Ophthalmic Surgery, Lasers and Imaging Retina

Clinical Science 

Comparison of Retinal Layer Intensity Profiles From Different OCT Devices

Zhihong Hu, PhD; Muneeswar G. Nittala, MD; Srinivas R. Sadda, MD

Abstract

BACKGROUND AND OBJECTIVE:

The purpose of this study is to use automated multiple retinal layer segmentation to compare retinal layer intensity profiles between different spectral-domain optical coherence tomography (SD-OCT) devices with and without normalization.

PATIENTS AND METHODS:

A graph-based multistage segmentation approach was used to identify 11 boundaries in horizontal SD-OCT B-scans passing through the foveal center. Four acquisition protocols from two different SD-OCT devices were applied on 34 eyes from 17 healthy participants. The mean intensity of the 11 layers was compared within each device and between the two devices. In addition, the intensity of the various layers was normalized against the vitreous and retinal pigment epithelium band, and the normalized intensity profiles were also compared.

RESULTS:

Within each OCT device, the mean intensity of the 11 retinal layers did not show significant difference ( P values of paired t test ranging between 0.15 and 0.30). For the two different devices, before normalization, the mean intensity of the 11 retinal layers differed significantly ( P = .02 to .03). After normalization, the mean intensity no longer differed significantly ( P = .26 to .51). Linear regression demonstrated an average R 2 of 0.94 ( P < .001) before normalization and 0.98 ( P < .001) after normalization between the two devices.

CONCLUSION:

The morphology of the intensity profiles was found to be similar within each SD-OCT device. Comparing intensity profiles from the two different devices, significant differences in unnormalized layer intensity were observed. Following normalization, the differences between OCT devices were no longer significant.

[ Ophthalmic Surg Lasers Imaging Retina. 2013:44:S5–S10.]

From the Doheny Eye Institute, University of Southern California, Los Angeles, California.

Supported by the Beckman Macular Degeneration Research Center and a Research to Prevent Blindness Physician Scientist Award.

Dr. Sadda is a consultant for and/or receives financial support from Carl Zeiss Meditec, Optos, Allergan, Genentech, Regeneron, and Optovue. The remaining authors have no financial or proprietary interest in the materials presented herein.

Address correspondence to SriniVas R. Sadda, MD, Doheny Eye Institute, 1450 San Pablo Street, Los Angeles, CA 90033; email: ssadda@doheny.org.

Received: June 11, 2013
Accepted: August 27, 2013

Abstract

BACKGROUND AND OBJECTIVE:

The purpose of this study is to use automated multiple retinal layer segmentation to compare retinal layer intensity profiles between different spectral-domain optical coherence tomography (SD-OCT) devices with and without normalization.

PATIENTS AND METHODS:

A graph-based multistage segmentation approach was used to identify 11 boundaries in horizontal SD-OCT B-scans passing through the foveal center. Four acquisition protocols from two different SD-OCT devices were applied on 34 eyes from 17 healthy participants. The mean intensity of the 11 layers was compared within each device and between the two devices. In addition, the intensity of the various layers was normalized against the vitreous and retinal pigment epithelium band, and the normalized intensity profiles were also compared.

RESULTS:

Within each OCT device, the mean intensity of the 11 retinal layers did not show significant difference ( P values of paired t test ranging between 0.15 and 0.30). For the two different devices, before normalization, the mean intensity of the 11 retinal layers differed significantly ( P = .02 to .03). After normalization, the mean intensity no longer differed significantly ( P = .26 to .51). Linear regression demonstrated an average R 2 of 0.94 ( P < .001) before normalization and 0.98 ( P < .001) after normalization between the two devices.

CONCLUSION:

The morphology of the intensity profiles was found to be similar within each SD-OCT device. Comparing intensity profiles from the two different devices, significant differences in unnormalized layer intensity were observed. Following normalization, the differences between OCT devices were no longer significant.

[ Ophthalmic Surg Lasers Imaging Retina. 2013:44:S5–S10.]

From the Doheny Eye Institute, University of Southern California, Los Angeles, California.

Supported by the Beckman Macular Degeneration Research Center and a Research to Prevent Blindness Physician Scientist Award.

Dr. Sadda is a consultant for and/or receives financial support from Carl Zeiss Meditec, Optos, Allergan, Genentech, Regeneron, and Optovue. The remaining authors have no financial or proprietary interest in the materials presented herein.

