Correlation of Fundus Autofluorescence Gray Values with Vision and Microperimetry in Resolved Central Serous Chorioretinopathy

Abstract

Purpose.: To evaluate the prognostic value of grayscale parameters in fundus autofluorescence (FAF) for visual function in resolved central serous chorioretinopathy (CSC).

Methods.: Seventy-six eyes of 67 patients with CSC that had been resolved for more than 4 months were analyzed retrospectively. Both the short-wavelength (SW)-FAF and near infrared (NIR)-FAF gray value parameters, including the mean, standard deviation, and coefficient of variation (CV), were calculated at 350-μm- and 1200-μm-diameter circles centered on the fovea. The FAF gray value parameters correlated with −logMAR best corrected visual acuity (BCVA) and mean microperimetry (MP) at the 2° and 4° diameters from the foveal center.

Results.: The mean −logMAR BCVA was 0.15 ± 0.23. The mean MP was 12.87 ± 3.79 dB at 2° and 13.54 ± 3.37 dB at 4°. The −logMAR BCVA correlated most strongly with the mean SW-FAF gray value at the 350-μm circle centered around the fovea (SW-M350; ρ = 0.353; P = 0.002), and the SD of the SW-FAF gray value at the 350-μm circle centered around the fovea (SW-SD350) correlated most strongly with the MP at 2° (ρ = −0.416, P < 0.0001) and 4° (ρ = −0.435, P < 0.0001). The NIR-FAF gray value parameters did not correlate with the macular function tests.

Conclusions.: In subjects with resolved CSC, FAF gray values correlated with visual function. BCVA correlated most strongly with SW-M350. MP at 2° and at 4° correlated most strongly with SW-SD350.

Central serous chorioretinopathy (CSC) appears to arise primarily from choroidal vascular abnormalities and subsequent dysfunction of the retinal pigment epithelium (RPE), resulting in detachment of the neurosensory retina or the RPE. ¹ After CSC resolves, patients may have residual visual symptoms including shallow relative scotoma, metamorphopsia, mild dyschromatopsia, or reduced contrast sensitivity, despite recovering normal visual acuity (VA). ² Abnormal fundus autofluorescence (FAF) remains in many cases, even though resolved CSC has a good long-term prognosis for visual function, whether it resolves spontaneously or after treatment. ³ Chronic forms of CSC are associated with atrophic and degenerative changes in the retina and RPE and, consequently, with decline in VA. ⁴ However, while the association between visual function and retinal degenerative change is fairly clear for resolved CSC, ^{5 –8} the degree to which degenerative changes of the RPE affect visual function is unknown. Many FAF studies on CSC explain best-corrected visual acuity (BCVA) from the perspective of autofluorescence and demonstrate that there is a moderate correlation between changes in FAF and BCVA and that some characteristics of lesions have prognostic value. ^{9 –12} These studies used qualitative analyses to subjectively grade images or FAF patterns as granular FAF, diffuse or confluent FAF, or descending tracts. One study adopted an objective measurable unit of mean and standard deviation to express the severity of FAF abnormalities. ⁹ A confocal scanning laser ophthalmoscope (SLO) has been used recently to obtain FAF images. This system provides a digital image and a viewer program with image analysis tools. Because a digital image consists of dots that can be expressed as gray values, the FAF gray value within a specific area delineated by a clinician can be expressed as a continuous variable such as the mean (M), standard deviation (SD), or coefficient of variation (CV). These variables can be easily obtained and adopted by clinics. However, the relationship between them variables and visual function has not been extensively investigated. This study assessed the correlation between FAF gray value parameters calculated in the viewer program and indicators of visual function such as BCVA or microperimetry (MP) in subjects with resolved CSC. 

Methods

Approval was obtained from the Korea University Medical Center Institutional Review Board, Seoul, Korea. All research and data collection complied with the Declaration of Helsinki. 

Case Selection

A retrospective chart review was performed for consecutive patients with resolved CSC at Korea University Medical Center between March 1, 2009, and March 31, 2011. Each patient had a documented episode of CSC diagnosed by fluorescein angiography (FA) and indocyanine green angiography (ICGA). CSC resolution was confirmed with spectral domain OCT (SD-OCT). Eyes were included if at least 4 months had elapsed since confirmation of CSC resolution by SD-OCT. When FAF and MP were examined, SD-OCT was performed again to verify complete attachment of the retina. Eyes with other intraocular diseases (including age-related macular degeneration, polypoidal choroidal vasculopathy, epiretinal membrane, diabetic retinopathy, or amblyopia), or a history of ocular surgery (including laser photocoagulation) were excluded. However, eyes with a history of photodynamic therapy (PDT) for CSC or cataract surgery were not excluded. 

