April 2008
Volume 49, Issue 4
Free
Retina  |   April 2008
Impact of Optic Media Opacities and Image Compression on Quantitative Analysis of Optical Coherence Tomography
Author Affiliations
  • Christoph Tappeiner
    From the Klinik und Poliklinik für Augenheilkunde, Inselspital, Bern, Switzerland.
  • Daniel Barthelmes
    From the Klinik und Poliklinik für Augenheilkunde, Inselspital, Bern, Switzerland.
  • Mathias H. Abegg
    From the Klinik und Poliklinik für Augenheilkunde, Inselspital, Bern, Switzerland.
  • Sebastian Wolf
    From the Klinik und Poliklinik für Augenheilkunde, Inselspital, Bern, Switzerland.
  • Johannes C. Fleischhauer
    From the Klinik und Poliklinik für Augenheilkunde, Inselspital, Bern, Switzerland.
Investigative Ophthalmology & Visual Science April 2008, Vol.49, 1609-1614. doi:10.1167/iovs.07-1264
  • Views
  • PDF
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Christoph Tappeiner, Daniel Barthelmes, Mathias H. Abegg, Sebastian Wolf, Johannes C. Fleischhauer; Impact of Optic Media Opacities and Image Compression on Quantitative Analysis of Optical Coherence Tomography. Invest. Ophthalmol. Vis. Sci. 2008;49(4):1609-1614. doi: 10.1167/iovs.07-1264.

      Download citation file:


      © ARVO (1962-2015); The Authors (2016-present)

      ×
  • Supplements
Abstract

purpose. To analyze the impact of opacities in the optical pathway and image compression of 32-bit raw data to 8-bit jpg images on quantified optical coherence tomography (OCT) image analysis.

methods. In 18 eyes of nine healthy subjects, OCT images were acquired from the central macula. To simulate opacities in the optical system, neutral-density (ND) filters with linear absorption spectra were placed between the OCT device and examined eyes. Light reflection profiles (LRPs) of images acquired with various ND filters were compared. LRPs of the 32-bit raw data were compared with those obtained from the 8-bit jpg compressed images.

results. ND filters induced a linear decrease of reflectivity in OCT images, depending on initial signal intensity. Quantitative OCT analysis showed no significant difference between 32-bit raw data and 8-bit jpg files (P > 0.05).

conclusions. Quantitative OCT analysis is not significantly influenced by data compression. A mathematical model can correct for optical opacities to improve OCT images.

