Bittencourt et al. Journal of Ophthalmic Inflammation and ORIGINAL RESEARCH(2019) 9:1Journal of OphthalmicInflammation and InfectionOpen AccessBlue light versus green light fundusautofluorescence in normal subjects and inpatients with retinochoroidopathysecondary to retinal and uveitic diseasesMillena Gomes Bittencourt1†, Muhammad Hassan2†, Muhammad Sohail Halim2, Rubbia Afridi2, Nam V. Nguyen1,2,Carlos Plaza2, Anh N. T. Tran2, Mohamed Ibrahim Ahmed1, Quan Dong Nguyen2 and Yasir Jamal Sepah2*AbstractPurpose: The aim of this study is to evaluate the differences in the fundus autofluorescence (FAF) signal betweenthe blue light autofluorescence (BAF) from Spectralis (Heidelberg, CA) and green light autofluorescence (GAF)200TxTM (OPTOS, UK, in normal subjects and in patients with retinochoroidopathies (RC).Methods: In this prospective study, FAF was performed using BL (λ 488 nm) and GL (λ 532 nm) on normal subjectsand patients with RC. The corresponding pairs of BAF and GAF images from both groups were analyzed usingPhotoshop. The strength of the FAF signal was measured on a gray scale, where optic disc was a standard toindicate absence of AF. In addition, gray values obtained from three identical points (foveal center, and points ofhypo and hyper autofluorescence) in the corresponding BAF and GAF images of normal and RC subjects weredivided by the optic disc value to calculate autofluorescence signal ratio (R). The R values at fovea (R1), hypoautofluorescentpoint (R2), and hyperautofluorescent point (R3) were compared between BAF and GAF modalities, in normaland in RC subjects separately.Results: One hundred six pairs (106 eyes) of FAF images analyzed (37 pairs: normal and 69 pairs: RC subjects). Innormal subjects, the mean R1, R2, and R3 values for BAF were (1.5 0.88, 1.23 0.58, and 4.73 2.85, respectively) andfor GAF were (0.78 0.20, 0.78 0.20, and 1.62 0.39, respectively). Similarly, in subjects with RC, the mean R1, R2, andR3 values for BAF were (1.68 1.02, 1.66 1.15, and 7.75 6.82, respectively) and for GAF were (0.95 0.59, 0.79 0.45,and 2.50 1.65, respectively). The mean difference in the R1, R2, and R3 ratios between BAF and GAF in normal and inRC subjects was statistically significant (p 0.001). The strength of the correlation (r) between ratios for BAF and GAFwas weak or not statistically significant in both normal and RC subjects (p 0.05).Conclusion: The distribution and intensity of the AF signal differ in BAF and GAF and cannot be used interchangeably.In BAF, optic disc signal is always weaker than in other areas, which was not true for GAF where optic disc signal wasstronger than fovea and hypoautofluorescent point in both groups.Keywords: Retinal imaging, Fundus autofluorescence, Autofluorescence imaging, Blue-light autofluorescence, Greenlight autofluorescence* Correspondence: [email protected] Gomes Bittencourt and Muhammad Hassan share the firstauthorship.†Millena Gomes Bittencourt and Muhammad Hassan contributed equally tothis work.2Byers Eye Institute, Stanford University, 2370 Watson Court, Suite 200, PaloAlto, CA 94303, USAFull list of author information is available at the end of the article The Author(s). 2019 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0International License (, which permits unrestricted use, distribution, andreproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link tothe Creative Commons license, and indicate if changes were made.