Fundus Autofluorescence

Short-wavelength (SW)-FAF and near-infrared (NIR)-FAF imaging were acquired simultaneously by retinal angiography (Heidelberg Retina Angiograph 2 [HRA2], Heidelberg Engineering, Heidelberg, Germany). A view mode of a 30° field was used, with an image resolution of 768 × 768 pixels. A wavelength of 488 nm was used for excitation in SW-FAF imaging, and emitted light above 500 nm was detected with a barrier filter. NIR-FAF images of ocular fundi were obtained at a 787-nm excitation wavelength with a barrier filter used for the detection of emitted light above 810 nm. For both SW- and NIR-FAF, up to 100 single images were averaged (4.7 frames/s), depending on patient fixation, using the ART (automated real-time) averaging mode to obtain a high-quality mean image. 

Gray values were calculated by the viewer program (Eye Explorer, ver. 1.6.1.0; Heidelberg Engineering). Numbers (measured as percentages) represent the gray value relative to the maximum possible value. For 8-bit gray value representation, an absolute value of 0 represents black, while 255 represents white. To measure FAF gray values, circles of 350- and 1200-μm diameter were centered on the fovea and applied to both SW- and NIR-FAF images (Fig. 1). ⁷ The mean (M), standard deviation (SD), and coefficient of variation (CV, CV = SD/M) of the gray value were calculated for each circle (M350, SD350, and CV350 for the 350-μm-diameter circle and M1200, SD1200, and CV1200 for the 1200-μm-diameter circle). 

Figure 1.

 

Short-wavelength FAF (left) and near-infrared-FAF (right) images acquired simultaneously with retinal angiography. Circles of 350- and 1200-μm diameter were centered on the fovea, and the mean and standard deviation of the gray value were calculated in each circle.

Microperimetry

Macular MP was performed using a spectral OCT/SLO (OTI, Toronto, ONT, Canada). The pupil was dilated and the patient was dark adapted for at least 15 minutes. Threshold fundus perimetry was performed on the central 2° and 4° (diameter) of the retina (1° = 250–300 μm, thus 2° = 500–600 μm, and 4° = 1000–1200 μm, encompassing the foveal area) using 13 stimulus points (five cross-shaped points for the central 2°, eight ring-shaped points for the central 2–4°) covering a central area 4° in diameter. ¹³ Goldmann III size stimuli were projected on a whitish green background illumination of 1.27 cd/m² (4 Asb) with a presentation time of 200 ms with a 1000-ms interval between stimuli. Stimuli intensity ranged from 0 dB (127 cd/m²) to 20 dB (2.54 cd/m²). The fixation target used for all subjects was a 1° single cross. The 4- to 2-dB staircase threshold strategy was used, and mean macular sensitivity was calculated at the 2° and 4° diameters from the fovea center. 

Statistics

The mean, standard deviation, and range were used, as appropriate. The Kolmogorov-Smirnov test verified the normality of the distribution of continuous variables. The M, SD, and CV of the fovea gray value obtained from SW- and NIR-FAF images were correlated with −logMAR BCVA and MP, using Spearman's correlation analysis as a nonparametric method. The estimated area, sensitivity, and specificity of the receiver operating characteristic (ROC) curve were calculated. The eyes were divided according to Snellen BCVA (group 1, ≥0.8; group 2, <0.8), ¹⁴ and PDT history. Macular function tests and gray value parameters for each group were compared using the Mann-Whitney test (statistical analyses: SPSS ver. 12.0; SPSS Inc., Chicago, IL). All statistics were two-tailed, and P < 0.05 indicated significance. 

