Identifying the Boundaries of Retinal Pigment Epithelial Detachments Using Two Spectral-Domain Optical Coherence Tomography Instruments

Transcription

1 CLINICAL SCIENCE Identifying the Boundaries of Retinal Pigment Epithelial Detachments Using Two Spectral-Domain Optical Coherence Tomography Instruments Fernando M. Penha, MD, PhD; Giovanni Gregori, PhD; Zohar Yehoshua, MD; William J Feuer, MS; Philip J. Rosenfeld, MD, PhD BACKGROUND AND OBJECTIVE: To compare two spectral-domain optical coherence tomography (SD- OCT) instruments in identifying the boundaries of retinal pigment epithelium detachment (PED). PATIENTS AND METHODS: 27 eyes were scanned with Cirrus and Spectralis SD-OCT instruments during a single visit. Two Cirrus scan patterns were used: the and covering a 6 6 mm (20 20 ) area. The Spectralis scan pattern consisted of seven B-scans, averaged 51 times, covering a 30 5 area. The main outcome measures were the retinal thickness at the foveal center and the number of segmentation failures on the central B-scan. RESULTS: The Spectralis algorithm failed to follow the proper retinal contour in 25 eyes (92.6%), while the segmentation on the Cirrus instrument was successful in every central B-scan. Spectralis yielded greater retinal thickness measurements in all cases, and the average difference between Cirrus and Spectralis was 139 µm (P <.001). The intraclass correlation coefficient between the two Cirrus scan patterns was 0.998, and Cirrus versus Spectralis was CONCLUSIONS: The Cirrus SD-OCT instrument identifies the appropriate segmentation boundaries in the presence PED. The Spectralis SD-OCT algorithm was unreliable in segmenting PEDs, leading to inaccurate retinal thickness measurements unless manual adjustments were performed. [Ophthalmic Surg Lasers Imaging Retina 2013;44:10-16.] INTRODUCTION Retinal pigment epithelium detachments (PEDs) have been associated most commonly with age-related macular degeneration (AMD), central serous chorioretinopathy, and polypoidal choroidal vasculopathy. 1 Historically, the diagnosis and follow-up of PEDs were based on imaging with fluorescein angiography (FA) and indocyanine green angiography (ICGA). More recently, optical coherence tomography (OCT) has proven useful for identifying and following PEDs. 1,2 The introduction of spectral-domain OCT (SD- OCT) instruments with high scanning speeds and dense scanning patterns covering a large retinal area has eliminated the need to interpolate between just six scans in time-domain technology and has led to a more reliable assessment of retinal thickness measurements. 3-6 However, the accuracy of any retinal thickness measurement depends primarily on the accuracy of the segmentation algorithms that identify the retinal boundaries, which include the internal limiting membrane (ILM) and the retinal pigment epithelium (RPE). The currently available SD-OCT instruments each use a different proprietary algorithm. Several papers have recently analyzed various characteristics of these segmentation algorithms,7,8 particularly their reproducibility in a number of different conditions The presence of a PED can present a challenge for a segmentation algorithm due to the deformation of the RPE. 8,17 Accurate identification of the RPE layer is required to obtain an accurate retinal thickness measurement in the setting of PEDs. 13,18 The impor- From the Department of Ophthalmology, Bascom Palmer Eye Institute, University of Miami Miller School of Medicine, Miami, Florida. Originally submitted December 9, Accepted for publication December 15, The research was supported by a grant from Carl Zeiss Meditec, Dublin, California, an unrestricted grant from Research to Prevent Blindness, NEI core center grant P30 EY to the University of Miami, the Jerome A. Yavitz Charitable Foundation, the Macula Vision Research Foundation, Feig Family Foundation, the Gemcon Family Foundation, the Carl and Lily Pforzheimer Foundation, and the Emma Clyde Hodge Memorial Foundation. Address correspondence to Philip J. Rosenfeld, MD, PhD, Bascom Palmer Eye Institute, 900 NW 17th St., Miami, FL prosenfeld@med.miami. edu. doi: / Healio.com/OSLIRetina

