OPTICAL COHERENCE TOMOGRAPHY

ANGIOGRAPHY OF RETINAL VENOUS

OCCLUSION
AMIR H. KASHANI, MD, PHD,* SUN YOUNG LEE, MD,* ANDREW MOSHFEGHI, MD,*
MARY K. DURBIN, PHD,† CARMEN A. PULIAFITO, MD, MBA*

Purpose: To noninvasively evaluate the retinal microvasculature in human subjects with

retinal venous occlusions using optical coherence tomography angiography and assess
potential clinical applications.
Methods: This was a prospective, observational study of adult human subjects with
clinical and imaging findings demonstrating retinal venous occlusion. Subjects underwent
complete ophthalmic examination and fluorescein angiography as appropriate for their
standard of care. Optical coherence tomography angiography was performed on a pro-
totype spectral domain-OCTA system in 3 mm · 3 mm and 6 mm · 6 mm regions centered
on the fovea and parafoveal areas. Retinal vasculature was assessed within three horizontal
slabs consisting of the superficial, middle, and deep retina. The vasculature within each
slab was reconstructed using intensity contrast-based algorithms and visualized as en-face
images. Optical coherence tomography angiograms were manually segmented to verify the
accuracy of the automated segmentation algorithms.
Results: Optical coherence tomography angiography was able to demonstrate almost all
of the clinically relevant findings in 25 subjects with acute and chronic retinal venous
occlusion. These findings were consistent with clinical, anatomic, and fluorescein
angiographic findings including areas of impaired vascular perfusion, retinal atrophy,
vascular dilation, shunt vessels, and some forms of intraretinal edema.
Conclusion: Optical coherence tomography angiography is an investigational method
that generates high-resolution, noninvasive angiograms that qualitatively illustrate most of
clinically relevant findings in retinal venous occlusion. Optical coherence tomography
angiography corresponds well with fluorescein angiograms and in many cases provides
more detailed anatomic and blood flow information. Optical coherence tomography
angiography, in conjunction with standard spectral domain-OCT, is at least equally as
effective as fluorescein angiography for evaluation and management of the macular
complications of retinal venous occlusions.
RETINA 35:2323–2331, 2015

O ptical coherence tomography angiography (OCTA)

is a novel method that measures depth resolved
blood flow in the retina using the variance in the
intensity and phase of the reflected OCT beam.1–4
The initial application of OCTA in Macular Telangi-
ectasia4 and in pilot studies of patients with macular
degeneration,3 glaucoma,2 and diabetes5 has been
promising. Retinal venous occlusions (RVO) are
a major cause of vision loss and impairment through-
out the world affecting at least 16 million people
worldwide.6 The commonly accepted pathophysiol-
ogy of central retinal venous occlusion and branch
retinal venous occlusion (BRVO) is thrombosis of
the central retinal vein and branch retinal vein,
respectively, leading to variably impaired capillary
perfusion and retinal ischemia.7,8 Optical coherence
tomography angiography can provide a simple and
noninvasive way to assess the extent of impaired
capillary perfusion in retinal vascular occlusion that
is currently not possible without injection of contrast
dye. Optical coherence tomography angiography of
retinal vein occlusion in a single case report shows
good correlation between OCTA images and “capil-
lary nonperfusion” on fluorescein angiograpy.9 In this
study, we report on the OCTA findings in 25 subjects
with various stages and types of RVO and demon-
strate the utility of OCTA in the assessment of the
major clinical and angiographic findings of RVO.