Bittencourt et al. Journal of Ophthalmic Inflammation and InfectionIntroductionIn recent years, the study of retinal fundus autofluorescence (FAF) has provided important information regarding the production of retinal fluorophores duringphysiological aging and in pathological events [1].Through noninvasive examination techniques, FAF images can map the metabolic status of both retinal pigmentepithelium (RPE) and photoreceptor outer segment.It is well known that the dominant source of FAF signalresults from light excitation of the fluorophores in lipofuscin (LF) [2, 3]. LF is physiologically produced within theRPE and reflects its metabolic activity, which is largely determined by the quantity of photoreceptor outer segmentrenewal [4]. Among the many bisretinoid LF fluorophores,vitamin A-derived pyridinium-bis-retinoid (A2-E) is oneof the most studied components. A2E along with otherfluorophores are subject to photooxidation and photodegradation with resulting damage to RPE secondary toformation of advanced glycosylated end products, complement activation, detergent like effect on lysosomal membranes, impaired lysosomal activity, and generation of freeradicals within the RPE cells [5–8].The fluorescence emission from these fluorophores hasa broad range which peaks at approximately λ610 nm [2,9]. However, these bisretinoids have different excitationmaxima’s in the visible spectrum including λ430 nm (alltrans-retinal dimer), λ439 nm (A2E), λ426 nm (iso-A2E),λ490 nm P-PE), and λ510 nm (all-trans-retinal dimer-phosphatidylethanolamine and all-trans-retinal dimer-ethanolamine) [10–13]. It was also noted that as the excitationwavelength increases, the spectral width of emissionspectrum decreases [3].Such wide range of excitation spectra means that thefluorophores responsible for autofluorescence (AF) inthe FAF imaging can vary depending on which excitationwavelength is used. The main excitation light more commonly used in commercial devices has been the bluelight (BL) (λ488 nm). However, more recently, greenlight (GL) (λ514 nm and λ532 nm) were introduced forclinical use in commercial confocal scanning laser ophthalmoscopes (cSLO) and adapted ultra-wide-field retinal imaging systems. In addition to potentially excitingdifferent fluorophores, GL is less absorbed by macularpigments compared to BL and enhances LF signal inmacula [14].Despite the proximity of BL and GL within the lightspectrum, a potential difference in FAF signal is yet tobe determined in images originated by various commercial devices. Thus, a detailed quantification of the bluelight AF (BAF) and green light AF (GAF) signal will facilitate precise interpretation of AF signals in a clinicalcontext. The index study is a prospective investigationwith the purpose of evaluating the differences in the(2019) 9:1Page 2 of 9FAF signal obtained by the BL used in Spectralis HRA OCT (Heidelberg Engineering Inc., Vista, CA, USA)and GL used in P200Tx (OPTOS Inc., Dunfermline,Scotland, UK), and to provide a better understanding ofthe relationships between the measurements obtainedfrom each device.MethodsThe index study enrolled normal subjects and patientswith an established diagnosis of retinochoroidopathy(RC) secondary to uveitis and other retinal diseases whowere being followed at a tertiary care ophthalmologyclinic. The study was conducted in compliance with thedeclaration of Helsinki, US Code of Federal RegulationsTitle-21, and the Harmonized Tripartite Guidelines forGood Clinical Practice (1996). The study was approvedby the local Institutional Review Board. A written informed consent was obtained from all participants. Allnormal subjects underwent fundus examination to document the health of their retina. Any subject with mediaopacity was excluded from the study.Inclusion and exclusion criteriaPatients were included in the study if they met the following criteria: (1) availability of BAF and GAF imagesof gradable quality, (2) no lesion involving the optic discwhich can affect the optic disc fluorescence, and (3) patients with retinochoroidopathy should have lesionswithin the posterior pole.Fundus autofluorescence imagingAfter pupillary dilation with topical tropicamide andphenylephrine, BAF images were taken using the Spectralis HRA OCT (Heidelberg Engineering Inc., Vista,CA, USA) after subjects were kept in dark room for30 min. No other