Address correspondence to SriniVas R. Sadda, MD, Doheny Eye Institute, 1450 San Pablo Street, Los Angeles, CA 90033; email: ssadda@doheny.org.

Received: June 11, 2013
Accepted: August 27, 2013

Introduction

Spectral-domain optical coherence tomography (SD-OCT) is an interference-based, noninvasive, in vivo three-dimensional imaging technique that allows direct visualization of retinal morphology and architecture. 1 It is an optical signal acquisition and processing method that captures the reflected signal from the retinal optical scattering media (ie, the retinal tissue), and thus can be used for the quantitative analysis of the tissue optical properties. 2 Because of the interferometric technique, the SD-OCT image is essentially the intensity profiles of the reflected light of retinal layers.

The various layers of the retina may exhibit different optical properties affected differentially by various diseases. Quantification of the optical properties of these layers may facilitate the understanding of retinal disease. However, existing commercial layer segmentation algorithms have largely focused on a few selected inner layers (such as the nerve fiber and ganglion cell layers). More recently, improvements to existing SD-OCT, such as frame averaging, despeckling, and enhanced image contrast, have allowed the outer retinal structures to be identified more precisely. 3–5

In addition, until recently, retinal layer analysis has largely been focused on retinal thickness. The reflected light from the retinal layers, however, carries more information characterizing the optical properties of tissue. Therefore, the changes in layer intensity may provide complementary information regarding the effects of retinal disease. Our group as well as others have demonstrated that intensity characteristics may be useful features for distinguishing phenotypes in age-related macular degeneration (AMD) 6 and diabetic eye disease. 7

Recently, we developed an automated graph-based multilayer approach to reliably segment 11 retinal layers, including the outer retinal bands in SD-OCT volumes to allow the properties of individual layers to be studied. 8–9 The purpose of this study is to utilize our automated multiple retinal layer segmentation approach to compare retinal layer intensity profiles within each SD-OCT device and between SD-OCT devices with and without normalization.

Patients and Methods

Subject Recruitment

Seventeen healthy participants (nine women and eight men; mean age: 31.25 ± 5.92 years; range: 24–42) with healthy eyes were enrolled in this study at the Doheny Eye Institute of the University of Southern California. The absence of any ocular disease in either eye was confirmed by ophthalmoscopic examination. All participants provided written informed consent. The study was approved by the Institutional Review Board of the University of Southern California and adhered to the tenets set forth in the Declaration of Helsinki.

OCT Imaging

For each participant, both eyes underwent SD-OCT imaging using a Heidelberg Spectralis (Heidelberg Engineering; Heidelberg, Germany) SD-OCT device and a Zeiss Cirrus OCT 4000 (6.5 software; Carl Zeiss Meditec, Dublin, CA) device, both with tracking functionality included (a new addition for the Cirrus). For all acquisitions, a single 6-mm horizontal B-scan through the foveal center was obtained. The Spectralis OCT features a scanning speed that is currently higher than Cirrus OCT 4000. At faster scanning speeds, there is sometimes greater compromise of the signal quality requiring averaging of more B-scan frames to improve image quality. To facilitate better comparison with the Cirrus, two different averaging protocols (both with tracking enabled) were used for the Spectralis OCT, 20× averaged Automatic Real-Time and 40× averaged Automatic Real-Time (referred to herein as “Spectralis 20× tracked” and “Spectralis 40× tracked,” respectively). For the Cirrus OCT, a 20× averaged scan was obtained with the tracking off and tracking on. The two Cirrus scan acquisitions are herein referred to as “Cirrus 20× untracked” and “Cirrus 20× tracked.” Although it is not relevant to the analyses in this report, it should be noted that the scans differed in number of A-scans and slightly in physical dimension. The Spectralis scans consisted of 512 (A-scans) × 496 pixels. The physical scan dimensions varied slightly between cases, but were on average 6.04 mm × 1.92 mm. The Cirrus scan consisted of 1024 (A-scans) × 1024 pixels. The physical scan dimension was 6 mm × 2 mm. The pixel depth for both the Spectralis and Cirrus scans were 8 bits in grayscale, and all scans were oriented for vitreous zero delay.