2323
2324 RETINA, THE JOURNAL OF RETINAL AND VITREOUS DISEASES
Methods
The study was approved by the institutional review
board of the University of Southern California and
a signed informed consent was obtained from each
subject before enrollment. All subjects underwent
standard clinical examination and testing as appropri-
ate for their clinical disease including assessment of
visual acuity, intraocular pressure, slit-lamp examina-
tion, indirect ophthalmoscopy, spectral domain OCT,
as well as color fundus photography, and fluorescein
angiography, if indicated. In some cases, wide-field
fundus photography and fluorescein angiography were
performed if appropriate. Fluorescein angiography was
performed using either a Spectralis (Heidelberg Engi-
neering, Carlsbad, CA) or Optos 200Tx (Optos,
Dunfermline, Scotland) device. Optical coherence
tomography data were also acquired using commer-
cially available Cirrus SD-OCT 5000 (Carl Zeiss
Meditec Inc, Dublin, CA) and Heidelberg Spectralis
(Heidelberg Engineering Inc, Heidelberg, Germany) in
addition to the OCTA device described below. Wide-
field fundus photography was acquired using com-
mercially available Optos 200Tx or the wide-field
montaging feature of the Heidelberg Spectralis. All
images were analyzed using the commercially available
software that was available with the imaging device.
Patients with a known clinical diagnosis of RVO
were recruited for the study, whereas those with
significant ophthalmological comorbidities such as
advanced macular degeneration or severe nonprolifer-
ative diabetic retinopathy were excluded. Subjects
with mild forms of diabetic retinopathy, well-
controlled hypertension, or hyperlipidemia were not
excluded. Ancillary imaging procedures were per-
formed on the same day or within 1 month of the
OCT angiograms used for this study except where
indicated otherwise. Optical coherence tomography
angiograms were acquired using a Cirrus with standard
SD-OCT hardware configuration and modified acqui-
sition and analysis software for SD-OCTA imaging
(Carl Zeiss Meditec).
From the *USC Eye Institute and Department of Ophthalmology,
Keck School of Medicine, University of Southern California, Los
Angeles, California; and †Carl Zeiss Meditec Inc, Dublin, California.
USC Eye Institute and Department of Ophthalmology would
like to acknowledge Research to Prevent Blindness for an unre-
stricted departmental grant to support this research.
All authors from USC would like to disclose that Carl Zeiss
Meditec has provided research support in the form of prototype
OCT angiography devices to USC for this study. The remaining
authors have no conflicting interests to disclose.
Reprint requests: Amir H. Kashani, MD, PhD, USC Eye
Institute and Keck School of Medicine, 1450 San Pablo Street,
Suite 4700, Los Angeles, CA 90033; e-mail: ahkashan@usc.edu
 2015  VOLUME 35  NUMBER 11
The prototype Cirrus SD-OCTA system used the
standard 840 nm central wavelength with a scan speed
of 67,500 kHz. Scans were typically acquired in
a 3 mm · 3 mm and 6 mm · 6 mm pattern. All
3 mm · 3 mm scans contained 300 A-scans per line
and 300 B-scans per image. All 6 mm · 6 mm scans
contained 300 A-scans per line and 350 B-scans per
image. B-scans were repeated four times over each
section. Subjects were instructed to focus on a fixation
target and at least three scans centered on the fovea,
nasal macula, and far temporal macula were taken.
Retinal vasculature was assessed within three horizon-
tal retinal slabs of the OCTA consisting of the super-
ficial retina (internal limiting membrane, retinal nerve
fiber layer, ganglion cell layer, and superficial inner
plexiform layer), middle retina (deep inner plexiform
layer, inner nuclear layer, outer plexiform layer, and
superficial outer nuclear layer), and outer retina (deep
outer nuclear layer to the external limiting membrane).
The vasculature within each retinal slab was recon-
structed using an intensity contrast-based algorithm
and visualized as separate en-face images.
Results
We report the results of 26 eyes in 25 subjects with
RVO that were imaged using OCTA in addition to
conventional imaging as part of their standard of care.
These subjects demonstrate a broad spectrum of RVO
presentations as summarized in Table 1. Briefly, this
cohort included subjects with central retinal venous
occlusion, major and minor BRVO, and hemiretinal
venous occlusion. In addition, the cohort represents
RVO duration ranging from less than 1 month to
greater than 5 years. The demographics of this cohort
are also summarized in Table 1.
Figure 1 illustrates the normal OCTA findings from
the unaffected eye of one of the subjects to serve as
a reference image and to illustrate the segmentation
Table 1. Demographics of Cohort
Number of subject 25
Number of eyes 26
Male:female 14:11
