The Retinal Disease Screening Study: Retrospective Comparison of Nonmydriatic Fundus Photography and Three-Dimensional Optical Coherence Tomography for Detection of Retinal Irregularities

Abstract

Purpose.: To determine the sensitivity of three-dimensional optical coherence tomography (3D-OCT) versus single field nonmydriatic fundus photography (FP) for detection of a variety of retinal abnormalities.

Methods.: Images from consecutive patients in a retina clinic undergoing simultaneous 3D-OCT (512×128) and single, foveal nonmydriatic 45° color fundus imaging with 3D-OCT-1000 in a 4 month-period were retrospectively collected. Findings from each modality were graded independently by two graders as present, questionable, or absent. Irregularities were separated into three categories for intermodality comparisons: epiretinal, retinal/subretinal, and RPE/choroidal irregularities. The approximate location of findings in relation to the 3D-OCT field was noted as in field and out of field. Findings from both modalities were combined to form the gold standard for comparison for each modality.

Results.: Five hundred and one sets of 3D-OCT scans and fundus images of 395 eyes of 223 patients were found in the study period, of which, 474 unique visits were included. Ninety-six percent of the scans had abnormal findings. Twenty-six fundus images (5.5%) were ungradable. 3D-OCT identified some abnormality in 25/26 (96.2%) of the ungradable fundus images. For overall detection of a variety of retinal irregularities or irregularity of each category (epiretinal, intraretinal, or RPE/choroidal irregularity), 3D-OCT was found to be more sensitive than that of nonmydriatic color fundus images. When single specific feature was speculated, 3D-OCT demonstrated various detection abilities: higher than FP for abnormal retinal thickness (or intraretinal hyporeflective features); similar as FP for RPE atrophy; however, lower for pigment migration (or intraretinal hemorrhage).

Conclusions.: In this study, sensitivities of 3D-OCT were higher than nonmydriatic fundus images for overall detection of retinal abnormalities or irregularities in each category. 3D-OCT demonstrated good ability to detect most features; however, with limitation to intraretinal hemorrhage and pigment migration. It is likely that OCT will be added to photography screening for chorioretinal diseases in the near future.

Introduction

Retinal and choroidal pathology, especially diabetic retinopathy (DR) and age-related macular degeneration (AMD), is the leading cause of visual loss in people of working age and older in the Western world. ¹ Diabetes mellitus is a condition that affects 180 million people worldwide. As a result of population growth, ageing, obesity, and sedentary lifestyle, the total number of diabetic patients is expected to rise to an estimated 439 million by 2030, ² in which the major part of this numerical increase will occur in the developing countries. ¹ While for the developed world, like the United States, the number of patients with AMD was estimated to be 8 million among people over 55 years old. ³ Of these people, 1.3 million would develop advanced AMD if no treatment were given to reduce their risk. ⁴ Globally, the elderly population will double by 2040. ⁵ Thus, effective methods for detection of retinal and choroidal diseases are vitally important both for patient care and socioeconomic concern. 

Nonmydriatic fundus photography (FP) was one of the most cost-effective methods for retinal and choroidal disease detection. However, with undilated pupils, its limitation due to possible inadequate image quality may be a major concern. Optical coherence tomography (OCT), as the methodology with the same ease and safety for a community-based, or even worldwide, screening program as the nonmydriatic methodology, is possible a preference for screening purpose in asymptomatic population. ^6,7 In this report, we evaluated the sensitivities of nonmydriatic FP and three-dimensional (3D)-OCT in the detection of retinal lesions in a high-risk population. 

Methods

Data Collection

In this study, images from consecutive patients in a retina clinic undergoing simultaneous 3D-OCT (512×128) and single, foveal nonmydriatic 45° color fundus image with 3D-OCT-1000 (Topcon Corp., Tokyo, Japan) in a 4 month-period were retrospectively collected. Written informed consent was obtained from all subjects. Approval for data collection and analysis was obtained from the institutional review board of the University of Southern California. The research adhered to the tenets set forth in the Declaration of Helsinki. 