Results

Baseline Characteristics

Seventy-six eyes of 67 patients were analyzed, including 33 (43.42%) right and 43 (56.58%) left eyes and 24 eyes with a history of reduced-fluence PDT. The average patient age was 50.07 ± 9.20 (range, 32–77) years and the male-to-female ratio was 53 (79.10%) to 14 (20.90%). Of the 76 eyes, 40 (52.63%) had acute CSC and 10 (13.16%) had recurrent CSC. The mean distance from the fovea center to the nearest leaking point was 1027 ± 806 μm (range, 230–3950) on FA or/and ICGA. The leaking point was located within the 1200-μm-diameter circle in 27 (35.53%) eyes, and none of the eyes had a leaking point within the 350-μm-diameter circle. Abnormal SW-FAF included the fovea center in 32 (42.11%) eyes, and abnormal NIR-FAF included the fovea center in 57 (75%) of 76 eyes. The mean duration of time from confirmation of CSC resolution to FAF and MP examination was 8.72 ± 6.30 (range, 4–30) months. There were 24 (31.58%) eyes with BCVA <0.8 and 52 (68.42%) with BCVA ≥0.8. Patient characteristics are listed in Table 1. 

Table 1.

 

Test Parameters in Patients with Resolved CSR

	Mean ± SD	Range
−LogMAR BCVA	0.15 ± 0.23	0.00–0.80
MP at 2°, dB	12.87 ± 3.79	0–18.7
MP at 4°, dB	13.54 ± 3.37	0.5–18.5
SW-FAF*
Mean, 350 μm	26.43 ± 10.27	6.48–57.06
SD, 350 μm	4.46 ± 1.56	2.07–9.20
CV, 350 μm	0.18 ± 0.07	0.09–0.57
Mean, 1200 μm	30.94 ± 10.42	13.31–63.57
SD, 1200 μm	6.31 ± 2.03	2.77–11.64
CV, 1200 μm	0.21 ± 0.05	0.10–0.38
NIR-FAF*
Mean, 350 μm	46.45 ± 13.03	6.81–77.06
SD, 350 μm	7.92 ± 2.43	3.49–15.84
CV, 350 μm	0.19 ± 0.08	0.06–0.62
Mean, 1200 μm	51.48 ± 11.73	15.85–76.64
SD, 1200 μm	9.59 ± 3.06	1.41–17.69
CV, 1200 μm	0.20 ± 0.09	0.02–0.52

*  Gray value.

Correlation Analysis

A correlation between −logMAR BCVA and MP at 2° and 4° (ρ = −0.562, P < 0.0001; and ρ = −0.588, P < 0.0001, respectively) was identified. For SW- and NIR-FAF images, SW-M350, SW-SD350, and SW-M1200 were significantly correlated with macular function tests. Among them, −logMAR BCVA had the strongest correlation with SW-M350 (ρ = 0.353, P = 0.002). MP at both 2° and 4° had the strongest correlation with SW-SD350 (ρ = −0.416, P < 0.001, and ρ = −0.435, P < 0.001, respectively). Among the NIR-FAF gray value parameters, no correlation was demonstrated with macular function tests (Table 2). 

Table 2.

 

Spearman Correlation between Gray Values of FAF and Macular Function Tests

	−LogMAR BCVA		MP at 2° (dB)		MP at 4° (dB)
	ρ	P	ρ	P	ρ	P
−LogMAR BCVA	1.000		−0.562	<0.0001	−0.588	<0.0001
MP at 2°, dB	−0.562	<0.0001	1.000		0.934	<0.0001
MP at 4°, dB	−0.588	<0.0001	0.934	<0.0001	1.000
SW-FAF*
Mean, 350 μm	0.353	0.002	−0.368	0.001	−0.434	<0.0001
SD, 350 μm	0.329	0.004	−0.416	<0.0001	−0.435	<0.0001
CV, 350 μm	−0.038	0.746	0.022	0.851	0.095	0.412
Mean, 1200 μm	0.254	0.027	−0.276	0.016	−0.317	0.005
SD, 1200 μm	0.159	0.169	−0.194	0.093	−0.186	0.108
CV, 1200 μm	−0.117	0.316	0.172	0.138	0.251	0.029
NIR-FAF*
Mean, 350 μm	−0.180	0.120	0.155	0.182	0.061	0.599
SD, 350 μm	−0.012	0.916	0.037	0.751	0.070	0.547
CV, 350 μm	0.111	0.342	−0.066	0.569	0.019	0.867
Mean, 1200 μm	−0.153	0.187	0.059	0.615	−0.033	0.776
SD, 1200 μm	0.087	0.455	−0.155	0.183	−0.097	0.405
CV, 1200 μm	0.115	0.323	−0.123	0.290	−0.035	0.764

  n = 76.

*  Gray value.