Age (mean ± SD) 65 ± 11
Age range 35–84
Initial Snellen visual acuity (LogMAR) 0.38 ± 0.28
Last Snellen visual acuity (LogMAR) 0.38 ± 0.23
Prevalence of dyslipidemia 33%
Prevalence of hypertension 63%
Prevalence of diabetes 29%
Duration of RVO
Less than 1 year 8
Greater than 1 year 17
 OCTA OF RETINAL VENOUS OCCLUSION  KASHANI ET AL
scheme used in this study. This segmentation scheme
has been described before1 and correlates with the
anatomic distribution of retinal vessels described in
histological studies.10 Figure 1D illustrates the bound-
aries of the superficial retinal slab which extend from
the internal limiting membrane to the superficial inner
plexiform layer. This slab contains the capillary net-
works feeding the nerve fiber layer, ganglion cell
layer, and boundary of the inner nuclear layer and
inner plexiform layer. Figure 1E illustrates the bound-
aries of the mid-retinal slab which extends from the
deep inner plexiform layer to the superficial outer
nuclear layer and contains the capillary network
located at the boundary of the inner nuclear layer
and outer plexiform layer. We do not show the deep
or outer retinal slab (extending from the deep outer
nuclear layer to the external limiting membrane) since
no blood flow was detected in that section in any of
our subjects as expected. We have previously shown
that using this segmentation scheme, we can approxi-
mate the published capillary density measurements of
the normal, healthy retina based on histological meas-
urements.1 Therefore, we continue to use this segmen-
tation scheme for consistency.
Optical coherence tomography angiography consis-
tently demonstrated almost all of the clinical and
fluorescein angiographic findings that are character-
istics of acute and chronic RVO. Figure 2 illustrates
representative findings of a subject with an acute (,1
month) twig or minor BRVO involving the inferior
nasal macula and fovea. Figure 2A shows the clinical
findings of resolving intraretinal hemorrhage, very
mild macular edema, venous dilation, and early
venous-venous shunting across the horizontal merid-
nuclear layer, outer plexiform layer, and superficial outer nuclear layer. Outer retinal slab is not shown but is defined by deep outer nuclear layer to the
external limiting membrane.
2325
ian. The attenuation of the OCTA signal in the region
directly involving the intraretinal hemorrhage is likely
confounded by the decreased OCT reflectivity through
the hemorrhage. This attenuation of the raw OCT sig-
nal is evident on the en-face intensity image (Figure
2M). Evaluation of the OCTA image and associated
B-scans (Figure 2, I–K) suggests that the intraretinal
hemorrhage is primarily in the mid-retinal slab since
OCTA signal from the superficial slab is largely pre-
served (Figure 2, B–D). Therefore, this case demon-
strates mild impairment of blood flow in the superficial
slab overlying the hemorrhage, but we cannot assess
the state of perfusion in the mid-retinal layer which
contains the hemorrhage. This anatomic location of the
intraretinal hemorrhage is consistent with histological
data from postmortem cadaver studies that show the
location of mild intraretinal hemorrhages in RVO is
commonly in the outer plexiform layer.11 Therefore,
we confirm previous reports3 that intraretinal hemor-
rhage obscures the OCTA assessment of blood flow.
We also demonstrate that retinal hemorrhage only ob-
scures OCTA signal from retinal layers involving the
hemorrhage and beneath the area of hemorrhage. This
particular case also illustrates that the OCTA segmen-
tation scheme that we are using grossly corresponds to
the distribution of superficial and deep capillary net-
works of the retina since the superficial capillary flow
signal is preserved in the area overlying the
hemorrhage.
Despite the intraretinal hemorrhage, the OCTA
findings in Figure 2 are largely consistent with the
fluorescein angiogram and even more useful than tra-
ditional angiography in some respects. For example, in
areas without intraretinal hemorrhage, OCTA shows
Fig. 1. Optical coherence
tomography angiogram of a 73-
year-old African American
female with normal macula. A.
Depth encoded, pseudocolored
OCT angiogram with 6 mm
field-of-view and (B) 3 mm · 3
mm field-of-view. In both pan-
els, red color represents super-
ficial retinal vasculature and
green represents mid-retinal
vasculature. Yellow color rep-
resents the overlay of the red
and green layers. C. B-scan
with total retinal thickness seg-
mentation lines. D. Superficial
retinal layer is defined by inter-
nal limiting membrane, retinal
nerve fiber layer, ganglion cell
layer, and superficial inner
plexiform layer. E. Middle reti-
nal layer is defined by deep
inner plexiform layer, inner
2326 RETINA, THE JOURNAL OF RETINAL AND VITREOUS DISEASES  2015  VOLUME 35  NUMBER 11