Information about age, sex, and ophthalmologic diagnosis was gathered. Raw FP and OCT data were exported from the imaging instruments for review at the reading center. 

Grading Methodology

Two graders (ACW and YO), certified for assessing color fundus images at the Doheny Image Reading Center (DIRC), evaluated each set of FP images for each eye independently. FP-based irregularities included: epiretinal membrane (ERM), macular hole (MH), macular edema, retinal atrophy, hard exudates, intraretinal hemorrhage, microaneurysms, intraretinal microvascular abnormalities (IRMA), cotton wool spots, pigment migration (defined as black, often spiculated, areas of pigment clumping within the macula), ⁸ subretinal fluid (SRF), cystoid edema, drusen, retinal pigment epithelium (RPE) atrophy, RPE detachment (PED), choroidal neovascularization (CNV)/subretinal fibrosis. Since peripapillary chorioretinal atrophy and pigmentary changes are relatively common clinical findings, ^9–11 neither of them were considered abnormal in our study; thus, were not evaluated. The presence or absence of the above features was graded as definite present (Y), questionable present (Q), absent (N), or cannot grade (CG). In order to compare findings with OCT images, areas of the fundus captured by OCT scans were defined as ‘‘in field''; regions not captured by OCT scanning were defined as ‘‘out field.'' The entirety of the fundus area captured, irrespective of OCT scanning, was defined as the ‘‘full field.'' ⁶ For each feature, both results for presence/absence and field were recorded. 

Two graders (YO and ACW), certified for assessing 3D-OCT images at DIRC, evaluated each set of 3D-OCT images for each eye independently. 3D-OCT scans were assessed using a previously described and validated spectral-domain OCT reading center software program (3D-OCTOR). ^6,7 OCT-based irregularities included ERM, MH, increased retinal thickness, decreased retinal thickness, intraretinal hyperreflective feature, intraretinal hyporeflective feature, photoreceptor inner segment/outer segment (IS-OS) irregularity, RPE irregularity, and RPE thinning/atrophy were recorded as Y, Q, N, or CG for each case. If intraretinal hyperreflective features were found, further assessment for its origin as RPE (intraretinal RPE migration), ^8,12 vascular, ¹³ or other were documented (categories could overlap for one case). All findings from 3D-OCT were regarded as in field in the study. 

The comparative relationship of observed features in FP and OCT was shown in Table 1. 

Table 1

 

Findings Assessed in Each Modality and Their Comparative Relationship

Epiretinal	Fundus Photography Findings	3D-OCT Findings
Epiretinal	ERM	ERM
	MH	MH
Retinal/subretinal	Aneurysms	Intraretinal hyperreflective features	Vascular origin
	Cotton wool spots	Intraretinal hyperreflective features	Vascular origin
	Exudate	Intraretinal hyperreflective features	Vascular origin
	Hemorrhage	Intraretinal hyperreflective features	Vascular origin
	Pigment migration	Intraretinal hyperreflective features	RPE origin
	Macular edema	Increased retinal thickness
	CME/SRF	Hyporeflective feature
	Retinal atrophy	Decreased retinal thickness
	IS-OS irregularity
RPE/choroidal	RPE atrophy	RPE thinning/atrophy
	Drusen	RPE irregularity
	CNV/subretinal fibrosis/PED	RPE irregularity

Statistical Analysis

Only gradable records were used for sensitivities and false negative (FN) analyses in the study. As published previously, partial credits was chosen to transform data with Q grades and for the purpose of calculating sensitivity and specificity, combined appearance of the abnormality was used as the ground truth. ^6,7,14 Commercially available software (SPSS, version 13; Statacorp LP, College Station, TX) was used for processing all statistical analysis. All P values were two-sided and were considered statistically significant when the values were less than 0.05. 

Results

Characteristics of the Study Population

Five hundred and one sets of OCT/FP images from 395 eyes of 223 patients were found for the study. Choosing one set of image from each unique visit only, 474 sets of images were evaluated in the study. 