imaging was performed on the studysubjects prior to fundus autofluorescence (FAF). Spectralis uses an optically pumped solid-state blue laser(λ488 nm) for excitation and uses a barrier filter to capture fundus emissions above λ500 nm of the spectrum.FAF images of a rectangular 30 30 field of view wererecorded with an ametropic corrector. To improve thesignal-to-noise ratio, nine images were aligned and anaverage image with 768 768 pixels was calculated withthe Spectralis HRA OCT software (Heidelberg EyeExplorer, v4.3; Heidelberg Engineering, Germany). Aftera gap of 30 min to account for photobleaching, greenlight FAF (GAF) images were taken using theOptomap-af function in 1000 RexMax mode of P200Tx(OPTOS Inc., Dunfermline, Scotland, UK). A singleGAF image was acquired using green light (λ532 nm)for excitation and by capturing fundus emissions between λ570 and 780 nm of the spectrum. The central

Bittencourt et al. Journal of Ophthalmic Inflammation and Infection30 30 field was manually obtained using the V2 Vantage software.FAF image registration and autofluorescence signal ratioThe pairs of BAF and GAF were co-registered and analyzed using Photoshop (V S5, Adobe Systems Inc., SanJose, CA, US). Only the central 30 30 images wereused for analysis. The BL and GL images were re-sizedto the same pixel/inch rate and aligned based on variousretinal landmarks such as retinal vessels. The strength ofthe FAF signal was defined by the absolute intensity in agray scale which ranges from 1 to 256. The value 1 isconsidered the absolute black color possible and 256 theabsolute white color possible. By definition, low pixelvalues (dark) represent low intensities of autofluorescence signal, and high pixel values (bright) representhigh intensities of AF signal.The optic disc AF intensity was used as a standard toindicate the absence of fluorescence, in both BAF andGAF. Multiple corresponding points (3–5) were selectedin the optic disc area of both BAF and GAF images andtheir intensities were averaged. In addition, the opticnerve head served as an index of background noise.Three points of interest were identified in both BAF andGAF images of normal and RC subjects. These pointsincluded the foveal center, one point in an area ofhypoautofluorescence, and one in an area of hyperautofluorescence. The software ruler and retinal landmarkswere used to guarantee the measurement of the samepoints in both BAF and GAF images.To compare the AF signal strength between the opticdisc and the points of interest, the gray values measuredin the three identified points (Foveal center, hypoautofluorescent, and hyperautofluorescent points) were divided from the values in the optic disc (Fig. 1), tocompute the AF signal ratio (R) in both BAF and GAFimages of the normal and RC subjects. The R values(2019) 9:1Page 3 of 9were calculated and used in this study to account for difference in the two imaging devices in terms of confocality, image capture, and image processing methods. TheR values were labeled as R1, R2, and R3 for foveal center,hypoautofluorescent, and hyperautofluorescent points,respectively.In normal subjects, the hypoautofluorescent point selected was in the vessels and the hyperautofluorescentpoint was chosen in the area 70–150 μm from the fovea.Similarly, in the subjects with RC, a point was selectedin the hypoautofluorescent lesion and another point wasselected in the hyperautofluorescent lesion.Blue light FAF vs green light FAFThe R1, R2, and R3 values were compared to test theagreement and the strength of correlation between BAFand GAF modalities, in both the normal and RC subjects.Statistical analysisThe SPSS (IBM Inc., Chicago, IL) release 19.0.0 wasused for statistical analysis. Demographic characteristicsof the patients were summarized using descriptive statistics and expressed as mean and standard deviations. Themean R values were calculated for points of interest inboth BAF and FAF eyes. To test the agreement betweenR values of BAF and GAF at same location, the meandifference and standard deviation were calculated, andBland-Altman scatter plots were generated. To verify thestrength of the correlation, the Pearson coefficients wereused.ResultsA total of 106 pairs (106 eyes) of FAF images were included in this prospective study. Each pair consisted ofBAF and GAF images of the same area. Thirty-sevenpairs (37 eyes) of the images were from the normal subjects. Seventy-eight pairs (78 eyes) were obtained fromFig. 1 Blue light and green light fundus autofluorescence images with points of interest identified. Fundus autofluorescence images acquiredwith blue light (a) and green light (b) in a normal subject showing the identical points of interest measured in both images. For both images, thegray values obtained in the fovea (F), hypoautofluorescent point (HO), and hyperautofluorescent point (HR) were divided from the gray values inthe optic disc (O) to calculate the autofluorescence signal ratios R1, R2, and R3 respectively