Segmentation of the Multiple Retinal Boundaries

The multiple retinal boundaries were identified based on a previously described multistage, multisurface segmentation approach first developed for SD-OCT volumes, 8–9 but tuned for the single line scan in this study. Briefly, the algorithm features a three-stage, graph-based approach 10–12 to identify 11 boundaries in a characteristic non-anatomic sequence beginning with the internal limiting membrane (ILM), ellipsoid zone (EZ), retinal pigment epithelium (RPE)/Bruch’s membrane (BM) complex, choroid-sclera (C-S) junction, interdigitation zone (IZ, also considered to be the inner border of the RPE layer for these analyses), nerve fiber layer (NFL)-ganglion cell layer (GCL) junction, inner plexiform layer (IPL)-inner nuclear layer (INL) junction, dendritic outer plexiform layer (OPL)-Henle Fiber layer (HFL) junction, GCL-IPL junction, INL-OPL junction, and finally concluding with the external limiting membrane (ELM).

For stage 1, the four most easily detectable boundaries were identified in four times downsampled images and were used as a priori positional information to limit the graph search for other boundaries at stage 2. Eleven boundaries were then detected in two times downsampled images at stage 2 and refined in the original image space at stage 3 using the graph search, integrating the estimated morphological shape model. A thin-plate spline fitting was applied to each segmented boundary to smooth the boundary. Note that while the multistage, multilayer algorithm was similar in the two different SD-OCT devices, the graph search parameters, such as the smoothness and interaction constraints, were different. Figure 1 is an example illustration of the multiple layer segmentation results of both devices. Although the accuracy of the segmentation algorithm has been previously described, the segmented B-scans were inspected by Doheny Image Reading Center graders to confirm accurate delineation of the retinal layers.

Typical results of automated segmentation of eleven boundaries in 20× averaged Spectralis (top) and Cirrus (bottom) horizontal B-scans with tracking enabled. The left panels show the original line scan through the foveal center without segmentation, and the right panels show the superimposed segmented boundaries. From the top to bottom, the 11 boundaries are 1. ILM; 2. NFL-GCL junction; 3. GCL-IPL junction; 4. IPL-INL junction; 5. INL-OPL junction; 6. Dendritic OPL-HFL junction; 7. ELM; 8. EZ; 9. IZ; 10. RPE/BM complex; and 11. C-S junction.ILM = internal limiting membrane; NFL = nerve fiber layer; GCL = ganglion cell layer; IPL = inner plexiform layer; INL = inner nuclear layer; OPL = outer plexiform layer; HFL = Henle Fiber layer; ELM = external limiting membrane; EZ = ellipsoid zone; IZ = interdigitation zone; RPE = retinal pigment epithelium; BM = Bruch’s membrane; C = choroid; S = sclera.

Figure 1.

Typical results of automated segmentation of eleven boundaries in 20× averaged Spectralis (top) and Cirrus (bottom) horizontal B-scans with tracking enabled. The left panels show the original line scan through the foveal center without segmentation, and the right panels show the superimposed segmented boundaries. From the top to bottom, the 11 boundaries are 1. ILM; 2. NFL-GCL junction; 3. GCL-IPL junction; 4. IPL-INL junction; 5. INL-OPL junction; 6. Dendritic OPL-HFL junction; 7. ELM; 8. EZ; 9. IZ; 10. RPE/BM complex; and 11. C-S junction.

ILM = internal limiting membrane; NFL = nerve fiber layer; GCL = ganglion cell layer; IPL = inner plexiform layer; INL = inner nuclear layer; OPL = outer plexiform layer; HFL = Henle Fiber layer; ELM = external limiting membrane; EZ = ellipsoid zone; IZ = interdigitation zone; RPE = retinal pigment epithelium; BM = Bruch’s membrane; C = choroid; S = sclera.

Characterization of the Retinal Layer Intensity

Eleven layers were defined as the spaces between the segmented boundaries and included: the vitreous, NFL, GCL, IPL, INL, dendritic OPL, outer nuclear complex, inner segments (defined for the purpose of analysis as the layer between the ELM and EZ, though the layer may only include the myoid portion of the inner segment), outer segments (between the EZ and the IZ), RPE layer, and the choroid. It should be noted that the above terminologies were based on the International Nomenclature for OCT Working Group (INOWG) classification (not yet published). The INOWG classification divides the outer nuclear complex into two sub-layers, the outer nuclear layer (ONL) and HFL, and divides the choroid into three sub-layers, the choroiocapillaris (CC), Sattler’s layer (medium vessels), and Haller’s layer (large vessels). However, without the use of directional OCT, 13 consistent segmentation of the sub-layers is difficult. For the available 11 segmented layers, the intensity of all pixels was averaged to generate a mean intensity for each layer, instrument, and acquisition (four for each eye). For the NFL, only pixels from the nasal NFL were used as the NFL temporal to the foveal center was very thin and difficult to distinguish from the ILM.