Fig. 2. Optical coherence tomography angiogram of a 61-year-old woman with an acute (,1 month) twig retinal vein occlusion with resolving in-
traretinal hemorrhage and 20/20 vision. A. Color fundus photograph shows intraretinal hemorrhage in the distribution of a small retinal venule after it
passes underneath a large artery in the classic configuration of a BRVO. B. Depth encoded, pseudocolored map of the superficial and mid-retinal
vasculature. Red = superficial slab. Green = middle retinal slab. Yellow = red/green overlay. Dotted blue lines correspond to B-scan sections below. C.
Raw OCTA intensity image of superficial slab demonstrates very mild attenuation of the OCTA signal in the vascular distribution of a single retinal vein
in the inferior nasal quadrant of the macula extending from the inferior arcade to the fovea. D. Raw OCTA intensity image of mid-retinal slab
demonstrates more extensive OCTA attenuation in the inferior nasal quadrant of the macula in the distribution of a single retinal vein. Note the initial
segment of the small vein is visible in the lower right corner of the image and quickly attenuates as it passes under the larger overlying artery in the
classic configuration of a BRVO. Some of the attenuated OCTA signal in this slab may be due to intraretinal hemorrhage although even regions
with minimal intraretinal hemorrhage show attenuated OCTA signal intensity suggesting attenuation of flow as well. E. Mid-late phase fluorescein
angiogram shows blocking in area of intraretinal hemorrhage. F–K. Segmentation of corresponding retinal layers shown in the B-scan at the level of the
fine dashed blue lines in above OCTA images. Top dashed blue line in panels B–D corresponds to B-scans in panels F–H. Bottom dashed blue line in
panels B–D corresponds to B-scans in panels I–K. Note the mild retinal thickening in sections I–K suggests the location of the intraretinal hemorrhage
is in the mid-retinal layer corresponding to the section with the most attenuation of OCTA signal above. L. 3 · 3 image with high magnification of area
in black dotted square in panel B shows vascular dilation and shunting (blue arrow) across the horizontal meridian in the region of the RVO compared
with the rest of the perifoveal vascular ring. M. En-face intensity image of the same 6 mm · 6 mm field-of-view as OCTA images shows acceptable
distribution of OCT reflectivity throughout the image except in the location of the intraretinal hemorrhage, and one area of signal artifact (blue arrow)
that would contribute to artifactual changes in OCTA.