2 Figure 1. Retinal pigmented epithelium (RPE) segmentation at the central B-san in a small pigment epithelium detachment (PED) by Cirrus and Spectralis SD-OCT. (A) Color fundus photo with the superimposed location of the SD-OCT dataset indicated by the white square; the yellow line is the scan position. (B) B-scan obtained by the Cirrus instrument without segmentation lines. (C) B-scan obtained by the Cirrus instrument with the internal limiting membrane segmentation (blue line) and RPE segmentation (red line). (D) Color fundus photo with the superimposed location of the SD-OCT scan raster (white horizontal lines); the yellow line is the scan position. (E) B-scan obtained by the Spectralis instrument without segmentation lines. (F) B-scan obtained by the Spectralis instrument with the segmentation lines. The outer boundary follows Bruch s membrane. tance of accurate and reproducible assessment of retinal thickness is apparent with the use of inhibitors of vascular endothelial growth factor (VEGF) for the treatment of neovascular AMD. A change in the retinal thickness measurement along with qualitative assessment of the B-scans is often used to determine re-treatment in eyes undergoing anti-vegf therapy, and vascularized PEDs are often a component of the neovascular lesions Due to the importance of retinal thickness measurements in the presence of PEDs when re-treating patients with anti-vegf therapy and when recruiting patients with neovascular AMD into clinical studies, we evaluated the segmentation performance of two SD-OCT instruments, the Cirrus SD-OCT (Carl Zeiss Meditec, Dublin, CA) and the Spectralis OCT (Heidelberg Engineering, Heidelberg, Germany) in patients with PEDs. METHODS This prospective study was approved by the institutional review board of the University of Miami Miller School of Medicine and was compliant with the Health Insurance Portability and Accountability Act of Consecutive patients with a clinical diagnosis of PED affecting the fovea were recruited from the retina service at the Bascom Palmer Eye Institute. All patients signed a research consent form after a detailed explanation of the study. The Bascom Palmer Eye Institute retina faculty confirmed the clinical diagnosis of a PED. Serous (vascular and nonvascular) and fibrovascular PEDs were included in this study. Digital color fundus imaging, Cirrus SD-OCT imaging, and Spectralis SD- OCT imaging were performed in all patients, and FA and ICGA were ordered according to clinical need. Eyes with serous PEDs were identified ophthalmoscopically by the presence of a circumscribed and translucent RPE detachment. FA, when available, demonstrated an early hyperfluorecense, which developed into a homogenous hyperfluorescent area with well-defined borders during the late phase. 1,23 A serous PED on SD-OCT had a dome-shaped RPE elevation that appeared optically empty. The serous vascularized PEDs on SD-OCT had a dome-shaped elevation of the RPE associated with subretinal or intraretinal fluid, as well as an area of hyperfluorescence, either as a hot spot or plaque seen on ICG angiography. 24 Fibrovascular PEDs presented as an RPE elevation with nonhomogenous hyper-reflective material below the RPE layer and overlying or adjacent intraretinal and/or subretinal fluid on SD-OCT imaging. FA revealed early stippled hyperfluorescence associated with late leakage. 1 Patients with other retinal pathologies were excluded. Other major exclusion criteria included PEDs too large to fit within the 6 6 mm 2 scanning area, PEDs touching the edge of the scan, and PEDs associated with disciform scars and areas of geographic atrophy. Each patient was imaged using both SD-OCT instruments on the same day. Each eye was imaged using two raster scan protocols available on the Cirrus Ophthalmic Surgery, Lasers & Imaging Retina Vol. 44, No. 1,

3 Figure 2. Retinal pigmented epithelium (RPE) segmentation at the central B-san in a large pigment epithelium detachment (PED) by Cirrus and Spectralis SD-OCT. (A) Color fundus photo with the superimposed location of the SD-OCT dataset indicated by the white square; the yellow line is the scan position. (B) B- scan obtained by the Cirrus instrument without segmentation lines. (C) B-scan obtained by the Cirrus instrument with the internal limiting membrane segmentation (blue line) and RPE segmentation (red line). (D) Color fundus photo with the superimposed location of the SD- OCT scan raster (white horizontal lines); the yellow line is the scan position. (E) B-scan obtained by the Spectralis instrument without segmentation lines. (F) B-scan obtained by the Spectralis instrument with the segmentation lines. The outer boundary first follows Bruch s membrane then the RPE. Figure 3. Retinal pigmented epithelium (RPE) segmentation at the central B-san in a complex pigment epithelium detachment (PED) by Cirrus and Spectralis SD- OCT. (A) A color fundus photo is shown with the superimposed location of the SD-OCT dataset indicated by the white square; the yellow line is the scan position. (B) B-scan obtained by the Cirrus instrument without segmentation lines. (C) B-scan obtained by the Cirrus instrument with the internal limiting membrane segmentation (blue line) and RPE segmentation (red line). (D) Color fundus photo with the superimposed location of the SD-OCT scan raster (white horizontal lines); the