Optical coherence tomography (OCT) is a noninvasive, noncontact imaging technique that allows in vivo visualization of tissue morphology with high resolution. 1 2 3 4 It has been established as a standard diagnostic tool to diagnose and to monitor diseases, especially macular diseases such as exudative forms of age-related macular degeneration and diabetic macular edema. 5 6 The examination is easy to perform and is used not only by retina specialists but also in glaucoma assessment to evaluate retinal nerve fiber layer thickness 7 8 and in ophthalmic surgery of the anterior segment (cataract or keratoplasty) (Fruh et al. IOVS 2007;48:ARVO E-Abstract 4675) to analyze retinal morphology before surgery. 9 Opacities of the optical media, such as cataract or corneal scarring, decrease signal intensity of the OCT examination, resulting in images with low detail and reduced information. 
OCT scans of a normal macula, running through the fovea, show clearly distinguishable reflective layers. Analysis of the reflectivity of these layers as a function of scan depth (Fig. 1)results in a curve (light reflectivity profile [LRP]) with six distinct peaks, hereafter referred to as P1 to P6. 10 11 Compared with retinal anatomy, 12 these peaks represent the retinal pigment epithelium (P1), a highly reflective layer between inner and outer segments of photoreceptors (P2), the external limiting membrane (P3), the outer plexiform layer (P4), the inner plexiform layer, (P5) and the nerve–fiber layer/vitreoretinal interface (P6). It has been suggested that P2 arises from tightly packed mitochondria in the ellipsoid region of the photoreceptors. 10 Because of its unique anatomy, the fovea lacks P4 and P5. The visibility of these peaks is crucial for image interpretation. 
Even though OCT provides an objective measure, interpretation is highly subjective because of the lack of standardized software revealing detailed analysis of the reflectivity of retinal structures. Thus image interpretation depends on the examiner’s experience. Various approaches have been described to provide an automated detailed analysis. 10 11 13 14 15 Controversies about this topic are ongoing, not only about the analytic approach but also about whether raw data from the OCT device or already processed data (built-in analysis software) should be used for analysis. 
In this study we investigated the impact that opacities of the optical media have on OCT images, whether it is possible to correct OCT images for the influence of opacities of the optical media, and whether there are differences between analysis of raw data and processing of jpg data by the OCT device. 
Materials and Methods
Subjects
Eighteen eyes of nine healthy volunteers (five men, four women; mean age, 30 years) were included. Examinations were performed after written informed consent was obtained. Before inclusion in the study, all test participants underwent slit lamp examination, fundus examination, measurement of refractive error, and determination of best-corrected visual acuity (Snellen). Approval for this study was obtained from the Ethics Committee of the University of Bern, Switzerland. The study was performed in accordance with the tenets of the Declaration of Helsinki 1975 (1983 revision). 
OCT Recordings
OCT scanning was performed (Stratus OCT; software version 4.01: Carl Zeiss Meditec AG, Oberkochen, Germany). Horizontal line scans of 5-mm length centered in the fovea were recorded in each eye. Each OCT scan consisted of a linear array of 512 individual depth scans, with each depth scan containing 1024 pixels. Test subjects were informed about the importance of focusing on the fixation point while scans were recorded. After each scan was completed, it was checked visually to ensure that the same region was registered each time; when variations were noted, a new scan was recorded. OCT recording was performed first without filters (native) and subsequently with neutral-density (ND) 11 filters, using the following optical densities: 0.10, 0.15, 0.30, 0.40, 0.60, and 0.80 (Schott AG, Mainz, Germany). This resulted in seven scans from the same region in each eye. Filter characteristics are shown in Figure 2
Data Analysis
Raw scan data and 8-bit grayscale images (jpg) of each scan were exported from the OCT device for further analysis. Raw data (32-bit) and 8-bit jpg images were displayed in grayscale (4096 and 256 grayscales, respectively). Because grayscale levels were used to analyze the reflectivity and not decibels, as provided by the OCT, arbitrary units were used instead of decibels. Calculation of light reflectivity profiles (LRPs) was performed using IGOR 5.04a (Wavemetrics Inc., Lake Oswego, OR). Raw OCT data were loaded into a two-dimensional data matrix of 512 columns in which each data column represented a single depth scan with 1024 data points, and jpg data were imported into a two-dimensional data matrix of 689 columns in which one column represented one pixel line within the jpg image containing 329 data points. No smoothing filters were applied to the imported data. LRP reflectivity values ranged from 0 to 4095 for raw data and from 0 to 255 for 8-bit images. LRPs were calculated for jpg and raw data every 100 μm along each OCT scan by averaging three adjacent single-depth scans for raw data and two adjacent single LRPs for jpg data. This procedure smoothed the curve and emphasized the peaks. Reflectivity in all LRPs was expressed in arbitrary units. P1 (retinal pigment epithelium), the peak that always showed the highest reflectivity, was used as a reference. Based on the findings in unfiltered OCT images, ranges for detecting peaks P2 to P6 were set as mean position in micrometers ±2 SD. 
LRPs were analyzed for reflectivity of the different peaks in raw and processed data in normal and ND filter readings. Peaks were detected using the built-in multipeak finding routine of IGOR. Background reflectivity (noise) was measured in the vitreous, 0.5 mm above the surface of the neuroretina. 
Mathematical Analysis
Datum points corresponding to similar locations in the reflectivity profile and thus to the same location in the retina were analyzed for reflectivity in raw and processed (compressed jpg images) OCT data. The datum points were compared between the two modes (raw and jpg). For each datum point analyzed, a factor was calculated that characterized the decrease in reflectivity (decrease factor). This factor was dependent on the density of the different filters used in the experiments. Correlation analysis between the initial reflectivity (without filters) and this decrease factor was performed to calculate an amplification curve. From this curve, a mathematical function was derived that allowed calculation of a factor characterizing the decrease in reflectivity induced by the neutral-density filter for every single point of the LRP. Calculations were performed using statistical software (Statistica 6; StatSoft Inc., Tulsa, OK). 
Statistical Analysis
Bivariate correlation analysis, regression analysis, and analysis of variance (ANOVA) were used where appropriate. Statistics were calculated (Statistica 6; StatSoft Inc.), and statistical significance was defined as P < 0.05. 
Results
Subjects
All eyes in all study subjects had normal visual acuity (range, 1.0–1.6; median, 1.2). Only minor refractive errors were present: the range of the spherical equivalent was −0.25 D to +0.5 D. Anterior segment morphology and fundus examination results were normal in all subjects. No subject had cataract or other opacities of the optical media. Intraocular pressure was within normal ranges (8–14 mm Hg). 
OCT Analysis