When the same correlation analyses were performed in eyes with spontaneously resolved CSC that did not receive PDT (n = 52), correlations between −logMAR BCVA and MP at 2° and 4° (ρ = −0.722, P < 0.0001, and ρ = −0.698, P < 0.0001, respectively) were higher than those calculated in all eyes. All macular function tests correlated most strongly with SW-SD350 (–logMAR BCVA: ρ = 0.419, P = 0.002; MP at 2°: ρ = −0.452, P = 0.001, and MP at 4°: ρ = −0.434, P = 0.001) and there was no correlation between NIR-FAF gray value parameters and macular function tests (Table 3). 

Table 3.

 

Spearman Correlation between Gray Values of FAF and Macular Function Tests in Spontaneously Resolved CSC without PDT

	−LogMAR BCVA		MP at 2° (dB)		MP at 4° (dB)
	ρ	P	ρ	P	ρ	P
−LogMAR BCVA	1	—	−0.722	<0.0001	−0.698	<0.0001
MP at 2°, dB	−0.722	<0.0001	1	—	0.951	<0.0001
MP at 4°, dB	−0.698	<0.0001	0.951	<0.0001	1	—
SW-FAF*
Mean, 350 μm	0.286	0.038	−0.364	0.007	−0.432	0.001
SD, 350 μm	0.419	0.002	−0.452	0.001	−0.434	0.001
CV, 350 μm	0.001	0.996	0.015	0.917	0.067	0.635
Mean, 1200 μm	0.220	0.113	−0.315	0.022	−0.355	0.009
SD, 1200 μm	0.298	0.030	−0.366	0.007	−0.346	0.011
CV, 1200 μm	0.029	0.834	0.023	0.868	0.086	0.541
NIR-FAF*
Mean, 350 μm	−0.229	0.098	0.169	0.226	0.075	0.596
SD, 350 μm	−0.141	0.313	0.136	0.333	0.159	0.257
CV, 350 μm	−0.014	0.922	0.015	0.918	0.081	0.562
Mean, 1200 μm	−0.208	0.136	0.192	0.169	0.092	0.513
SD, 1200 μm	0.097	0.488	−0.150	0.284	−0.104	0.460
CV, 1200 μm	0.145	0.300	−0.203	0.145	−0.117	0.404

  n = 52.

*  Gray value.

A significant correlation was found between SW-M and -SD in both the 350-μm-diameter circle (ρ = 0.679, P < 0.0001) and the 1200-μm-diameter circle (ρ = 0.736, P < 0.0001). However, no significant correlation was seen between IR-M and -SD for the 350- or the 1200-μm-diameter circle. There were significant correlations between SW- and NIR-M and between SW- and NIR-CV (Table 4). 

Table 4.

 

Spearman Correlation for Gray Values between SW- and NIR-FAF

	Mean	SD	CV
350-μm-Diameter circle
ρ	0.319	0.145	0.248
P	0.005	0.213	0.030
1200-μm-Diameter circle
ρ	0.440	0.147	0.428
P	<0.0001	0.205	<0.0001

  n = 76.

Intergroup Comparison

When the MP and gray values were compared between groups 1 and 2, group 1 maintained a better MP for both the 350-μm-diameter circle (14.43 ± 2.02 vs. 9.37 ± 4.45, P < 0.0001) and the 1200-μm-diameter circle (14.92 ± 1.91 vs. 10.44 ± 3.86, P < 0.0001). There was a significant difference between the two groups for SW-M350, SW-SD350, and NIR-M350. However, there was no significant difference between the two groups for SW-FAF parameters at the 1200-μm-diameter circle and the other NIR-FAF parameters at either the 350- or 1200-μm circles (Table 5). When visual function was compared between the normal and the abnormal subfoveal FAF groups, there were statistically significant differences in logMAR BCVA (P = 0.002), MP at 2° (P = 0.010), and MP at 4° (P = 0.013) for SW-FAF, but there was no difference in macular functions for NIR-FAF (Table 6). When the visual function tests and gray values between the PDT (−) group and the PDT (+) group were compared, there were significant differences between the two groups in −logMAR BCVA, MP at 4°, SW-M350, and SW-M1200 (Table 7). 

Table 5.