yellow line is the scan position. (E) B-scan obtained by the Spectralis instrument without segmentation lines. (F) B-scan obtained by the Spectralis instrument with the segmentation lines. The outer boundary does not follow either the RPE or Bruch s membrane. SD-OCT instrument: and A-scans. The protocol consisted of 200 B-scans, each containing 200 A-scans. This is a homogeneous sample grid with A-scans 30 µm apart. The protocol consists of 128 B-scans approximately 47 µm apart. The length of each B-scan is 6 mm. On the Spectralis OCT, a scan pattern was chosen that consisted of 7 B-scans, each averaged 51 times, covering a 30 5 area. The length of each B-scan was 9 mm. The operator assessed the quality of the scan during its acquisition and, whenever possible, an effort was made to delete datasets with poor signal strength or with significant motion artifacts. The patient was repositioned during the scanning session as necessary, but there was no specific requirement to reset the patient or the instrument after acquiring each dataset. The main outcome measure was to compare the segmentation performance on the B-scan that included the foveal center. For the Spectralis OCT, the automatic central retinal thickness measurement was ob- 12 Healio.com/OSLIRetina

4 Figure 4. Retinal pigmented epithelium (RPE) segmentation agreement at the central B-san in a pigment epithelium detachment (PED) by Cirrus and Spectralis SD-OCT. (A) Color fundus photo with the superimposed location of the SD-OCT dataset indicated by the white square; the yellow line is the scan position. (B) B-scan obtained by the Cirrus instrument without segmentation lines. (C) B-scan obtained by the Cirrus instrument with the internal limiting membrane segmentation (blue line) and RPE segmentation (red line). (D) Color fundus photo with the superimposed location of the SD-OCT scan raster (white horizontal lines); the yellow line is the scan position. (E) B- scan obtained by the Spectralis instrument without segmentation lines. (F) B-scan obtained by the Spectralis instrument with the segmentation lines. The outer boundary follows the RPE. tained from the B-scan through the fovea. The foveal B-scans were then selected from the or cube obtained from the Cirrus SD-OCT. Retinal thickness, ie, the distance between the automated segmentations of the ILM and the RPE, was measured manually at the center of the B-scan. A retinal specialist (FP) examined all the scans and evaluated the quality of the segmentation. A segmentation failure was defined as a failure to follow the appropriate retinal boundaries along at least 5% of the B-scan length. Bland-Altman plots were used to evaluate the agreement of the central retinal thickness measurements between the two SD-OCT instruments (Cirrus: and vs. Spectralis) and the two Cirrus acquisition protocols ( and ). All statistical calculations were carried out with the SPSS V16.0 software (SPSS Inc., Chicago, IL). RESULTS Twenty-seven eyes of 25 patients with different types of PEDs were enrolled. The sample included five men and 20 women with a mean age of years (range: 51 to 89 years, SD: 8.43). Six eyes (22.23%) were classified as serous nonvascularized PEDs (Fig. 1), 10 eyes (37.03%) as serous vascularized PED (Fig. 2), and 11 eyes (40.74%) as fibrovascular PEDs (Fig. 3). There are well-known differences between the segmentation algorithms on the two instruments, particularly regarding the location of the RPE boundary. Figure 5. Bland-Altman plot comparing the central retinal thickness (CRT) measurements at the central B-scan by two Cirrus acquisition protocols: and The acquisition protocol tends to give somewhat higher CRT. The solid line represents the average difference that is about 3 µm, close to zero. The Cirrus segmentation algorithm placed the RPE boundary line just below the front edge of the hyperreflective band interpreted as the anatomical RPE. The Spectralis OCT typically segmented the back edge of the same hyper-reflective band, leading to somewhat thicker retinal thickness measurements. 13 However, the Spectralis algorithm failed to properly segment the central B-scan in 25 eyes (92.6%). In 20 of these eyes, the algorithm followed Bruch s Ophthalmic Surgery, Lasers & Imaging Retina Vol. 44, No. 1,