OCT images were analyzed for all structures yielding reflectivity peaks (P1-P6). All peaks could be detected in native OCT scans. With ND11 filters 0.6 (approximately 25% of transmission) and denser, no intraretinal signals (except P2) could be recorded (Fig. 3) . In OCT images recorded with ND11 filters (optical densities of 0.10, 0.15, 0.30, and 0.40), only reflectivity peaks for retinal pigment epithelium (P1), the ellipsoid region of the photoreceptors (P2), the external limiting membrane (P3), and the vitreoretinal interface (P6) could be detected. Although the outer plexiform layer (generating P4) and the inner plexiform layer (generating P5) could still be distinguished on the cross-sectional OCT scans using an ND filter of OD = 0.1 (Fig. 3) , because of the poor signal-to-noise ratio, the peak-finding algorithm failed to reliably detect peaks 4 (outer plexiform layer) and 5 (the inner plexiform layer) on images recorded with any of the ND filters. Background reflectivity (noise) in the vitreous did not show much variation (Fig. 4)
OCT images recorded with ND filters of increasing absorption rates resulted in a linear decrease of reflectivity for each peak examined (Fig. 4) . Depending on the initial reflectivity of every peak examined, the factor determining the amount of decrease of reflectivity varied as a function of the density of the filter. Highly reflective peaks decreased more than less reflective peaks (Fig. 4)
Regression analysis of the initial reflectivity of the peaks evaluated with the respective decrease factor (for the different ND filters) resulted in the curve shown in Figure 5
A mathematical function x′ = a + b × x × log (x), where x′ is the new (corrected) value within the LRP and x the original value, could be derived from this correlation. The parameters a and b are dependent on the type of data and the density of the filter used. 
Application of this mathematical function, which describes this logarithmic curve, allows a correction of LRPs taken from OCT images recorded with ND filters if the ND filter absorption rate is known (Fig. 6) . Comparing an LRP from an unfiltered OCT scan with the corrected LRP from the same retinal location (Fig. 6)shows good agreement between the curves when looking at the intraretinal portion of the curve. However, the noise (region in the vitreous) shows a higher reflectivity than the original curve. This can also be observed in the corrected OCT image (Fig. 6)
LRPs from the same retinal location of OCT images recorded with the same parameters did not differ significantly between the test subjects (P > 0.05). Using this very high interindividual consistency makes it possible to calculate a factor describing the optical properties of the filters by dividing the measured reflectivity of a peak by the normative value for this peak. 
Comparison and statistical analysis of corresponding LRPs derived from raw data and processed jpg data did not differ significantly (P > 0.05). No significant differences were seen in reflectivities of similar peaks of LRPs from different retinal locations (central vs. peripheral) and in recordings without and with ND filters. 
No significant statistical differences could be detected if detail level and reflectivities of similar peaks were compared between LRPs from jpg and raw data (Fig. 7) . Although there was no statistical significance between LRPs, small differences related to the higher spatial resolution of the raw data could be observed (Fig. 7) . As can be seen for P5 or P6 in Figure 7B , smaller subpeaks, as shown in the raw data LRP, appear merged in the jpg data LRP. 
Discussion
We were able to show that opacities of optic media result in a decrease in reflectivity in OCT images of the retina and that it is possible to correct OCT images using a mathematical formula. To date, the subjective assessment of OCT images is the de facto standard for OCT image interpretation. Whereas experienced examiners see small details, less experienced examiners might only look at gross abnormalities and possibly miss important details. The current approaches to interpreting OCT images quantitatively were mainly focused on the distances between intraretinal reflectivity layers, but the different levels of reflectivity should also be included in the analysis because they enhance the significance of the observations. 
Only minor variations were seen in LRPs of different test images recorded from comparable retinal locations under equal OCT recording conditions. Given this, it is possible to calculate corrected OCT images using the algorithm described. This could be of interest for examination of patients before cataract surgery or keratoplasty. Often biomicroscopy does not allow exact fundus examination in eyes with opacities in optic media, but OCT images are still of fair quality to diagnose a retinal disease limiting visual prognosis. Especially in patients with combined pathologic conditions, such as a cataract and inherited retinal disease or a cataract and glaucoma, a quantitative analysis of OCT images is helpful, as in assessing the central retinal tissue or the retinal nerve fiber layer (RNFL) before interventions. As shown previously in healthy controls, optical densities—such as cataracts—significantly influence the measurements of the RNFL. 16 Using methods to correct for these opacities offers the possibility of detecting neuroretinal changes without bias by filtering effects. 
As described in Results, it is possible to calculate a factor describing the decrease of light reflectivity caused by a filter or a cataract. In the present study, a decrease of transmission down to only 30% still allowed reliable analyses. Of course, this implies the presence of healthy retinal tissue. If a retinal disease is suspected, several retinal locations have to be considered for calculating the influence of the opacities. For example, measurements from the fovea in a patient with a dry form of age-related macular degeneration will lead to a false calculation of this factor, whereas reliable results can be obtained by selecting another retinal location with healthy retinal tissue. In the study cohort, there was only minor variation in the reflectivity of the RPE (Fig. 7) , which is in accordance with previously published results in healthy controls. 10 Because of this minor variation and because the RPE can be detected in virtually every OCT scan, we propose the use of the retinal pigment epithelial peak as a basis for the calculation of the so-called decreasing factor. 
Interestingly, there were no significant differences in the analysis of raw data and data processed by the OCT software. This means that it is possible to directly analyze the processed data by the OCT machine. The advantage of analyzing processed images is that the area of interest can be selected directly from an image displayed on the computer screen, thus skipping the time-consuming exportation and analysis of the raw data. In addition, the compression applied by the OCT software reduces the amount of data by 90%. However, taking into account the more detailed information, raw data should be favored for quantitative analysis because small details may be missed when analyzing the compressed jpg data. 
The study presented has some limitations. The Stratus OCT yields signals only within a certain range of filters (up to 0.60). As a result, image alterations resulting from opacities can be corrected only within a certain bandwidth. Besides the technical limitations, there are problems related to fixation. The healthy study subjects were well instructed and could fixate very well. Patients with dense cataract or retinal disease may have problems fixating, however, making it difficult to record OCT images for further analysis in desired regions or resulting in the acquisition of distorted images from fast eye movements when the patient tries to fixate. 
 