Characteristics of the patients' demographics were shown in Table 2. Of all the eligible subjects, 49.3% (110/223) were women. The mean age of patients was 72.2 years, ranging from 17 to 96 years. The range of best-corrected visual acuities (BCVA) was 20/20 to hand motion with a mean visual acuity of 20/40. 

Table 2

 

Demographics of Patients in the Study

Patients	223
Eyes	395
Sex (men:women)	49.3%: 50.7% (110:113)
Mean age (range)	72.2 years (17–96 years)
Mean vision (range)	20/40 (20/20–Hand motion)
Distribution of eyes (%)
Macular disease
Age-related macular degeneration	241 (50.8)
Branch/central retinal vein occlusion	22 (4.6)
Diabetic retinopathy	77 (16.2)
Epiretinal membrane/vitreomacular traction	74 (15.6)
Myopia	33 (7.0)
Other retinal diseases*	27 (5.7)
Lens status
Phakic eyes	282 (59.5)
Aphakic/pseudophakic eyes	192 (40.5)

*  Including hereditary retinal diseases (e.g., best vitelliform macular dystrophy), chronic lymphocytic leukemia, hypertensive retinopathy, choroidal nevus, presumed ocular histoplasmosis syndrome, retinal detachment, and eyes within normal limit.

All OCT images included in the study met reading center criteria for sufficient image quality. FP images were ungradable in 5.5% (26/474). In total, 448 sets of gradable FP and OCTs were included in the sensitivities and specificities analyses. 

Detection of Irregularities in Eyes With Gradable OCT and FP Images

Grading results of each chorioretinal abnormality were shown in Table 3. Sensitivities of FP and 3D-OCT in field and in full field were shown in Table 4. 3D-OCT was more sensitive than FP for detection of epiretinal irregularity (Table 4). The examples of FN for 3D-OCT and FP were shown in Figure 1. 

Figure 1

 

Examples of false negative cases for epiretinal irregularity for nonmydriatic fundus photography and 3D-OCT. (A1, A2) Color images graded as absent for epiretinal changes and OCT B-scans of the same eye with definite ERM. (B1, B2) Color images graded as absent for epiretinal irregularity and OCT B-scans of the same eye with definite MH. (C, D) Color images graded as questionable ERM. (E, F) Color image graded as questionable MH.

Table 3. 

 

Grading Results for Chorioretinal Abnormalities Observed by Nonmydriatic FP and 3D-OCT

	Epiretinal Irregularity			Retinal/Subretinal Irregularity			RPE/Choroidal Irregularity
	FPI	FPF	OCT	FPI	FPF	OCT	FPI	FPF	OCT
Y	20	24	272	182	270	379	247	280	347
	4.5%	5.4%	60.7%	40.6%	60.3%	84.6%	55.1%	62.5%	77.5%
Q	61	64	94	69	94	25	112	99	38
	13.6%	14.3%	21.0%	15.4%	21.0%	5.6%	25.0%	22.1%	8.5%
N	367	360	82	197	84	44	89	69	63
	81.9%	80.4%	18.3%	44.0%	18.8%	9.8%	19.9%	15.4%	14.1%
Total	448	448	448	448	448	448	448	448	448

 FPI, fundus photography in field; FPF, fundus photography in full fields.

Table 4

 

Comparison of Sensitivities of Nonmydriatic FP to 3D-OCT if Using Findings From Both FP and OCT as Ground Truth