 

Comparison of MP and Gray Values of FAF between Group 1 (Snellen BCVA ≥0.8) and Group 2 (Snellen BCVA <0.8)

	Group 1 (n = 52)		Group 2 (n = 24)		P
	Mean ± SD	Mean Rank	Mean ± SD	Mean Rank
MP at 2°, dB	14.43 ± 2.02	46.97	9.37 ± 4.45	20.15	<0.0001
MP at 4°, dB	14.92 ± 1.91	47.45	10.44 ± 3.86	19.10	<0.0001
SW-FAF*
Mean, 350 μm	24.53 ± 9.50	34.35	30.57 ± 10.84	47.50	0.016
SD, 350 μm	4.13 ± 1.26	34.46	5.18 ± 1.91	47.25	0.019
CV, 350 μm	0.18 ± 0.08	38.75	0.17 ± 0.04	37.96	0.884
Mean, 1200 μm	30.08 ± 11.09	35.99	32.81 ± 8.74	43.94	0.145
SD, 1200 μm	6.18 ± 2.09	36.94	6.60 ± 1.92	41.88	0.365
CV, 1200 μm	0.215 ± 0.06	39.92	0.20 ± 0.04	35.42	0.408
NIR-FAF*
Mean, 350 μm	48.51 ± 13.45	41.97	41.99 ± 11.07	30.98	0.044
SD, 350 μm	7.92 ± 2.52	38.33	7.93 ± 2.29	38.88	0.920
CV, 350 μm	0.18 ± 0.09	35.77	0.20 ± 0.05	44.42	0.113
Mean, 1200 μm	52.67 ± 12.33	40.88	48.90 ± 10.09	33.33	0.166
SD, 1200 μm	9.40 ± 3.17	36.79	10.00 ± 2.85	42.21	0.320
CV, 1200 μm	0.19 ± 0.09	36.17	0.21 ± 0.08	43.54	0.176

 Significances determined by the Mann-Whitney test.

*  Gray value.

Table 6.

 

Comparison of the Visual Function Tests between the Normal Subfoveal FAF Group and the Abnormal Subfoveal FAF Group

	Subfoveal Involvement				P
	No		Yes
	Mean ± SD	Mean Rank	Mean ± SD	Mean Rank
SW-FAF abnormality	(n = 43)		(n = 33)
−LogMAR BCVA	0.08 ± 0.16	N/A	0.25 ± 0.27	N/A	0.002
MP at 2°, dB	13.85 ± 2.94	N/A	11.5 ± 4.39	N/A	0.010
MP at 4°, dB	14.34 ± 2.79	N/A	12.42 ± 3.80	N/A	0.013
NIR-FAF abnormality	(n = 17)		(n = 59)
−LogMAR BCVA	0.07 ± 0.17	30.82	0.18 ± 0.24	40.71	0.081*
MP at 2°, dB	13.58 ± 3.63	43.85	12.61 ± 3.85	36.96	0.256*
MP at 4°, dB	13.85 ± 3.46	42.15	13.41 ± 3.38	37.45	0.439*

*  Mann-Whitney test.

Table 7.

 

Comparison of Visual Function Tests and Gray Values of FAF between the PDT (−) Group and the PDT (+) Groups

	PDT (−) (n = 52)		PDT (+) (n = 24)		P
	Mean ± SD	Mean Rank	Mean ± SD	Mean Rank
−LogMAR BCVA	0.12 ± 0.21	35.34	0.23 ± 0.25	45.78	0.042
MP at 2°, dB	13.24 ± 3.67	41.24	11.88 ± 4.00	32.20	0.101
MP at 4°, dB	13.97 ± 3.28	41.86	12.44 ± 3.46	30.76	0.044
SW-FAF*
Mean, 350 μm	24.36 ± 8.76	34.61	31.23 ± 11.98	47.46	0.020
SD, 350 μm	4.24 ± 1.40	35.89	4.96 ± 1.81	44.52	0.117
CV, 350 μm	0.19 ± 0.08	40.60	0.16 ± 0.03	33.65	0.207
Mean, 1200 μm	29.02 ± 9.58	34.69	35.37 ± 11.15	47.28	0.022
SD, 1200 μm	6.08 ± 2.00	35.82	6.85 ± 2.05	44.67	0.108
CV, 1200 μm	0.22 ± 0.05	39.66	0.20 ± 0.05	35.83	0.487
NIR-FAF*
Mean, 350 μm	46.93 ± 13.24	39.45	45.35 ± 12.76	36.30	0.568
SD, 350 μm	7.89 ± 2.56	37.43	7.99 ± 2.16	40.96	0.523
CV, 350 μm	0.19 ± 0.09	37.42	0.18 ± 0.05	41.00	0.516
Mean, 1200 μm	51.62 ± 11.63	39.23	51.15 ± 12.21	36.83	0.663
SD, 1200 μm	9.45 ± 3.09	37.91	9.92 ± 3.05	39.87	0.722
CV, 1200 μm	0.20 ± 0.09	38.30	0.21 ± 0.09	38.96	0.905

 Significance determined by Mann-Whitney test.