Figure 1.
 
(A) Six-millimeter OCT scan through the foveola calculated from raw data. (B) Enlarged area from the scan from (A) illustrating the different retinal reflective layers from the normal perifovea. (C) Corresponding LRP calculated from the region illustrated in (B). Each peak (P1–P6) represents a highly reflective layer. For better visualization, only the part of the LRP illustrating the peaks generated by the neuroretinal layers is shown.
Figure 1.
 
(A) Six-millimeter OCT scan through the foveola calculated from raw data. (B) Enlarged area from the scan from (A) illustrating the different retinal reflective layers from the normal perifovea. (C) Corresponding LRP calculated from the region illustrated in (B). Each peak (P1–P6) represents a highly reflective layer. For better visualization, only the part of the LRP illustrating the peaks generated by the neuroretinal layers is shown.
Figure 2.
 
Filter characteristics of the ND11 filter set. Transmission (in percent, y-axis) is plotted against optical density (x-axis). Filter densities that were used in the experiments are marked by points.
Figure 2.
 
Filter characteristics of the ND11 filter set. Transmission (in percent, y-axis) is plotted against optical density (x-axis). Filter densities that were used in the experiments are marked by points.
Figure 3.
 
Grayscale OCT images illustrating the effect of ND filters on OCT image acquisition. (A) Native image. (BD) OCT images recorded in the same test subject using ND filters with various optical densities. (B) OD 0.1. (C) OD 0.4. (D) OD 0.6. Increasing the optical density of the ND filter results in stronger signal attenuation. At an OD of 0.6 (D), no intraretinal structures are visible except for the photoreceptor signal and RPE.
Figure 3.
 
Grayscale OCT images illustrating the effect of ND filters on OCT image acquisition. (A) Native image. (BD) OCT images recorded in the same test subject using ND filters with various optical densities. (B) OD 0.1. (C) OD 0.4. (D) OD 0.6. Increasing the optical density of the ND filter results in stronger signal attenuation. At an OD of 0.6 (D), no intraretinal structures are visible except for the photoreceptor signal and RPE.
Figure 4.
 