Abnormal Findings		Ground Truth		Fundus Photography			3D-OCT
		n, Y/Q	n, Y/Q	Sensitivities		Eyes With Irregularities Missed by FP n, Y/Q	Sensitivities		Eyes With Irregularities Missed by 3D-OCT n, Y/Q
		In Field	Full Field	In Field	Full Field	In Field	In Field	Full Field	In Field	Full Field
Epiretinal irregularity
ERM		272/96	272/96	14.8%	16.6%	209/127	99.7%*	99.7%*	0/2	0/2
MH		18/27	18/27	20.6%	20.6%	13/24	92.1%*	92.1%*	0/5	0/5
Total		272/97	272/97	14.8%	17.5%	205/130	99.7%*	99.5%*	0/3	0/3
Retina/subretinal irregularity
Abnormal thickness		191/29	191/29	9.7%	10.0%	157/57	99.5%*	99.5%*	0/2	0/2
Hyperreflective feature		281/25	282/25	67.0%	84.5%	72/61	93.6%*	78.4%	11/18	55/50
Hyporeflective feature		230/29	230/90	3.5%	3.5%	214/44	100%*	100%*	0/0	0/0
Total		381/127	399/25	54.9%	77.0%	128/91	99.2%*	95.1%*	1/4	11/18
RPE/choroid irregularity
RPE atrophy		149/60	153/61	76.0%	76.8%	27/32	79.3%	77.4%	9/56	13/57
Drusen/CNV/SR fibrosis/PED		364/56	380/48	77.3%	74.7%	70/105	93.4%*	90.7%*	13/26	25/25
Total		364/56	380/48	77.3%	81.6%	40/98	93.4%*	90.6%*	13/26	25/26
Total abnormalities		424/19	436/10	82.5%	90.7%	29/94	97.6%*	95.9%*	6/9	11/14

 SR Fibrosis, subretinal fibrosis.

*  Represents the sensitivity of 3D-OCT is statistically higher than that of FP (P < 0.05).

Retinal/subretinal irregularity was found positive in 91.1% in field and 94.4% in full field. FP was more sensitive in detecting intraretinal hyperreflective feature (67.0% infield and 84.5% in full field) than abnormal retinal thickness (9.7% in field and 10.0% in full field) or hyporeflective feature (3.5% in field and in full field). 3D-OCT showed relatively consistent sensitivities (from 78.4% to 99.5%) in detecting all retinal/subretinal irregularities in the study. 

Compared with FP, 3D-OCT was more sensitive to detect abnormal retinal thickness (retinal edema or retinal atrophy) both in field and in full field. Similarly for hyporeflective feature, including CME or SRF, 3D-OCT was able to detect 100.0% of abnormalities; while FP only had sensitivity of 3.5% both in field and in full field. 

For detection of intraretinal hyperreflective feature, the in field sensitivity of 3D-OCT (93.6%) was higher than FP (67.0%); however, adding out field, 3D-OCT (78.4%) was no longer more favorable than FP (84.5%). Using combined findings from OCT and in field FP as ground truth, a total of 20.0 cases with irregularities were missed by 3D-OCT, including 3.0 for hemorrhage, 15.5 for pigment migration, 1.5 for aneurysms, 2.0 for abnormal vessels, and none for exudates or cotton wool spots or IRMA (a total of 29 eyes with more than one category graded by FP). Using combined findings from OCT and full field FP, this number increased to 75.0, including 1.0 for exudate, 4.0 for hemorrhage, 74.5 for pigment migration, 1.5 for aneurysms, 9.5 for abnormal vessels, and none for cotton wool spots or IRMA (a total of 105 eyes with more than one category graded by FP) (Fig. 2). Altogether, a total of 102.5 in field hyperreflective irregularities were missed by FP compared with the ground truth (Table 4). Among them, 44.5 were considered related to vascular abnormality, due to its clues for retinal edema or cystoid changes in the same or adjacent B-scans; and 83.5 related to RPE abnormality, characterized by hyperreflective foci in several adjacent B-scans in 3D-OCT, mainly in the outer nuclear layer with evidence of disruptive RPE integrity (Fig. 3). 

Figure 2

 

Examples of false negative cases of hyperreflective features for 3D-OCT. (A–C) Color images graded as definite pigment migration in field. (D) Color image graded as definite pigment migration in field, hemorrhage in field and out of field. (E) Color image graded as definite abnormal vessels in field and out of field. (F) Color image graded as definite hemorrhage in field and out of field and pigment out of field. (G) Color image graded as definite hemorrhage in field, out of field, questionable aneurysms in field and out of field, pigment migration out of field. (H) Color image graded as questionable aneurysms in field, definite pigment migration out of field. (I) Color image graded as questionable hemorrhage in field and out of field, questionable aneurysms in field and out of field, definite pigment migration in field and out of field.