Bittencourt et al. Journal of Ophthalmic Inflammation and Infection(2019) 9:1subjects with various RCs. Table 1 outlines the demographic characteristics of the study population.Autofluorescence signal ratio for BAF and GAF in normalsubjectsThe average R in the fovea (R1) of normal subjects was1.5 0.88 and 0.78 0.20 as determined by BAF andGAF, respectively (Fig. 2). Similarly, the average R2 inthe vessels was 1.23 0.58 and 0.61 0.20 for BAF andGAF images, respectively. Finally, the average R3 in thepoint located 70–150 μm from the fovea was 4.73 2.85and 1.62 0.39 as assessed by BAF and GAF, respectively(Table 2). Figure 2 plots the 95% confidence interval distribution across for ratios measured by BAF and GAF inthe normal subjects.Autofluorescence signal ratio for BAF and GAF in RCsubjectsIn the group of eyes with RCs, the average R valuesfollowed the same trend seen in normal subjects. Theaverage R1 in RC subjects was 1.68 1.02 and 0.95 0.59 as determined by BAF and GAF in the GL FAF images. Similarly, the average R2 in point within hypofluorescent lesion was 1.66 1.15 and 0.79 0.45 for BAFPage 4 of 9and GAF images, respectively. Finally, the average R3 inthe point located within hyperfluorescent lesion was7.75 6.82 and 2.50 1.65 as assessed by BAF and GAF,respectively (Table 2). Figure 2 plots the 95% confidenceinterval distribution across for ratios measured by BAFand GAF in the RC subjects.Agreement between BAF and GAF autofluorescencesignal ratiosThe mean difference in the R1, R2, and R3 ratios between BAF and GAF in normal eyes was statistically significant (R1: 0.72 0.9, p 0.0001 (Fig. 3)); R2: 0.62 0.52, p 0.0001; R3: 3.10 2.81, p 0.0001) (Table 3).Similar results were also observed in eyes with RCs,where difference in the R1, R2, and R3 ratios betweenBAF and GAF was 0.73 1.09 (p 0.0001), 0.86 1.05(p 0.0001), and 4.76 6.56 (p 0.0001), respectively(Table 3). The bland-Altman scatterplots for differencesbetween BAF and GAF ratios in normal and RC subjectsare shown in Fig. 4. The scatterplots show no agreementbetween the AF signal ratios of BAF and GAF images,with a statistically significant difference between bothgroups.Table 1 Demographic characteristics of the study populationDiagnosisNumber of eyesMean age years (SD)Gender (F:M)EthnicityNumber of pair of images37 (17 subjects)32 ( 7)17 F:14 M9C: 2AA37Normal subjects with no known ocular diseaseNormal subjectsRetinochoroidopathy secondary to retinal diseasesHigh myopia1329 ( 6.43)11 M:2 F9A: 4C13Hydroxychloroquine retinal toxicity238 (N/A)2F2C2Aged-related macular degeneration (drusen)174 (NA)1M1C1Central serous chorioretinopathy248 ( 8.66)2M2A2Cone-rod dystrophy475 ( 0.57)4M2C: 2A4Diabetic retinopathy268 (N/A)2M2C2Sickle cell retinopathy148 (N/A)1F1AA1Lymphoma (intraocular)174 (N/A)1M1C1Uveitic retinochoroidopathiesPanuveitis (idiopathic)130 (N/A)1F1C1Vogt-Koyanagi Harada225 (N/A)2M2AA2Acute zonal occult outer retinopathy344 ( 21)3F3C3Punctate inner choroidopathy1031 ( 6.67)8 F:2 M10C10Birdshot choroidoretinopathy554 ( 4.54)3 F:2 M5C5Multifocal choroiditis1640 ( 8.03)16 F12C: 2AA: 2A16Sarcoidosis with posterior uveitis283 (N/A)2F2C2Serpiginous choroiditis238 (N/A)2M2C2231 (N/A)2F2AA269 (42 patients)41 ( 16)38 F:31 M47C: 15A: 7AA69Retinochoroiditis of unclear etiologyTotalAge: SD standard deviation and N/A not applicable; gender: F female and M male; ethnicity: C Caucasian, AA African-American, and A Asian