5 Figure 6. Bland-Altman plot comparing the central retinal thickness (CRT) measurements at the central B-scan by Cirrus and Spectralis machine. The solid line represents the average difference, which is about 139 µm. All points are above zero, showing that the Spectralis CRT values are consistently greater than the Cirrus measurements. (A) Represents Cirrus scan pattern versus Spectralis. (B) Represents Cirrus scan pattern versus Spectralis. membrane instead of the RPE (Fig. 1). In the remaining five eyes, the segmentation errors were of a more complex nature, as shown in Figures 2 and 3. Figure 4 illustrates a case in which the Spectralis RPE segmentation was judged successful. The Cirrus instrument correctly segmented the RPE layer in every central B- scan. The mean center point thickness recorded by the Spectralis was µm ± µm ( µm). The mean center point thickness using the Cirrus was µm ± µm ( µm) and µm ± µm ( µm), using the or scan patterns, respectively. There was a small (3 µm) but statistically significant difference between the two Cirrus scan patterns (P =.012) (Fig. 5). There was a large and significant difference between the Cirrus and the Spectralis central retinal thickness measurements (P <.001). The average difference between Cirrus versus Spectralis was 139 µm (SD: 140) and Cirrus versus Spectralis was 136 µm (SD: 137), with consistently larger measurements for the Spectralis instrument. The intraclass correlation between the two Cirrus scan patterns was 0.998, and the Cirrus protocol versus Spectralis was The Bland- Altman plot illustrates this important systematic difference, showing a greater disagreement between the instruments for larger lesions (Fig. 6). DISCUSSION Both time-domain OCT and SD-OCT can provide very reproducible retinal thickness measurements. However, the vast majority of the reproducibility studies with different OCT machines were conducted in healthy subjects A study compared the measurements of central retinal thickness using seven OCT devices in patients without retinal diseases. 26 Not surprisingly, the mean CRT was similar using Spectralis and Cirrus SD-OCT but not identical, with a difference of 20 µm. Giani et al. conducted a study that evaluated retinal thickness measurements using six OCT devices, including both normal and pathologic eyes. 28 They concluded that retinal thickness measurements are different between the instruments due to differences in the algorithm analysis. The greatest difference was 175 µm between SD-OCT Copernicus HR (Optopol Technology SA, Zawiercie, Poland) and Spectralis, because Copernicus identifies the outer boundary as the inner edge of the RPE complex. While there are known differences between the instruments based on the location of the RPE segmentation boundary in normal retinas, the true test of an algorithm is its ability to follow a boundary when the normal anatomy is disturbed. Imaging neovascular AMD patients with SD-OCT can be challenging, especially when there is complex anatomy and the retinal thickness measurements are needed for deciding when to re-treat with anti-vegf therapy or for use in recruitment or as an endpoint in clinical trials. The present research compared the central retinal thickness measurements (on a single B-scan) in patients with PEDs associated with wet AMD using two common SD-OCT commercial instruments. This study was performed to simulate a standard clinical situation faced by physicians when monitoring these 14 Healio.com/OSLIRetina

6 types of patients. Typically, patients with PEDs are monitored by SD-OCT, and qualitative as well as quantitative changes in anatomy, such as the status of macular fluid and central retinal thickness, are followed. The retinal thickness is a meaningful clinical parameter only if the underlying segmentation accurately reflects the anatomical condition. The results of the present paper suggest that the Cirrus SD-OCT algorithm is significantly more accurate than the Spectralis SD-OCT algorithm in the presence of a PED. The Spectralis SD-OCT algorithm produces slightly larger retinal thickness measurements than the Cirrus SD-OCT when both RPE segmentations follow the RPE layer in a normal macula; however, in the presence of PED, the difference in retinal thickness measurements differs dramatically because the Spectralis algorithm fails to follow the RPE layer. Moreover, our study shows that the segmentation algorithm on the Spectralis instrument does not behave consistently with respect to the choice of the outer boundary. While it followed Bruch s membrane in most cases, there were five eyes in which the Spectralis algorithm attempted to follow the RPE layer but failed and two eyes in which it actually followed the RPE anatomy correctly. An analysis of the example in Figure 1 illustrates the inconsistencies. Here the Spectralis segmentation algorithm ignores the large RPE deformation and follows Bruch s membrane but then abandons it to follow the smaller RPE deformation at the left of the scan. Krebs et al. also reported greater retinal thickness with the Spectralis instrument (370.9 ± μm) compared with the Cirrus instrument (271 ± 87.3 μm) in wet AMD patients. 