve fiber layer using optical coherence tomography.
Ophthalmology 1999;106:570–9.
30. Mwanza JC, Chang RT, Budenz DL, et al. Reproducibility of
peripapillary retinal nerve fiber layer thickness and optic nerve
head parameters measured with Cirrus HD-OCT in glaucomatous
eyes. Invest Ophthalmol Vis Sci 2010;51:5724–30.
31. Vizzeri G, Balasubramanian M, Bowd C, et al. Spectral
domain-optical coherence tomography to detect localized retinal
nerve fiber layer defects in glaucomatous eyes. Opt Express
2009;17:4004–18.
1047

Footnotes and Financial Disclosures
Originally received: April 10, 2010.
Final revision: September 2, 2010.
Accepted: October 14, 2010.
Available online: April 26, 2011. Manuscript no. 2010-533.
Department of Ophthalmology and Visual Sciences, Kyoto University
Graduate School of Medicine, Kyoto, Japan.
Financial Disclosure(s):
The author(s) have made the following disclosure(s):
Masanori Hangai - Consultant - NIDEK Co., Ltd., Topcon Corporation.
1048
Ophthalmology Volume 118, Number 6, June 2011
Nagahisa Yoshimura - Consultant - NIDEK Co., Ltd., and Topcon
Corporation.
Supported in part by a Grant-in-Aid for Scientific Research (no.:
20592038) from the Japan Society for the Promotion of Science (JSPS),
Tokyo, Japan.
Correspondence:
Masanori Hangai, MD, Department of Ophthalmology and Visual Sciences,
Kyoto University Graduate School of Medicine, 54 Kawahara-cho,
Shogoin, Sakyo-ku, Kyoto 606–8507, Japan. E-mail: Hyperlinkhangai@
kuhp.kyoto-u.ac.jp.