there is vascular dilation of the perifoveal capillaries
that are downstream of the occlusion site (Figure 2L
blue arrow). This level of capillary resolution is hard to
reliably achieve with standard fluorescein angiography.
In addition, OCTA demonstrates mild vascular shunting
across the horizontal meridian that is also confirmed in
the fluorescein angiogram (Figure 2, E and L). Optical
coherence tomography angiography findings in other
subjects with acute RVO and intraretinal hemorrhage
were similar. This case also illustrates that artifacts in
the raw OCT signal intensity (Figure 2M middle right
of panel; blue arrow) can simulate decreased perfusion
(Figure 2B middle right of panel; blue arrow). There-
fore, it is necessary to carefully evaluate raw OCT
reflectivity in regions of attenuated OCTA findings to
rule out artifactual changes.
Figure 3 illustrates diffusely impaired capillary
perfusion in a subject with chronic (.1 year) BRVO
 OCTA OF RETINAL VENOUS OCCLUSION  KASHANI ET AL 2327

Fig. 3. Optical coherence tomography angiogram of a 73-year-old man with chronic (1 year) retinal vein occlusion involving the superior and nasal
macula and fovea. A. Depth encoded, pseudocolored map of the superficial and mid-retinal vasculature. Red = superficial vasculature. Green = middle
retinal vasculature. Yellow = red/green overlay. B. Optical coherence tomography angiography image of superficial layer demonstrates attenuation of
perifoveal blood flow in most of superior retina and some parts of inferior perifoveal vasculature. C. Optical coherence tomography angiography image
of mid-retinal vasculature demonstrates more extensive loss of blood flow in superior and inferior macula than images of superficial layer. D. Optical
coherence tomography angiography image of the deep retinal vasculature shows no retinal blood flow as expected. E–H. Segmentation of corre-
sponding retinal layers shown in the B-scan at the level of the fine blue line in above images. I. Color fundus photograph of posterior pole demonstrates
ischemic appearing macula with dot-blot hemorrhages and tortuous retinal vessels. J. En-face intensity image of the same 3 mm · 3 mm field-of-view as
OCTA images shows an acceptable distribution of OCT reflectivity throughout the image without any artifacts that would contribute to artifactual
changes in OCTA.