13 In this study, the posterior line was positioned along Bruch s membrane in 84.4% of PED patients. Moreover, in 44.4% of those cases, identification of the posterior boundary failed, following partially both Bruch s membrane and RPE. 13 In principle, the choice of acquisition protocol can influence the performance of the segmentation algorithms. In particular, the Cirrus SD-OCT uses a three-dimensional algorithm that can be affected by the geometry of the sampling grid. Malamos et al. conducted a study to evaluate the segmentation performance in the two principal scan patterns from the Cirrus instrument, the and cubes, in patients with wet AMD. They found that RPE and ILM segmentation were excellent in both scan patterns, and severe failures were less common with the raster scan pattern. 21 For this reason, we decided to use both of the Cirrus SD-OCT scan options in our study. The choice of scan patterns on the Spectralis OCT is not known to have an effect on the segmentation, which is consistent with the fact that the algorithm works on each B-scan separately. The amount of averaging could be a factor, but Sadda et al. demonstrated that B-scan density can be reduced to 32 B-scans (spaced 188 µm apart) with minimum change in resultant retinal thickness measurements when compared to 128 B-scans (spaced 47 µm apart). 29 Our study found a small but statistically significant difference in retinal thickness measurement in the central B-scans between the two Cirrus scan patterns. The difference of 3 µm is of little practical consequence and might be an artifact of our study design because the central B-scan was manually selected to correlate with the Spectralis central scan and measured using the caliper available in the Cirrus machine software, so some variation was to be expected. However, the effects of small changes in the position of the central B-scan are dwarfed by the segmentation differences when comparing the retinal thickness measurements across the instruments. Reproducibility is somewhat simpler to analyze than accuracy in this context, both for practical reasons, because all that is needed is to compare numbers produced by the instruments, and for theoretical considerations, because no real gold standard exists. Therefore, there has been an emphasis on computing and comparing measures of reproducibility. This is certainly important; however, it is only a piece of a complex picture, and the interplay between accuracy and reproducibility is not always clearly illustrated. Obviously, it is very simple to produce extremely reproducible and completely inaccurate measurements. In practice, there is often a trade-off between these two goals, and the most important characteristic of a particular algorithm is the quality of the compromises between acceptable levels for both accuracy and reproducibility. Accurate and reproducible retinal thickness measurements do not give a full picture of a PED and its evolution over time. SD-OCT datasets can offer a wealth of information about the geometry of a PED, which could be clinically important. For example, we have shown that it is possible to produce fully automated measurements of PED volume and area that are very reproducible. 17 In summary, the Cirrus SD-OCT algorithm identifies the appropriate retinal boundaries in the presence of PEDs and provides accurate retinal thickness measurements. The Spectralis SD-OCT algorithm is inconsistent when segmenting PEDs, producing inaccurate retinal thickness measurements that require manual adjustment. It is always important to check whether the automated segmentation is correct before accepting the retinal thickness measurements on any instrument, but this is particularly true when using the Spectralis OCT to image eyes with PED. Ophthalmic Surgery, Lasers & Imaging Retina Vol. 44, No. 1,

7 REFERENCES 1. Zayit-Soudry S, Moroz I, Loewenstein A. Retinal pigment epithelial detachment. Surv Ophthalmol. 2007;52(3): Holz FG, Jorzik J, Schutt F, Flach U, Unnebrink K. Agreement among ophthalmologists in evaluating fluorescein angiograms in patients with neovascular age-related macular degeneration for photodynamic therapy eligibility (FLAP-study). Ophthalmology. 2003;110(2): Chan A, Duker JS, Ko TH, Fujimoto JG, Schuman JS. Normal macular thickness measurements in healthy eyes using Stratus optical coherence tomography. Arch Ophthalmol. 2006;124(2): Kiernan DF, Hariprasad SM, Chin EK, Kiernan CL, Rago J, Mieler WF. Prospective comparison of cirrus and stratus optical coherence tomography for quantifying retinal thickness. Am J Ophthalmol. 2009;147(2): Legarreta JE, Gregori G, Punjabi OS, Knighton RW, Lalwani GA, Puliafito CA. Macular thickness measurements in normal eyes using spectral domain optical coherence tomography. Ophthalmic Surg Lasers Imaging. 