*  Gray value.

ROC Curve

The ROC curve for SW-M350 with BCVA ≥0.8 or BCVA <0.8 are shown in Figure 2. The area under the curve for SW-M350 was 0.673. When a value greater than 26.34 was used as the threshold, a sensitivity of 66.7% and a specificity of 69.2% were achieved. 

Figure 2.

 

Receiver operating characteristic curve of the mean of the SW-FAF gray value in the 350-μm zone with BCVA ≥0.8 or BCVA <0.8.

Discussion

Recent reports have suggested that FAF can be used to predict VA in eyes with CSC. ^9,12 Although the use of FAF in eyes with either resolved or unresolved CSC has been shown, including both resolved and unresolved CSC could interfere with interpreting FAF as related to visual function, because the presence of a detached retina may compromise the function of the neurosensory retina. ^8,15 When only eyes with resolved CSC were included, a moderate correlation (−0.56 to −0.59) was revealed between MP and −logMAR BCVA, which was slightly higher than that reported by Reibaldi et al. ¹⁵ (−0.52 to −0.54). The data of the two studies may vary because of a difference in testing area and the number of eyes included. BCVA correlated with MP at the foveal center. Because BCVA was expected to reflect the function of the central macula, ¹⁶ our study focused on central macular sensitivity and measured the retinal sensitivity of the central 2° and 4°, whereas Reibaldi et al. ¹⁵ scanned the central 12°. The larger the MP measuring area, the smaller the portion occupied by the macula. Therefore measuring the retinal sensitivity of a large area may mask the correlation between BCVA and MP at the fovea. 

SW-M350 and -SD350 had weak but significant correlations with macular function for various FAF parameters. As the gray value increased, both BCVA and MP sensitivity decreased. SW-M350 was also diagnostically useful in differentiating VA ≥0.8 from <0.8. In a study by Spaide and Klancnik, ⁹ of all the covariates tested, normalized central macular autofluorescence (SW-FAF) was the single best predictor of VA, which is consistent with the results of this study. Spaide and Klancnik reported a correlation coefficient for a VA of −0.50, which is higher than the correlation coefficient reported in the present study. ⁹ This difference may be owing to our not normalizing gray values in our study. Spaide and Klancnik normalized the FAF mean gray value of the 1.07-mm-diameter circle by dividing it by the FAF mean gray value of the whole image. However, in the present study, this important process could not be performed because the images were artificially averaged composite images made using the ART technique. Averaging usually enhances contrast by spreading the gray values over the full range from 0 to 255. Accurate measurements can be performed only in single-image frames before averaging. Therefore, instead of normalizing the image, the correlation between macular function and the nonnormalized FAF gray value of the SLO image was evaluated. The goal was to remove the effect of subretinal fluid (SRF) blocking the underlying fluorophores by analyzing cases of CSC that had been resolved for longer than 4 months, thereby allowing for an evaluation of the effects of irreversible degeneration of the retina photoreceptor and RPE, instead of reversible variables such as lipofuscin precipitates. In CSC, increased SW-FAF in the previous detachment area generally returns to baseline about 4 months after SRF resolution, suggesting that some of the hyper-AF deposit can be cleared with time. ¹⁰ Because it was not known precisely when the CSC resolved completely or whether the hyper-AF material was completely removed, a 4-month interval was chosen between the first and last OCT scans, confirming CSC resolution. This 4-month reference period was intended to assure complete clearing of the macular hyper-AF material. 