involving the superior macula and fovea. In this case,
the en-face OCT intensity image did not demonstrate
any irregular fluctuations in OCT reflectivity that
would artifactually create regions of low OCTA sig-
nal (Figure 3J). A fluorescein angiogram was not
available in this case due to infiltration at the injec-
tion site. The OCTA demonstrates a large area of
impaired capillary perfusion extending from the supe-
rior arcade and involving the perifoveal ring. The
depth encoded, pseudocolored image demonstrates
impaired blood flow in both the superficial and
mid-retinal slabs. In all of our subjects, we found that
areas of attenuated OCTA signal were almost always
qualitatively larger in the mid-retinal slab than corre-
sponding superficial retinal slabs. Evaluation of the
B-scan in these regions did not demonstrate any nota-
ble loss in the OCT reflectivity with increasing depth
through the retina. Therefore, it is unlikely that this
observation is an artifact of decreased OCT reflectiv-
ity through retinal tissue. As in Figure 3D, in no case
could we identify blood flow signal in the deep retinal
slab as expected.
Figure 4 illustrates a chronic hemiretinal venous
occlusion with intraretinal edema and diffusely
impaired capillary perfusion involving the whole supe-
rior macula and fovea. As in the previous cases, the
attenuation of the OCTA signal from the mid-retinal
slab was more prominent than the superficial slab.
There was no evidence of decreased reflectivity in
the corresponding B-scans (Figure 4, F–N) that sug-
gests this was due to artifactual loss of signal. There-
fore, we believe, this represents true layer specific
impairment of flow in the mid-retinal layer.
Figure 4 also illustrates 2 notable observations
regarding intraretinal edema and OCTA. First, the B-
scans in Figure 4, I–K illustrate typical hyporeflective
2328 RETINA, THE JOURNAL OF RETINAL AND VITREOUS DISEASES  2015  VOLUME 35  NUMBER 11

Fig. 4. Optical coherence tomography angiography of a 50-year-old Hispanic female with chronic (.1 year) hemiretinal vein occlusion and diffuse
areas of capillary nonperfusion. A. Mid-late phase fluorescein angiogram of the macula demonstrates diffuse loss of capillary perfusion in the superior
macula with focal areas of staining consistent with intraretinal fluid. Dotted blue lines correspond to B-scans shown below. B. Depth encoded,
pseudocolored map of the superficial and mid-retinal vasculature. Red = superficial vasculature. Green = middle retinal vasculature. Yellow = red/green
overlay. C. Optical coherence tomography angiography of superficial layer demonstrates lack of capillary blood flow in most of superior retina with
preservation of most of the inferior macular blood flow. Mainly, the larger and superficial vessels demonstrate blood flow in the superior macula. D.
Optical coherence tomography angiography of mid-retinal vasculature demonstrates more extensive loss of blood flow in superior macula and relative
preservation of inferior macular blood flow. E. En-face intensity image of the same 3 mm · 3 mm field-of-view as OCTA shows normal and even
distribution of OCT reflectivity throughout the image without any artifacts that would contribute to artifactual changes in OCTA. F–N. Segmentation of
corresponding retinal layers shown on the B-scan at the level of the fine dashed blue lines in above OCTA. Top dashed blue line in panels B–D
corresponds to B-scans in panels F–H. Middle dashed blue line in panels B–D corresponds to B-scans in panels I–K. Bottom dashed blue line in panels
B–D corresponds to B-scans in panels L–N. Note that the area of retinal thickening in the OCTA corresponds to panels F–H. This area seems to
correlate with the diffuse OCTA signal in the mid-retinal layer of the superior macula that is not the same consistency as retinal microvasculature. This
area of retina demonstrates intraretinal fluid in the mid-retinal layer (H) that is relatively hyperreflective on OCT and seems to generate some OCTA
signal, whereas the hyporeflective pockets of intraretinal fluid through the fovea (I–K) do not demonstrate OCTA signal. The bright hyperreflective dots
likely represent early hard exudates. Therefore, these areas are referred to as intraretinal fluid associated with hard exudate.

pockets within the retina that are consistent with macu-
lar edema. These hyporeflective pockets of edema are
not visible on OCTA images. In all cases where intra-
retinal hyporeflective cystoid spaces were found, there
was also evidence of impaired perfusion in one or both
retinal slabs as in Figure 4, B–D and I–K. This has not
been demonstrated previously using fluorescein angiog-
raphy. In contrast, the hyperreflective pockets of edema
in Figure 4, F–H do appear on OCTA images. Hyper-
reflective intraretinal pockets of edema have been pre-
viously described on SD-OCT alone12 but not in OCTA.
In Figure 4, these pockets of hyperreflective edema cor-
respond to diffuse and splotchy OCTA signal in the
superior macula (Figure 4, B–D top blue dashed
line and corresponding B-scans in Figure 4, F–H). We
have observed similar cases of increased OCTA signal
intensity corresponding to hyperreflective pockets of
intraretinal edema in a few other subjects. These hyper-
reflective pockets of edema correlate with the location of
hard exudates on color fundus photography and are
usually near hyperreflective pin-point spots consistent
with hard exudate on B-scans as well. Therefore, we
refer to these areas as intraretinal fluid associated with
hard exudate. This finding suggests that some forms of
intraretinal edema may be visualized on OCTA images.
The clinical utility, if any, of this finding is not clear. We
hypothesize that the OCTA signal from these hyperre-
flective pockets of edema originates from Brownian
motion of particulate material that later consolidates into
hard exudate.
 OCTA OF RETINAL VENOUS OCCLUSION  KASHANI ET AL 2329