(A) Decrease of reflectivity of several peaks depending on the filter used in jpg images. There is a linear decrease of reflectivity for every peak. Peaks with a higher initial reflectivity show a more pronounced decrease than peaks with low initial reflectivity. (B) Same analysis in LRP from raw data.
Figure 4.
 
(A) Decrease of reflectivity of several peaks depending on the filter used in jpg images. There is a linear decrease of reflectivity for every peak. Peaks with a higher initial reflectivity show a more pronounced decrease than peaks with low initial reflectivity. (B) Same analysis in LRP from raw data.
Figure 5.
 
(A) Curve representing the correction factor for every initial reflectivity in LRP from jpg images. (B) Same curve for LRP from raw data.
Figure 5.
 
(A) Curve representing the correction factor for every initial reflectivity in LRP from jpg images. (B) Same curve for LRP from raw data.
Figure 6.
 
Example of an LRP correction (A) and a whole image correction (BD). (A, dashed line) LRP from the fovea from an unfiltered OCT scan. Dotted curve: LRP from an OCT scan with an ND11 filter in the same location. Correcting the ND11 dotted curve using the correction parameters shown in Figure 4results in the solid line curve, which is equivalent to the initial curve. Differences in the two curves are induced by small motion artifacts. Although the intraretinal curve fits the original curve well, noise (region of the vitreous) shows a little higher reflectivity than the original image. For better visualization, only the part of the LRP illustrating the peaks generated by the neuroretinal layers is shown. (B) Unfiltered OCT scan. (C) OCT scan from the same retinal location in the same eye using an ND filter (OD 0.4). (D) Applying the correction results in a corrected OCT image.
Figure 6.
 
Example of an LRP correction (A) and a whole image correction (BD). (A, dashed line) LRP from the fovea from an unfiltered OCT scan. Dotted curve: LRP from an OCT scan with an ND11 filter in the same location. Correcting the ND11 dotted curve using the correction parameters shown in Figure 4results in the solid line curve, which is equivalent to the initial curve. Differences in the two curves are induced by small motion artifacts. Although the intraretinal curve fits the original curve well, noise (region of the vitreous) shows a little higher reflectivity than the original image. For better visualization, only the part of the LRP illustrating the peaks generated by the neuroretinal layers is shown. (B) Unfiltered OCT scan. (C) OCT scan from the same retinal location in the same eye using an ND filter (OD 0.4). (D) Applying the correction results in a corrected OCT image.
Figure 7.
 
Comparison of LRP from raw data (red curve) and from a jpg data (black curve). (A) Central fovea and (B) peripheral macula depict two examples of LRPs from native OCT scans of one test subject. There are only minor differences that are not statistically significant. (CF) Averaged LRPs calculated from data of all patients of the same retinal location. Average values are shown as a thick line, the 95% confidence intervals as a shaded area. Red displays data of compressed images (jpg data), and black or gray shows raw data. (C) Central fovea and (D) peripheral macula show averaged LRPs of native OCT scans calculated from data of all patients in the same retinal location. There are no statistical differences between these curves. (E, F) Same LRPs of OCT scans with a 0.4 filter from the same retinal location as demonstrated in (C, D). A decrease in reflectivity can be detected. There are only minor differences that are not significant. For better visualization, only the part of the LRP illustrating the peaks generated by the neuroretinal layers is shown.
Figure 7.
 