SD were analyzed because a healthy macula is expected to have a relatively uniform fluorophore distribution, whereas an unhealthy macula is expected to have more variance in gray values because of the localized loss or accumulation of abnormal fluorophores. ^11,17,18 The SD of FAF reflects the variation in gray values within a specific area and could be an indicator of the degree of irregularity. As SW-M increased, SW-SD increased with a moderate correlation for both the 350- and 1200-μm-diameter circles. No correlation was found between NIR-M and -SD. Only SW-SD350 was found to correlate with visual function, and the correlation was weak to moderate (−logMAR BCVA: 0.329, MP: −0.416 to −0.435). Of the SW- and NIR-CV parameters, only SW-CV1200 showed weak correlation with MP at 4°, perhaps because CV was calculated as SD/mean. 

NIR-FAF changes in eyes with CSC have been described in previous studies. ^11,17 Increased granular FAF corresponding to the previously detached area appears earlier in SW-FAF images, but disappears later in NIR-FAF images after CSC resolution. ¹⁷ Sekiryu et al. ¹¹ found that final BCVA was significantly worse in eyes with granular hypo-NIR-FAF than in eyes without these findings. ¹¹ In the present study, although the gray value parameters of NIR-FAF did not correlate with BCVA, a significant difference was found between the NIR-M350 of eyes with good (≥0.8) or low (<0.8) BCVA. 

The PDT(+) group had higher mean gray values in SW-FAF and lower macular function tests than the PDT(−) group, perhaps because the PDT(+) group had a chronic and recurrent type of CSC and consequently showed more pre-PDT pigmentary changes. 

SW-FAF reflects subretinal and RPE lipofuscin deposits, ^9,17,18 and NIR-FAF is generated from the melanin in the RPE and the choroid. ^{19 –21} The metabolism of lipofuscin and melanin depends on both the photoreceptor and RPE. Therefore, abnormal FAF could be an indirect indicator of metabolic status. In this study, NIR-FAF parameters showed no correlation with visual function, whereas SW-FAF parameters showed a weak to moderate correlation, perhaps because variables other than FAF influence visual function and photoreceptors, and RPE with abnormal FAF may maintain function to some degree during changes in FAF. This metabolic change may not correlate directly with functional change. 

This study has some limitations. The most significant one is that it was not possible to normalize FAF gray values because the images were acquired using the ART technique, which meant that the individual image was not a single frame but a composite image derived from averaging multiple frames. The quality of the HRA2 FAF image is highly dependent on ocular media status and patient cooperation, which may compromise the reproducibility and repeatability of FAF images. Second, because of the study's retrospective nature, the patient group was somewhat unevenly distributed according to VA. The proportion of eyes with normal BCVA was relatively high and the proportion with poor vision was relatively low (skewed distribution). CSC has a good long-term prognosis with respect to visual function. Therefore, in this study, 52 (68.42%) of 76 eyes had VA ≥0.8 after resolution of CSC. The large number of eyes in the good VA group could be one of the reasons for the weak-to-moderate correlation observed between VA and gray values. Even 20/20 BCVA eyes had a relatively wide distribution of mean gray values. Third, in many cases, the FAF pattern in CSC has simultaneous irregular bidirectional changes (increased FAF and decreased FAF). Therefore, simply calculating the mean gray value may underestimate changes in FAF. In terms of indicating disease severity, the standard deviation could be an auxiliary variable to complement the mean gray value, because the standard deviation reflects signal irregularities. However, as the disease progresses to a more advanced stage, FAF signals regain homogeneity but show diffuse or confluent hypo-FAF by RPE atrophy. In the multiple linear regression analysis with SW-M350 and -SD350 for −logMAR BCVA, only SW-M350 remained in the final model (data not shown). Further long-term studies on changes in FAF based on time since CSC resolution and the correlation of sequential changes in FAF and visual function are needed to elucidate the relation of FAF gray values on visual function more precisely. Finally, this study lacked OCT image analysis. In previous studies, OCT parameters were highly predictive of visual function. ⁷ However, this study focused on quantified FAF parameters. Further research on the relationship between changes in FAF and RPE morphologic changes and on the relationship between changes in FAF and OCT are needed. 

In summary, SW-M350 correlated most strongly with VA, and SW-FAF SD350 correlated most strongly with MP. However, NIR-FAF did not correlate with visual function. Analysis of the functional significance of FAF gray value parameters will enhance the usefulness of FAF imaging in CSC. 

Footnotes

 Supported by Grant A102024 from the Korean Health Technology R&D Project, Ministry for Health, Welfare and Family Affairs, Republic of Korea.

Footnotes

 Disclosure: J. Oh, None; S.-W. Kim, None; S.-S. Kwon, None; I.K. Oh, None; K. Huh, None