Bittencourt et al. Journal of Ophthalmic Inflammation and Infection(2019) 9:1Page 5 of 9Fig. 2 Confidence interval plots. Confidence interval plots for autofluorescence signal ratios (R) measured in normal (a) and RC (b) subjects. The Rvalues are represented by blue and green circle for BAF and GAF, respectively. The R values above dotted lines represent signals more intensethan the optic disc signal (dotted line). Similarly, below the dotted are located the AF signals weaker than the signal in the optic discCorrelation between BAF and FAF autofluorescence signalratiosThe strength of the correlation (r) between R values forBAF and GAF was weak or not statistically significant inboth normal eyes (R1 r 0.015, p 0.05; R2 r 0.308,p 0.05; and R3 r 0.158, p 0.05) and RC eyes (R1 r 0.179, p 0.05; R2 r 0.407, p 0.05; and R3 r 0.276,p 0.05).DiscussionWith the advent of cSLO, FAF images have left the research laboratories and have become more popular as apropaedeutic tool to assist in the diagnosis, management, and monitoring of certain conditions. However,despite its incontestable importance, FAF image interpretation can be particularly challenging and are subjectto individual interpretation of gray intensity and contrastbetween normal an abnormal area. Many papers haveTable 2 Autofluorescence signal ratio (R) in normal and RCsubjectsAutofluorescence signal ratio (R)Blue light (λ488 nm)Green light (λ532 nm)Mean95% ConfidenceintervalMeanSD95% ConfidenceintervalSDNormal 93–5.761.620.391.50–1.74RC 98–8.782.501.652.19–2.86SD standard deviationqualitatively characterized the FAF patterns and theirprogression in various retinal pathologies.Lois and colleagues were among the first to quantitatively assess the differential distribution of FAF signal inthe images obtained with cSLO in normal volunteersand in patients [15]. Their work was primarily based onBL FAF images and did not include other wavelengths.As a first effort to explore the difference between blueand green excitatory lights, Wolf-Schnurrbusch et al.evaluated differences in size of geographic atrophy lesionas assessed by BL (λ488 nm) and GL (λ514 nm) using amodified cSLO (HRAmp; Heidelberg Engineering) [16].There was a significant difference in measurements ofGA lesion size between the two wavelengths. Additionally, GAF was better at delineation of borders of atrophic patches in foveal region and preserved fovealisland. Similarly, another study demonstrated slightly larger GA lesion size of GAF compared to BAF with significantly higher intergrader reliability in GAF comparedto BAF [17]. Both of these findings were attributed tobetter contrast offered by the GAF. These findings showthat wavelength used to acquire the images may play arole in the visualization of lesions.Our group proposed a method similar to the one described by Lois et al. to assess the strength of AF signalsand to evaluate the relation, in this case ratio, betweenBL and GL FAF [15]. However, unlike the method ofLois et al., which subtracts the optic nerve head intensityfrom the retinal signal, our method estimates the ratiobetween AF signals in various retinal areas and opticdisc and does not entail the primary purpose of describing the distribution of the AF in degrees of eccentricity.Additionally, using the R values allows us to comparetwo different imaging devices and account for their differences in terms of confocality, image capture, and

Bittencourt et al. Journal of Ophthalmic Inflammation and Infection(2019) 9:1Page 6 of 9Fig. 3 Blue light vs green light autofluorescence signals at the fovea. Blue light autofluorescence (BAF) (a) and green light autofluorescence (GAF)(b) images at the foveal center demonstrating the differences in the foveal autofluorescence (AF) signal. The AF signal due to BAF (a) appears muchdarker compared to the GAF (b) as the blue light is more strongly absorbed by the macular pigments compared to the green lightimage processing methods. In our study, the relation between optic disc and fovea (R1) hypo (R2) and hyperautofluorescent (R3) areas was consistent across normalsubjects and subjects with RC within same modality.However, the R values were not equivalent across BAFand GAF images. The agreement and the correlation ofboth FAF modalities were significantly different supporting the findings seen by Wolf-Schnurrbusch and Pfau etal. [16, 17].In BAF imaging of the normal eyes, the gray intensity in the optic nerve head consisted of dark blackdue to the absence of LF in this area. Similarly, theblood vessels showed a weak strength of the AF signal due to absorption of BL by hemoglobin, yet itwas higher than the optic disc. Normally in the BAF,the foveal center appears