The mean intensity was compared between the two instruments and the two acquisition protocols for each device. To homogenize the intensity between the devices/acquisitions, the intensity was normalized against the vitreous and RPE. More specifically, the mean intensity of the vitreous was set to zero (ie, subtracting the vitreous intensity from the intensity of the layer of interest), and all the layers were then normalized to the RPE and clipped to a range of 0 to 255. The vitreous and RPE were specifically chosen because they were consistently the darkest and brightest layers on the scan. For consistency of the statistical analysis, the right eyes were horizontally flipped in the x-direction. To increase the statistical power, all 34 eyes from the 17 participants were included for the subsequent intensity analysis.

Results

A typical automated segmentation result for the 11 boundaries from Spectralis and Cirrus OCT (both with 20× averaging and tracking enabled) B-scans are shown in Figure 1 . For rapid qualitative comparisons of layer intensity characteristics, plotting of mean intensity in each layer in a bar graph is useful. Figure 2 (page S8) shows the unnormalized mean intensity profiles of the 11 segmented layers for the two acquisition protocols of the two devices for the 34 eyes.

Mean intensity profiles of two protocols within Spectralis devices and two protocols within Cirrus devices for the 11 layers of all 34 eyes, respectively.Note: 1. Vitreous; 2. Nerve fiber layer; 3. Ganglion cell layer; 4. Inner plexiform layer; 5. Inner nuclear layer; 6. Dendritic outer plexiform layer; 7. Outer nuclear complex (Henle Fiber layer and outer nuclear layer); 8. Inner segments; 9. Outer segments; 10. Retinal pigment epithelium layer; 11. Choroid layer (choriocapillaris layer, Sattler’s layer, and Haller’s layer).

Figure 2.

Mean intensity profiles of two protocols within Spectralis devices and two protocols within Cirrus devices for the 11 layers of all 34 eyes, respectively.

Note: 1. Vitreous; 2. Nerve fiber layer; 3. Ganglion cell layer; 4. Inner plexiform layer; 5. Inner nuclear layer; 6. Dendritic outer plexiform layer; 7. Outer nuclear complex (Henle Fiber layer and outer nuclear layer); 8. Inner segments; 9. Outer segments; 10. Retinal pigment epithelium layer; 11. Choroid layer (choriocapillaris layer, Sattler’s layer, and Haller’s layer).

The Table (page S8) is a summary of all the P values of the paired t test of the unnormalized and normalized mean intensity of the 11 layers between the four different protocols of the two different devices for the 34 eyes. For the two different protocols within each device, the unnormalized mean layer intensity between all the layers for the 20 times and 40 times averaged Spectralis scans appeared similar (with a P value of the paired t test of .30). For Cirrus scans, tracking did not appear to alter mean intensity to any significant extent ( P = .15) between the mean intensity of all the layers.

Paired t Test of the Mean Intensity Between Four Different Protocols of the Different Devices for the 11 Retinal Layers From all 34 Eyes

Table:

Paired t Test of the Mean Intensity Between Four Different Protocols of the Different Devices for the 11 Retinal Layers From all 34 Eyes

When comparing between devices, significant differences in unnormalized mean intensity for all the layers were observed. For instance, the acquisitions of “Spectralis 20× tracked” and “Cirrus 20× tracked,” as shown in the top panel of Figure 3 (page S9), presented a significant difference with a P value of .03. Following normalization, however, the mean intensity values for each of the layers, were more similar ( P = .42). For all four protocols, before normalization, a significant difference in layer intensity was observed between devices with an average P value of .03. After normalization, the difference was no longer significant with an average P value of .38.

Mean profiles of 20 times averaging for the 11 layers of all 34 eyes in Spectralis and Cirrus scans before and after the normalization against the vitreous and retinal pigment epithelium.Note: 1. Vitreous; 2. Nerve fiber layer; 3. Ganglion cell layer; 4. Inner plexiform layer; 5. Inner nuclear layer; 6. Dendritic outer plexiform layer; 7. Outer nuclear complex (Henle Fiber layer and outer nuclear layer); 8. Inner segments; 9. Outer segments; 10. Retinal pigment epithelium layer; 11. Choroid layer (choriocapillaris layer, Sattler’s layer, and Haller’s layer).

Figure 3.