Figure 3

 

Examples of false negative cases of retinal/subretinal irregularity for nonmydriatic FP and 3D-OCT. (A1–A3) Color image graded as absent for abnormal retinal thickness and macular edema and OCT B-scans of the same eye with intraretinal hyporeflective feature (circle in [A2]) and increased retinal thickness (A2, A3). (B1–B3) Color image graded as absent for retinal/sub retinal irregularity and OCT B-scans of the same eye with definite intraretinal hyperreflective feature (circle in [B2]) hyporeflective feature (arrows in [B3]), and increased retinal thickness (B2, B3). (C1–C3) Color image graded as absent for pigment migration and SRF and B-scan with definite intraretinal hyperreflective feature (circle in [C2]) and questionable SRF (circle in [C3]).

Looking at pigment migration only, using combined findings (hyperreflective features with RPE origin as the OCT findings and/or pigment clumping from FP) as ground truth, 39.7% of cases from our study had pigment migration in field. The sensitivities of FP and 3D-OCT were 82.2% and 19.4%, respectively. A total of 126.5 FN cases in field and 229.5 in full field were found for 3D-OCT; while a total of 30.5 FN cases in field and 17.0 in full field were found for FP. 

For hemorrhage only, using maximum findings from both OCT (hyperreflective features with vascular origin) and FP (hemorrhage) as ground truth, the sensitivity of FP versus 3D-OCT was 86.7% vs. 14.1% in field and 88.7% vs. 13.2% in full field. 3D-OCT failed to detect a total of 85.0 cases with hemorrhage in field and 92.5 in full field. And FP was unable to detect a total of 14.0 eyes with hemorrhage in field and 13.0 in full field. 

RPE/choroidal irregularity was found positive in 93.8% in field and 95.5% in full field. FP and 3D-OCT demonstrated similar sensitivities to detect RPE atrophy (Table 4). For the ability to detect drusen/CNV/subretinal fibrosis/PED, 3D-OCT was more sensitive than FP both in field and in full field (Table 4). FN examples for both modalities were shown in Figure 4. 

Figure 4

 

Examples of false negative cases of RPE/choriodal irregularity for nonmydriatic FP and 3D-OCT. (A1, A2) Color image graded as definite present for RPE atrophy and OCT B-scans of the same eye without RPE thinning (A2). (B1, B2) Color image graded as definite present for drusen and OCT B-scans of the same eye without RPE irregularity. (C1, C2) Color image graded as absent for RPE/choroidal irregularity and OCT B-scans of the same eye with definite PED. (D1, D2) Color image graded as absent for RPE/choroidal irregularity and OCT B-scans of the same eye with definite CNV/subretinal fibrosis.

Altogether, as shown in Table 4, the sensitivities for 3D-OCT were higher than FP for overall detection of a variety of abnormalities assessed in the study or irregularities for each category (Table 4). For specific features, 3D-OCT was more sensitive than nonmydriatic FP in most features (e.g., abnormal retinal thickness); however, with limitation for the full field sensitivity of hyperreflective feature in comparison of FP. 

Abnormalities Detected by 3D-OCT for the Eyes With Ungradable FP

For 26 eyes with ungradable FP, 3D-OCT detected additional 19.5 eyes with epiretinal irregularity, 15.5 eyes with decreased retinal thickness, 9.0 eyes with increased retinal thickness, 15.5 eyes with hyperreflective feature, 12.0 eyes with hyporeflective foci. RPE/choroidal irregularity was found in a total of 23.0 eyes. Altogether, 24.0 eyes with at least one kind of abnormalities were detected by 3D-OCT only in this group. 

Discussion

In this study, the sensitivity of 3D-OCT scanning was found to be higher than that of nonmydriatic FP for overall detection of a variety of retinal irregularities or irregularities categorized as epiretinal, intraretinal, or RPE/choroidal, evaluated in our high-risk population. When a single specific feature was speculated, 3D-OCT demonstrated various detection abilities: higher than FP for abnormal retinal thickness (or intraretinal hyporeflective features, et al.); similar as FP for RPE atrophy; however, lower for intraretinal hyperreflective features (e.g., pigment migration or intraretinal hemorrhage). 