hypoautofluorescent due toabsorption of BL by luteal pigment and melanin.However, the AF signal as detected by our study wasTable 3 Difference between autofluorescence signal ratios (R)measured by blue light and green light at the fovea (R1),hypoautofluorescent point (R2), and hyperautofluorescent point (R3)Mean differenceSDSig. (p)Difference between blue and green ratios: normal subjects (N 39images)R1 Difference0.720.900.000R2 Difference0.620.520.000R3 Difference3.102.810.000Difference between blue and green ratios: patients (N 80 images)R1 Difference0.731.090.000R2 Difference0.861.050.000R3 Difference4.766.560.000SD standard deviationstill brighter than the optic disc. In GAF imaging ofnormal eyes, the AF signal at the optic disc washigher than at the foveal center even though normallythe AF signal at fovea is higher in GAF due toweaker absorption of GL by macular pigment compared to the foveal AF signal in BAF as BL is stronger absorbed by the macular pigments (Fig. 3) [18].Similarly, the AF signal at optic disc in GAF washigher than that of the blood vessels. In RC eyes, theR values for BAF and GAF signal followed similarpattern to normal eyes. Figure 5 describes a casedemonstrating difference in the AF signal of ahypoautofluorescent lesion as captured by BAF andGAF. The optic disc had the lowest AF signal in BAFimages compared to fovea and even hypoautofluorescent lesions. On the other hand, optic disc AF signalintensity was higher in GAF images compared tofovea and hypoautofluorescent lesions. The differencein the autofluorescent pattern of the optic disc between BAF and GAF can be because of excitation ofdifferent fluorophores by the two wavelengths (Fig. 6).These differences in the AF signals captured by BAFand GAF underscore the importance of not using theFAF images captured by the Heidelberg Spectralis andOptos P200Tx interchangeably.In our study, we also noted that the range of the confidence interval of the R values also differed between BAFand GAF modalities. The intervals were broader in theBAF images compared to the GAF images. Sparrow etal. demonstrated that emission spectra of the fluorophores present in the LF varied depending on the wavelength used [3]. As the wavelength increased, theemission spectral width decreased considerably from190 nm at λ330 nm to 60 nm at λ545 nm. Therefore,

Bittencourt et al. Journal of Ophthalmic Inflammation and Infection(2019) 9:1Page 7 of 9Fig. 4 Bland-Altman scatterplots. Bland-Altman scatterplots for differences between BAF and GAF autofluorescence ratios (R) in normal (a fovea; bhypoautofluorescent point; c hyperautofluorescent point) and RC (d fovea; e hypoautofluorescent point; f hyperautofluorescent point) subjectsthe variance in emission intensities may be responsibleto a wider confidence interval of gray scale measurements in BAF noted in our study.Our study has provided valuable insight about FAFimage analysis. A variety of diagnoses, from retinalvascular to uveitic diseases, although small in number,provide a diverse sampling of how BAF and GAF maydiffer among different entities. However, our study alsocontained some limitations. The absolute values of theAF signal could not be used for comparison due to theFig. 5 A case outlining discrepancy in blue and green light autofluorescence signals of similar hypoautofluorescent lesions in posterior uveitis.Blue light autofluorescence (BAF) (a) and green light autofluorescence (GAF) (b) images of an 83-year-old Caucasian woman with posterior uveitissecondary to sarcoidosis (retinochoroiditis), showing hypoautofluorescent lesions (yellow arrows) spread throughout the posterior pole. The fundusautofluorescence images show noticeable discrepancies in the autofluorescence AF signals of the hypofluorescent lesions captured by the BAF (a) andGAF (b). Additionally, in the lesion inferior to the optic nerve (yellow circle, red arrow), there is a small area of deep loss of AF signalwithin the hypoautofluorescent lesion as shown by the GAF (b). The detail is not revealed by the BAF image