Mean profiles of 20 times averaging for the 11 layers of all 34 eyes in Spectralis and Cirrus scans before and after the normalization against the vitreous and retinal pigment epithelium.

Note: 1. Vitreous; 2. Nerve fiber layer; 3. Ganglion cell layer; 4. Inner plexiform layer; 5. Inner nuclear layer; 6. Dendritic outer plexiform layer; 7. Outer nuclear complex (Henle Fiber layer and outer nuclear layer); 8. Inner segments; 9. Outer segments; 10. Retinal pigment epithelium layer; 11. Choroid layer (choriocapillaris layer, Sattler’s layer, and Haller’s layer).

The mean intensity values for all layers between the two devices for each case were plotted in Figure 4 (page S10). Linear regression demonstrated an average R 2 of 0.94 ( P < .001) before normalization and 0.98 ( P < .001) after normalization.

Scatter plots of the mean intensity of the 11 layers across the four different protocols of the two different devices from all 34 eyes. The blue points correspond to the mean intensity of the layers before normalization, and the red points represent the post-normalization values. The green lines represent a non-biased perfect correlation of R2 = 1. In all comparisons, normalization improved the correlation.

Figure 4.

Scatter plots of the mean intensity of the 11 layers across the four different protocols of the two different devices from all 34 eyes. The blue points correspond to the mean intensity of the layers before normalization, and the red points represent the post-normalization values. The green lines represent a non-biased perfect correlation of R 2 = 1. In all comparisons, normalization improved the correlation.

Discussion

In this study, multiple retinal layer intensity profiles of four different acquisition protocols between two different SD-OCT devices were compared. A normalization technique against the vitreous and RPE was applied to homogenize the intensity between the two different devices.

The morphology of the intensity profiles was similar for the two different acquisition protocols (tracked vs untracked for the Cirrus and 20× vs 40× averaging for the Spectralis) for each device. Interestingly, the use of tracking in Cirrus had no significant difference in the mean intensity for any layer. However, it should be noted that the study sample was relatively small, and the study was not designed to identify extremely small differences between acquisitions. In addition, this initial study only included healthy eyes, which likely have good fixation and thus less of an effect or benefit with tracking.

The significant differences were found in the unnormalized intensity between the two different devices. The overall shape or pattern of the intensity profile, however, was similar between the devices, with the RPE layers demonstrating the brightest intensity. After applying the normalization against the vitreous and RPE, the intensity differences between devices were no longer significant ( P = .38 on average), and showed an even stronger correlation with R 2 of 0.98 ( P < .001) on average.

There are several limitations to consider when evaluating this preliminary study. First, the image and physical size were different in the Spectralis and Cirrus devices. Specifically, the image size for the Cirrus and Spectralis was 1024 (A-scans) × 1024 pixels and 512 (A-scans) × 496 pixels, respectively. The physical scan size for Cirrus was 6 mm × 2 mm and was on average 6.04 mm × 1.92 mm for Spectralis. Although we would not expect these differences to have affected the intensity measurements significantly, a small effect cannot be excluded. Second, though the B-scans were inspected for segmentation errors, extremely small errors of only 1 to 2 pixels may have been difficult to recognize. Because some of the layers were extremely thin, spanning only a few pixels, even these small errors could have affected the results. Third, the “shadowing” effect of retinal blood vessels was not accounted for. This was not a major issue in the present study because only a single B-scan through the foveal center was used and there was a relative paucity of vessels, however this would be a more significant concern for volume/cube scans. For future studies, detection of vessels and exclusion of the portions of the layer affected by the vessel may be important strategies. Fourth, this pilot study only included healthy participants with a relatively narrow age range (mean age: 31.25 ± 5.92 years; 24–42 years). Because retinal thickness is known to decrease with age, 14 it is possible that intensity may change as well. Thus, it is unknown if our findings will extrapolate to other age groups. Finally, our normalization strategy used the RPE and, to lesser extent the vitreous, which could be affected in the setting of disease. Whereas this is not an issue in normal eyes, other layers, such as NFL (less sensitive to disease), may need to be considered for normalization in disease eyes.

In summary, in this study, multiple retinal layer intensity profiles of four different acquisition protocols between two different SD-OCT devices were compared. The morphology of the intensity profiles was found to be similar for the two different acquisition protocols for each device. Comparing OCT intensity profiles from two different SD-OCT devices, differences in unnormalized intensity were observed. These differences, however, could be reduced using normalization strategies that incorporated the RPE (and vitreous) intensity as a reference. This technique for normalization may be of value for homogenizing OCT data across devices and acquisitions to aid in a more consistent interpretation of the retinal morphology and pathology.