Screening for retinal and choroidal pathology, especially DR and AMD, is a challenging and important task for both the undeveloped and the developed world. The traditional concept to screen for asymptomatic patients is important because, for example, the majority of patients who develop DR have no symptoms until macular edema and/or proliferative diabetic retinopathy (PDR) are already present. In addition, the efficacy of laser photocoagulation in preventing visual loss from PDR and macular edema is well established in randomized trials. However, the current screening methods may be inadequate to handle the increased burden brought on by aging populations and eye disease epidemics. The growing disparity between the number of eye care providers and number of patients in need of care suggests that improving the efficiency of eye care delivery at all levels is also important. ¹⁵ A more efficient and cost effective method to screen the high-risk population is expected. 

With the advent of OCT imaging, high-quality, cross-sectional images of the neurosensory retina can be acquired without pupil dilatation in a matter of seconds. In addition, (semi-) automated segmentation of the OCT images allowed thickness measurement and qualitative estimation, ^16–19 which could be possibly adapted for evaluation in a screening setting. As a result, volume scanning with OCT has potential as a method for screening for chorioretinal disease. Recently, a much higher sensitivity of 3D-OCT for overall detection of a variety of retinal irregularities in comparison with nonmydriatic photography was found in an asymptotic population (only a total of 25.5 retinal/choroidal irregularities were observed by in field FP). ⁶ However, in the current report, the sensitivities of 3D-OCT and FP were tested and compared in a population with much higher risk, in which, a total of 438.5 retinal/choroidal irregularities were observed in field. 

Similarly, as reported in the previous study, 3D-OCT again, showed almost perfect sensitivity to detect epiretinal irregularity (including ERM and MH) and some retinal/subretinal irregularity (including abnormal retinal thickness, or intraretinal hyporeflective features) compared with FP both in field and in full field. Fundus images, on the other hand, failed to detect quite a large number of abnormalities observed by 3D-OCT (272.5 ERM, 25.0 MH, and 185.5 abnormal thickness) in the same field and only had sensitivity of 3.5% both in field and in full field for detection of hyporeflective features. In addition, even though nonmydriatic color images had higher quality limitations, 3D-OCT could detect more abnormalities in the eyes with ungradable FP, which enabled 3D-OCT to be more capable in the screening studies. The large advantage of 3D-OCT to detect these potential vision-threatening features makes 3D-OCT an irreplaceable candidate as both a screening tool and a treatment guidance in addition to FP for retinal/choroidal diseases. 

However, 3D-OCT did not always demonstrate more favorable ability compared with FP. For detection of intraretinal hyperreflective feature, the sensitivities of 3D-OCT and FP vary upon different etiologies in the study. Overall, 3D-OCT was more sensitive than FP in field (93.6% vs. 67.0%); however, less sensitive than FP in full field (78.4% vs. 84.5%). The inability for 3D-OCT in field, in order was pigment migration, hemorrhage, abnormal vessels, and aneurysms, with a total of 20.0 FN cases; while in full field, it was pigment migration, abnormal vessels, hemorrhage, and exudates (Fig. 2). 

Intraretinal hemorrhage is a common feature for assessment of diabetic retinopathy and is widely used for its severity classification. ²⁰ In addition, new hemorrhage is also a sign of activity in neovascular AMD (nAMD) or other vision threatening diseases, which probably requires immediate treatment decisions. To possibly recognize hemorrhage with the existing screening methodology in a high-risk population is quite an important pursue. It was reported that hyperreflective features represented hemorrhage in the epiretinal space ^21,22 or subretinal space. ^23,24 Thus, by using the intraretinal hyperreflective features seen by 3D-OCT, detection of intraretinal hemorrhage could be possibly achieved. Our result did not show favorable evidence: 3D-OCT was far less sensitive to detect hemorrhage than FP both in field and in full field. This finding is, however, consistent with the previous reports. ^6,25 Thus, with the current method, 3D-OCT as a single screening modality is not sufficient for hemorrhage screening. However, since hemorrhage is often accompanied by exudates or edema, detection of hemorrhage with the accompanying features improves the detection of observable/referable DR or nAMD. ²⁶ For example, considering hemorrhage as part of the intraretinal hyperreflective features, 3D-OCT only missed a total of 3.0 cases compared with the ground truth, counting for 3.3% of total hemorrhage cases. As a result, a lesion with hemorrhage as its component could possibly not be missed by 3D-OCT. Thus, 3D-OCT as an easy imaging modality could be considered as a candidate in addition to FP for retinal/choroidal disease screening. 