Figure 5 illustrates the findings from a subject
with a chronic BRVO that was followed in our clinic
from the time of the original BRVO onset. The color
fundus images in this figure show the original distri-
bution of the intraretinal hemorrhages. These hemor-
rhages were not present during the OCTA scanning
which occurred 2 years after the inciting BRVO.
Therefore, the attenuated OCTA signal in this case
is not caused by intraretinal hemorrhage. This patient
received regular antivascular endothelial growth fac-
tor injections for treatment of the macular edema from
the time of diagnosis. Optical coherence tomography
angiography images demonstrate patchy areas of
impaired capillary perfusion that are slightly more
prominent in the mid-retinal slab than the superficial
retinal slab as in our previous cases (Figure 5B).
B-scans through the area of inferior macula demon-
strate persistent intraretinal fluid corresponding to the
area with attenuated OCTA signal and decreased
blood flow. The OCTA also shows vascular dilation
and shunting across the horizontal meridian that is
characteristic of chronic RVO (Figure 5L). The

Fig. 5. Optical coherence tomography angiography of a 66-year-old man with chronic (.1 year) branch retinal vein occlusion in the macula. A. Color
fundus photograph from the time of the initial vein occlusion approximately 2 years before OCTA was performed demonstrates diffuse intraretinal
hemorrhage in the distribution of the inferior temporal vein. These hemorrhages were resolved at the time of OCTA imaging. B. Depth encoded,
pseudocolored map of the superficial and mid-retinal vasculature. Red = superficial vasculature. Green = middle retinal vasculature. Yellow = red/green
overlay. C. Optical coherence tomography angiography of superficial layer demonstrates attenuation of the OCTA signal from the superficial layer in
a nonspecific vascular distribution of the superior temporal and inferior macula extending from the inferior arcade to the fovea. Dotted blue lines
correspond to B-scans shown below. D. Optical coherence tomography angiography of mid-retinal vasculature demonstrates more extensive loss of
capillary blood flow than the superficial retinal vasculature. E. Mid-late phase angiogram of the vein occlusion at the initial presentation 2 years before
OCTA imaging. F–K. Segmentation of corresponding retinal layers shown in the B-scan at the level of the fine dashed blue line in above OCTA. Top
blue line in panel C and D corresponds to B-scans in panels F–H. Bottom dotted blue lines in panels C and D correspond to B-scans in panels I–K. L. 3
· 3 image with high magnification of area in white dotted square in panel A shows mild vascular dilation and shunting (blue arrowheads) across the
horizontal meridian in the region of the RVO. M. En-face intensity image of the same 6 mm · 6 mm field-of-view as OCTA demonstrates normal
distribution of OCT reflectivity throughout the image without any artifacts that would contribute to artifactual changes in OCTA.
2330 RETINA, THE JOURNAL OF RETINAL AND VITREOUS DISEASES
en-face OCT intensity image demonstrates normal
distribution of OCT reflectivity throughout the image
without any artifacts that would contribute to artifac-
tual changes in OCTA intensity.
Discussion
We show that OCTA can qualitatively demonstrate
most of the major clinical and fluorescein angiographic
features of acute and chronic RVO including impaired
capillary perfusion, intraretinal hemorrhage, some
forms of macular edema, vascular dilation, and
venous-venous shunting. This suggests that OCTA, in
conjunction with standard SD-OCT, is at least equally
as effective as fluorescein angiography for evaluation
and management of the macular complications of
RVO. We also demonstrate 3 novel findings of retinal
venous occlusion on OCTA, namely (1) layer specific
impairment of blood flow, (2) detection of intraretinal
edema associated with hard exudate, and (3) impaired
blood flow in regions of intraretinal edema. Although
our segmentation scheme may not be the best for every
disease process, we have previously showed that we
can approximate the histological capillary density
measurements of the normal, healthy retina using this
segmentation scheme.1 In addition, our experience with
this segmentation scheme suggests that it accurately
corresponds to the known pathology of at least some
diseases, like RVO. For example, in Figure 2, the seg-
mentation scheme nicely isolates the attenuated OCTA
signal from the hemorrhage to the mid-retinal layers
and the relative preservation of blood flow in the over-
lying layer. The location of the attenuated OCTA signal
corresponds to the location of intraretinal hemorrhage
in the outer plexiform layer in pathologic studies of
RVO11 and the OCT B-scan findings in that subject.
The segmentation of the blood flow in the superficial
slab is largely preserved in that case.
Our study has a number of interesting and clinically
useful findings. For example, OCTA may be just as
useful in assessment of the macular changes of acute
RVO as fluorescein angiography but without the risk
of intravenous dye injection. Although intraretinal
hemorrhage does obscure the OCTA signal, just as it
does in fluorescein angiography, the microvascular
findings surrounding the region of RVO are more
easily visualized on OCTA than fluorescein angiogra-
phy. For example, the OCTA in Figure 2 shows pres-
ervation of the blood flow in the superficial slab
overlying the hemorrhage. This example and others
in our series suggest that loss of perfusion in the
mid-retinal layers may occur earlier than that in super-
ficial layers. However, the presence of hemorrhage in
 2015  VOLUME 35  NUMBER 11
this case confounds our ability to assess the blood flow
in the layer of the hemorrhage. Nevertheless, this sug-
gests that layer specific changes in retinal perfusion
can be detected with very high accuracy and with
much more precision than fluorescein angiography.
The ability to perform OCTA as frequently as
standard SD-OCT scans (i.e., monthly) will likely
provide a useful tool in assessing the correlation of
retinal edema and blood flow in RVO that has not been
possible previously. In our cohort, we observed that in
all cases where classic hyporeflective intraretinal edema
was present, there was also corresponding impaired
blood flow in that region (Figures 4 and 5). It is pos-
sible that this blood flow attenuation is somehow an
artifact of the OCTA signal from altered retinal anat-
omy rather than a true reflection of blood flow, but we
could find no evidence to support this possibility. It is
unlikely that there is significant signal attenuation over
100 microns of retinal tissue and examination of the
B-scans in those regions do not show notable signal
attenuation with increasing retinal depth. In addition,