References

  1. Drexler W, Fujimoto JG. State-of-the-art retinal optical coherence tomography. Progress Retinal Eye Res . 2008;27:45–88. doi:10.1016/j.preteyeres.2007.07.005 [CrossRef]
  2. Schmitt JM, Xiang SH, Yung KM. Differential absorption imaging with optical coherence tomography. J Opt Soc Am A . 1998;15:2288–2296. doi:10.1364/JOSAA.15.002288 [CrossRef]
  3. Nielsen FD, Thrane L, Black J, Hsu K, Bjarklev A, Andersen P E. Swept-wavelength source for optical coherence tomography in the 1μm range. In Proc. of SPIE New Light Sources, Technologies, and Signal Postprocessing: Optical Coherence Tomography and Coherence Techniques II. 2005;5861:58610H.
  4. Sander B, Larsen M, Thrane L, Hougaard J L, Jorgensen TM. Enhanced optical coherence tomography imaging by multiple scan averaging. Br J Ophthalmol . 2005;89(2):207–212. doi:10.1136/bjo.2004.045989 [CrossRef]
  5. Ferguson RD, Hammer DX, Paunescu LA, Beaton S, Schuman JS. Tracking optical coherence tomography. Opt Lett . 2004;29(18):2139–2141. doi:10.1364/OL.29.002139 [CrossRef]
  6. Lee SY, Stetson PF, Ruiz-Garcia H, Heussen FM, Sadda S. Automated characterization of pigment epithelial detachment by optical coherence tomography. Inv Ophthalmol Vis Sci . 2012;53(1):164–170. doi:10.1167/iovs.11-8188 [CrossRef]
  7. Gao W, Tatrai E, Ölvedy V, et al. Investigation of changes in thickness and reflectivity from layered retinal structures of healthy and diabetic eyes with optical coherence tomography. J Biomed Sci Eng . 2011;4:657–665. doi:10.4236/jbise.2011.410082 [CrossRef]
  8. Hu Z, Wu X, Hariri A, Sadda S. Automated multilayer segmentation and characterization in 3-D spectral-domain optical coherence tomography images. Proc SPIE . 2013;8567.
  9. Hu Z, Wu X, Hariri A, Sadda SR. Multiple layer segmentation and analysis in 3-D spectral-domain optical coherence tomography volume scans. J Biomed Opt . 2013;18(7):76006. doi:10.1117/1.JBO.18.7.076006 [CrossRef]
  10. Hu Z, Abràmoff MD, Kwon YH, Lee K, Garvin MK. Automated segmentation of neural canal opening and optic cup in 3-D spectral optical coherence tomography volumes of the optic nerve head. Inv Ophthalmol Vis Sci . 2010;51:5708–5717. doi:10.1167/iovs.09-4838 [CrossRef]
  11. Lee K, Niemeijer M, Garvin M, Kwon Y, Sonka M, Abràmoff M. Segmentation of the optic disc in 3-D OCT scans of the optic nerve head. IEEE Trans Med Imag . 2010;29:159–168. doi:10.1109/TMI.2009.2031324 [CrossRef]
  12. Li K, Wu X, Chen D, Sonka M. Optimal surface segmentation in volumetric images — a graph-theoretic approach. IEEE Trans Pattern Anal Mach Intell . 2006;28:119–134. doi:10.1109/TPAMI.2006.19 [CrossRef]
  13. Lujan BJ, Roorda A, Knighton RW, Carroll J. Revealing Henle’s fiber layer using spectral domain optical coherence tomography. Inv Ophthalmol Vis Sci . 2011;52:1486–1492. doi:10.1167/iovs.10-5946 [CrossRef]
  14. Alamouti B, Funk J. Retinal thickness decreases with age: an OCT study. Ophthalmol . 2003;87:899–901.

Paired t Test of the Mean Intensity Between Four Different Protocols of the Different Devices for the 11 Retinal Layers From all 34 Eyes

Cirrus 20× (tracked)Cirrus 20× (untracked)
.15 /.90 ††
Spectralis 20× (tracked).3 /.44 ††.03 /.42 ††.03 /.26 ††
Spectralis 40× (tracked).02 /.51 ††.02 /.31 ††

10.3928/23258160-20131101-02

Sign up to receive

Journal E-contents