Bittencourt et al. Journal of Ophthalmic Inflammation and Infection(2019) 9:1Page 8 of 9Fig. 6 Blue light vs green light autofluorescence signals at the optic disc. Blue light autofluorescence (BAF) (a) and green light autofluorescence(GAF) (b) images of optic disc. There are noticeable discrepancies in the autofluorescence signal at the optic disc as captured by BAF (a) andGAF (b)difference in the optics of both devices, translating indifferent intensity of fundus exposure, gain, and distortions. Also, the sample size of our study was relativelysmall, especially in specific categories.ConclusionThe quantitative evaluation of AF signal is an importantstep toward the correct interpretation of BAF and GAFimages in clinical practice. According to our results, thedistribution and intensity of the FAF signal differ in BAFand GAF images acquired by Heidelberg Spectralis andOptos P200Tx , respectively. Therefore, the outcomesfrom these two devices are not simply interchangeabledespite using wavelengths which are in close proximityin visual spectrum. The difference was consistent acrossnormal and RC subjects. Further analyses are indicatedto confirm if the images from Spectralis and P200Tx may provide similar or different AF characterization ofdiseases. Moreover, the discrepancies in our findingsunderscore the importance of evaluating the consequences of design choices made by the manufacturers.AbbreviationsAF: Autofluorescence; BAF: Blue-light autofluorescence; BL: Blue light;FAF: Fundus autofluorescence; GAF: Green-light autofluorescence; GL: Greenlight; LF: Lipofuscin; R: Autofluorescence signal ratio;RC: Retinochoroidopathy; RPE: Retinal pigment epitheliumAuthors’ contributionsMB contributed in the study concept and design, acquisition of data, manuscriptdrafting, and revising. MH contributed in the study concept and design, dataanalysis and interpretation, manuscript drafting, review, and revising.MSH contributed in the analysis and interpretation, manuscript review,and revision. RA contributed in the manuscript drafting. NN contributedin the manuscript review and revision. CP contributed in the manuscriptreview and revision. AT contributed in the manuscript drafting andrevision. MIA contributed in the study concept and design, data analysis.QDN contributed in the study concept and design, interpretation, manuscriptrevision, and approval to publish. YJS contributed in the study concept anddesign, manuscript revision, and final approval to publish. All authors read andapproved the final manuscript.Ethics approval and consent to participateThe study was conducted in compliance with the declaration of Helsinki, USCode of Federal Regulations Title-21, and the Harmonized Tripartite Guidelinesfor Good Clinical Practice (1996). The study was approved by the localInstitutional Review Board. A written informed consent was obtained from allparticipants.Consent for publicationNot applicable.Competing interestsYJS has received research support from Astellas, Genentech, and Optovue,and serves on the Scientific Advisory Board for Genentech/Roche, Optos, andRegeneron.QDN is a recipient of a Physician Scientist Award from Research to PreventBlindness, New York, NY, and serves on the Scientific Advisory Board forAbbVie, Bayer, Genentech, Regeneron, and Santen, among others. QDN alsochaired the Steering Committee for the RISE and RIDE studies and was onthe Steering Committee for the VISTA Study, and other studies sponsored byGenentech and Regeneron.No other authors have received any financial funding or support.AcknowledgementsNot applicable.Publisher’s NoteFundingThere was no funding received for this study.Availability of data and materialsThe datasets used during the current study are available from the correspondingauthor on reasonable request.Springer Nature remains neutral with regard to jurisdictional claims in publishedmaps and institutional affiliations.Author details1Ocular Imaging Research and Reading Center, Menlo Park, CA, USA. 2ByersEye Institute, Stanford University, 2370 Watson Court, Suite 200, Palo Alto, CA94303, USA.

Bittencourt et al. Journal of Ophthalmic Inflammation and InfectionReceived: 23 September 2018 Accepted: 20 December 2018References1. Bindewald A, Bird AC, Dandekar SS, Dolar-Szczasny J, Dreyhaupt J, FitzkeFW, Einbock W, Holz FG, Jorzik JJ, Keilhauer C et al (2005) Classification offundus autofluorescenc

the Spectralis HRA OCT software (Heidelberg Eye Explorer, v4.3; Heidelberg Engineering, Germany). After a gap of 30 min to account for photobleaching, green light FAF (GAF) images were taken using the Optomap-af function in 1000 RexMax mode of P200Tx (OPTOS Inc., Dunfermline, Scotland, U