Pigment migration was documented as intraretinal hyperreflective foci with high speed ultrahigh resolution OCT ⁸ and was described as a risk factor for eyes to develop advanced AMD. ²⁷ Thus, the ability to document pigment migration is useful not only for detecting retinal/choroidal disease, but also for monitoring progression of eyes with high risk. Interestingly, our results were quite different from the previous report, in which, 30 out of 31 eyes with RPE clumping seen by FP were also observed with OCT. ⁸ In our study, 3D-OCT only was able to detect 19.4% cases with evidence of pigment migration observed by FP. Different from the prospective nature of the previous report, in our retrospective study, a relatively overall lower fundus image quality might lead to the over evaluation of RPE clumping, resulting in some false negative cases of pigment migration in our study. However, this only explained a portion of our results. Meanwhile, FP also showed weakness to detect pigment migration, with a total of 30.5 FN cases. As a result, both FP and 3D-OCT might be needed in assessment of pigment migration in a future study. 

In the current study, a higher percentage of RPE atrophy was found, with similar sensitivities from FP and 3D-OCT (Table 4). For detection of drusen/CNV/subretinal fibrosis/PED, 3D-OCT was more sensitive than FP both in field and in full field. However, since a detailed grading for single component of drusen/CNV/subretinal fibrosis/PED is practically not always possible, it was not done in our study. Thus, using RPE irregularity observed by 3D-OCT was not adequate enough to indicate specific etiologies. However, with its higher sensitivity, 3D-OCT could possibly pick up more abnormal cases. With the initial aid of OCT, further detailed examination could be done in an early stage, which makes OCT a good candidate for screening purpose. 

Our study has limitations. Although clinical diagnosis for each eye was available, ground truth for each feature (e.g., macular edema) was obtained by using combined appearance from FP and 3D-OCT, without true clinical confirmation of the presence of each feature. In addition, applying the questionable grading protocol and the calculating method for Q grades, led the sensitivities incomparable with most other studies. However, our previous work using a similar method has provided adequate evidence to draw the conclusion for the current study. ¹ In addition, we used Topcon 3D-OCT-1000 (Topcon Corp.) for this study. Recently, a newer model incorporating a digital fundus camera with considerably higher resolution was introduced by the same company. However, the data was acquired in a real clinic setting, which mimics the actual screening process. Under this situational presumption, 3D-OCT demonstrated even more flexibility and better sensitivity. 

Conclusions

In this study, the sensitivity of 3D-OCT was higher than nonmydriatic imaging for overall detection of retinal abnormalities in our high-risk population. 3D-OCT was also more sensitive than FP for irregularities categorized as epiretinal, intraretinal, or RPE/choroidal. However, OCT itself demonstrated limitation to detect intraretinal hemorrhage and pigment migration. It is likely that OCT will be added to photography screening for chorioretinal diseases in the future. 

Acknowledgments

Supported by grants from Research to Prevent Blindness; Carl Zeiss Meditec, Optos, and Optovue, Inc. (SRS); and the Department of Health's NIHR Biomedical Research Centre for Ophthalmology at Moorfields Eye Hospital and UCL Institute of Ophthalmology (PAK). 

Disclosure: Y. Ouyang, None; F.M. Heussen, None; P.A. Keane, None; S.R. Sadda, Optos (F, C), Carl Zeiss Meditec (F, C), Optovue (F), Allergan (C), P; A.C. Walsh, Heidelberg Engineering (C), P