we observe similar findings in diabetic subjects with
macular edema, suggesting that intraretinal edema may
be caused by compromised capillary blood flow in
other retinal vascular diseases (AHK; data submitted
for publication). It is important to note that decreased
blood flow as measured by OCTA can occur without
intraretinal edema as well, at least in this cohort.
We demonstrate that hyperreflective pockets of
intraretinal edema generate OCTA signal as in Figure
4. The presence of hyperreflective pockets of intrare-
tinal fluid on SD-OCT has been described before,12
but the clinical significance is not known. We show
that at least in 2 cases we have observed so far, the
presence of hyperreflective pockets of intraretinal
edema correlates exactly with the location of diffuse
OCTA signal from retinal tissue and not retinal ves-
sels. These areas are also always associated with hard
exudate on color photographs and SD-OCT. We
hypothesize that this finding on OCTA is due to the
Brownian motion of particulate material within the
pockets of intraretinal edema. In essence, these areas
are the intermediary step between resorption of intra-
retinal fluid and formation of hard exudate. The clin-
ical significance of these findings is not clear, and the
rarity with which they have been reported in the liter-
ature and in our case series suggests that this is not
a frequently occurring phenomenon or it is very under-
appreciated on current SD-OCT scans. Because hard
exudate formation in the macula is often a harbinger of
poor visual acuity, this finding may have prognostic
significance.
We observe that areas of impaired capillary perfu-
sion are more distinct on OCTA than corresponding
 OCTA OF RETINAL VENOUS OCCLUSION  KASHANI ET AL
fluorescein angiography images of capillary nonperfu-
sion. There are a few reasons for this observation.
First, the assessment of capillary perfusion on fluores-
cein angiography is only possible in a very short
transit window while fluorescein dye is predominantly
in the capillaries. This ideal transit interval is probably
between 10 to 15 seconds of a well-timed fluorescein
angiogram.13 Beyond this interval, fluorescein extrav-
asation into the tissue progressively decreases signal-
to-noise from the capillaries and prevents visualization
of the capillaries. In addition, it is possible that fluo-
rescein dye flows through (or leaks into) capillaries
that are not patent to passage of red blood cells.
Optical coherence tomography angiography has
a number of limitations which will likely be overcome
as the technology matures and scanning methods
improve. First, the field-of-view is relatively small
because of limitations in the scanning speed and image
tracking software. Therefore, OCTA is only useful for
assessment of macular disease. Fluorescein angiogra-
phy will still be required for assessment of peripheral
vascular disease. These limitations will be overcome
with advances in SS-OCTA and improvements in
image tracking algorithms and scan speeds. Neverthe-
less, at this point, it is not always possible to get
a complete OCTA assessment of the macula, espe-
cially in subjects with poor fixation. Second, OCTA
cannot detect intraretinal fluid on a consistent basis.
Our results, as well as unpublished observations,
suggest that OCTA can detect “chronic” appearing
intraretinal edema that is typically associated with hard
exudates and is relatively hyperreflective on SD-OCT
sections.12 These areas typically do not have any fluo-
rescein angiographic findings. However, in none of
our subjects were we able to detect typical, hypore-
flective intraretinal edema on OCTA images. The clin-
ical significance of these findings remains to be
determined. Because OCTA are based on and coregis-
tered with the underlying B-scans from the OCT, it
does not seem that this will be a significant problem
in detecting intraretinal edema. The most important
limitation of OCTA is that it cannot assess “leakage”
as in fluorescein angiography. Therefore, it is unlikely
that OCTA will completely replace fluorescein angi-
ography since the latter modality has both wide-field
capabilities and time-encoded information in the form
of a transit phase that allows assessment of leakage.
2331
Key words: angiography, optical coherence tomog-
raphy, retinal vein occlusion, retinal vascular disease.
Acknowledgments
The authors would like to thank Douglas Matsunaga
and Jack Yi for help with OCTA imaging and
preparation of tables and figures.
References
1. Matsunaga D, Yi J, Puliafito CA, Kashani AH. OCT angiog-
raphy in healthy human subjects. Ophthalmic Surg Lasers
Imaging Retina 2014;45:510–515.
2. Jia Y, Wei E, Wang X, et al. Optical coherence tomography
angiography of optic disc perfusion in glaucoma. Ophthalmol-
ogy 2014;121:1322–1332.
3. Moult E, Choi W, Waheed NK, et al. Ultrahigh-speed swept-
source OCT angiography in exudative AMD. Ophthalmic Surg
Lasers Imaging Retina 2014;45:496–505.
4. Thorell MR, Zhang Q, Huang Y, et al. Swept-source OCT
angiography of macular telangiectasia type 2. Ophthalmic Surg
Lasers Imaging Retina 2014;45:369–380.
5. Schwartz DM, Fingler J, Kim DY, et al. Phase-variance optical
coherence tomography: a technique for noninvasive angiogra-
phy. Ophthalmology 2014;121:180–187.
6. Rogers S, McIntosh RL, Cheung N, et al. The prevalence of
retinal vein occlusion: pooled data from population studies
from the United States, Europe, Asia, and Australia. Ophthal-
mology 2010;117:313–319 e311.
7. Frangieh GT, Green WR, Barraquer-Somers E, Finkelstein D.
Histopathologic study of nine branch retinal vein occlusions.
Arch Ophthalmol 1982;100:1132–1140.
8. Green WR, Chan CC, Hutchins GM, Terry JM. Central retinal
vein occlusion: a prospective histopathologic study of 29 eyes
in 28 cases. Retina 1981;1:27–55.
9. Kuehlewein L, An L, Durbin MK, Sadda SR. Imaging areas of
retinal nonperfusion in ischemic branch retinal vein occlusion
with swept-source OCT microangiography. Ophthalmic Surg
Lasers Imaging Retina 2015;46:249–252.
10. Tan PE, Yu PK, Balaratnasingam C, et al. Quantitative confo-
cal imaging of the retinal microvasculature in the human retina.
Invest Ophthalmol Vis Sci 2012;53:5728–5736.
11. Green WR, Chan CC, Hutchins GM, Terry JM. Central retinal
vein occlusion: a prospective histopathologic study of 29 eyes
in 28 cases. Trans Am Ophthalmol Soc 1981;79:371–422.
12. Liang MC, Vora RA, Duker JS, Reichel E. Solid-appearing retinal
cysts in diabetic macular edema: a novel optical coherence tomog-
raphy finding. Retin Cases Brief Rep 2013;7:255–258.
13. Johnson RN, Fu AD, McDonald HR, et al. Chapter 1:
fluorescein angiography: basic principles and interpretation.
In: Ryan SJ, Hinton DR, Schachat AP, et al, eds. Retina. 5th
ed. London, United Kingdom: W.B. Saunders; 2013:2–50.e51.