2008;39(4 Suppl):S Geitzenauer W, Kiss CG, Durbin MK, et al. Comparing retinal thickness measurements from Cirrus spectral domain- and Stratus time domain-optical coherence tomography. Retina. 2010;30(4): Mylonas G, Ahlers C, Malamos P, et al. Comparison of retinal thickness measurements and segmentation performance of four different spectral and time domain OCT devices in neovascular age-related macular degeneration. Br J Ophthalmol. 2009;93(11): Song Y, Lee BR, Shin YW, Lee YJ. Overcoming segmentation errors in measurements of macular thickness made by spectral-domain optical coherence tomography. Retina. 2012;32(3): Gregori G, Wang F, Rosenfeld PJ, et al. Spectral domain optical coherence tomography imaging of drusen in non-exudative age-related macular degeneration. Ophthalmology. 2011;118(7): Schutze C, Ahlers C, Sacu S, et al. Performance of OCT segmentation procedures to assess morphology and extension in geographic atrophy. Acta Ophthalmol. 2011;89(3): Abramoff MD, Lee K, Niemeijer M, et al. Automated segmentation of the cup and rim from spectral domain OCT of the optic nerve head. Invest Ophthalmol Vis Sci. 2009;50(12): Matt G, Sacu S, Buehl W, et al. Comparison of retinal thickness values and segmentation performance of different OCT devices in acute branch retinal vein occlusion. Eye (Lond). 2011;25(4): Krebs I, Smretschnig E, Moussa S, Brannath W, Womastek I, Binder S. Quality and reproducibility of retinal thickness measurements in two spectral-domain optical coherence tomography machines. Invest Ophthalmol Vis Sci. 2011;52(9): Heussen FM, Ouyang Y, McDonnell EC, et al. Comparison of manually corrected retinal thickness measurements from multiple spectraldomain optical coherence tomography instruments. Br J Ophthalmol. 2012;96(3): Lammer J, Scholda C, Prunte C, Benesch T, Schmidt-Erfurth U, Bolz M. Retinal thickness and volume measurements in diabetic macular edema: a comparison of four optical coherence tomography systems. Retina. 2011;31(1): Grover S, Murthy RK, Brar VS, Chalam KV. Comparison of retinal thickness in normal eyes using Stratus and Spectralis optical coherence tomography. Invest Ophthalmol Vis Sci. 2010;51(5): Penha FM, Rosenfeld PJ, Gregori G, et al. Quantitative imaging of retinal pigment epithelial detachments using spectral-domain optical coherence tomography. Am J Ophthalmol. 2012;153(3): Krebs I, Hagen S, Smretschnig E, Womastek I, Brannath W, Binder S. Reproducibility of segmentation error correction in age-related macular degeneration: Stratus versus Cirrus OCT. Br J Ophthalmol. 2012;96(2): Khetpal V, Heimmel MR, Rao SK, Scott IU, Greenberg PB. Resolution of retinal pigment epithelial detachment associated with exudative age-related macular degeneration following intravitreal ranibizumab therapy. Acta Ophthalmol. 2009;87(1): Keane PA, Liakopoulos S, Ongchin SC, et al. Quantitative subanalysis of optical coherence tomography after treatment with ranibizumab for neovascular age-related macular degeneration. Invest Ophthalmol Vis Sci. 2008;49(7): Malamos P, Ahlers C, Mylonas G, et al. Evaluation of segmentation procedures using spectral domain optical coherence tomography in exudative age-related macular degeneration. Retina. 2011;31(3): Baumann B, Gotzinger E, Pircher M, et al. Segmentation and quantification of retinal lesions in age-related macular degeneration using polarization-sensitive optical coherence tomography. J Biomed Opt. 2010;15(6): Barbazetto I, Burdan A, Bressler NM, et al. Photodynamic therapy of subfoveal choroidal neovascularization with verteporfin: fluorescein angiographic guidelines for evaluation and treatment TAP and VIP report No. 2. Arch Ophthalmol. 2003;121(9): Yannuzzi LA, Slakter JS, Sorenson JA, Guyer DR, Orlock DA. Digital indocyanine green videoangiography and choroidal neovascularization. Retina. 1992;12(3): Grover S, Murthy RK, Brar VS, Chalam KV. Normative data for macular thickness by high-definition spectral-domain optical coherence tomography (Spectralis). Am J Ophthalmol. 2009;148(2): Pierro L, Giatsidis SM, Mantovani E, Gagliardi M. Macular thickness interoperator and intraoperator reproducibility in healthy eyes using 7 optical coherence tomography instruments. Am J Ophthalmol. 2010;150(2): Sull AC, Vuong LN, Price LL, et al. Comparison of spectral/fourier domain optical coherence tomography instruments for assessment of normal macular thickness. Retina. 2010;30(2): Giani A, Cigada M, Choudhry N, et al. Reproducibility of retinal thickness measurements on normal and pathologic eyes by different optical coherence tomography instruments. Am J Ophthalmol. 2010;150(6): Sadda SR, Keane PA, Ouyang Y, et al. Impact of scanning density on measurements from spectral domain optical coherence tomography. Invest Ophthalmol Vis Sci. 2010;51(2): Healio.com/OSLIRetina