Comparison of LRP from raw data (red curve) and from a jpg data (black curve). (A) Central fovea and (B) peripheral macula depict two examples of LRPs from native OCT scans of one test subject. There are only minor differences that are not statistically significant. (CF) Averaged LRPs calculated from data of all patients of the same retinal location. Average values are shown as a thick line, the 95% confidence intervals as a shaded area. Red displays data of compressed images (jpg data), and black or gray shows raw data. (C) Central fovea and (D) peripheral macula show averaged LRPs of native OCT scans calculated from data of all patients in the same retinal location. There are no statistical differences between these curves. (E, F) Same LRPs of OCT scans with a 0.4 filter from the same retinal location as demonstrated in (C, D). A decrease in reflectivity can be detected. There are only minor differences that are not significant. For better visualization, only the part of the LRP illustrating the peaks generated by the neuroretinal layers is shown.
The authors thank Jeannie Wurz for her careful editing of the manuscript. 
DrexlerW, SattmannH, HermannB, et al. Enhanced visualization of macular pathology with the use of ultrahigh-resolution optical coherence tomography. Arch Ophthalmol. 2003;121(5)695–706. [CrossRef] [PubMed]
FercherAF. Optical coherence tomography. J Biomed Opt. 1996;1(2)157–173. [CrossRef] [PubMed]
FujimotoJG, BrezinskiME, TearneyGJ, et al. Optical biopsy and imaging using optical coherence tomography. Nat Med. 1995;1(9)970–972. [CrossRef] [PubMed]
HuangD, SwansonEA, LinCP, et al. Optical coherence tomography. Science. 1991;254(5035)1178–1181. [CrossRef] [PubMed]
HeeMR, PuliafitoCA, WongC, et al. Optical coherence tomography of macular holes. Ophthalmology. 1995;102(5)748–756. [CrossRef] [PubMed]
SchaudigUH, GlaefkeC, ScholzF, RichardG. Optical coherence tomography for retinal thickness measurement in diabetic patients without clinically significant macular edema. Ophthalmic Surg Lasers. 2000;31(3)182–186. [PubMed]
MumcuogluT, WollsteinG, WojtkowskiM, et al. Improved visualization of glaucomatous retinal damage using high-speed ultrahigh-resolution optical coherence tomography. Ophthalmology. 2007.Sept 18 [Epub ahead of print]
BudenzDL, FredetteMJ, FeuerWJ, AndersonDR. Reproducibility of peripapillary retinal nerve fiber thickness measurements with stratus OCT in glaucomatous eyes. Ophthalmology. 2007.Aug10 [Epub ahead of print]
BrierleyL. Preoperative assessment of cataract patients. J Cataract Refract Surg. 2006;32(1)2–3.author reply 3
BarthelmesD, SutterFK, Kurz-LevinMM, et al. Quantitative analysis of OCT characteristics in patients with achromatopsia and blue-cone monochromatism. Invest Ophthalmol Vis Sci. 2006;47(3)1161–1166. [CrossRef] [PubMed]
HuangY, CideciyanAV, PapastergiouGI, et al. Relation of optical coherence tomography to microanatomy in normal and rd chickens. Invest Ophthalmol Vis Sci. 1998;39(12)2405–2416. [PubMed]
YamadaE. Some structural features of the fovea centralis in the human retina. Arch Ophthalmol. 1969;82(2)151–159. [CrossRef] [PubMed]
IshikawaH, SteinDM, WollsteinG, BeatonS, FujimotoJG, SchumanJS. Macular segmentation with optical coherence tomography. Invest Ophthalmol Vis Sci. 2005;46(6)2012–2017. [CrossRef] [PubMed]
JacobsonSG, CideciyanAV, AlemanTS, et al. Crumbs homolog 1 (CRB1) mutations result in a thick human retina with abnormal lamination. Hum Mol Genet. 2003;12(9)1073–1078. [CrossRef] [PubMed]
MassinP, VicautE, HaouchineB, ErginayA, PaquesM, GaudricA. Reproducibility of retinal mapping using optical coherence tomography. Arch Ophthalmol. 2001;119(8)1135–1142. [CrossRef] [PubMed]
El-AshryM, AppaswamyS, DeokuleS, PagliariniS. The effect of phacoemulsification cataract surgery on the measurement of retinal nerve fiber layer thickness using optical coherence tomography. Curr Eye Res. 2006;31(5)409–413. [CrossRef] [PubMed]
Figure 1.
 
(A) Six-millimeter OCT scan through the foveola calculated from raw data. (B) Enlarged area from the scan from (A) illustrating the different retinal reflective layers from the normal perifovea. (C) Corresponding LRP calculated from the region illustrated in (B). Each peak (P1–P6) represents a highly reflective layer. For better visualization, only the part of the LRP illustrating the peaks generated by the neuroretinal layers is shown.
Figure 1.
 
(A) Six-millimeter OCT scan through the foveola calculated from raw data. (B) Enlarged area from the scan from (A) illustrating the different retinal reflective layers from the normal perifovea. (C) Corresponding LRP calculated from the region illustrated in (B). Each peak (P1–P6) represents a highly reflective layer. For better visualization, only the part of the LRP illustrating the peaks generated by the neuroretinal layers is shown.
Figure 2.
 
Filter characteristics of the ND11 filter set. Transmission (in percent, y-axis) is plotted against optical density (x-axis). Filter densities that were used in the experiments are marked by points.
Figure 2.
 
Filter characteristics of the ND11 filter set. Transmission (in percent, y-axis) is plotted against optical density (x-axis). Filter densities that were used in the experiments are marked by points.
Figure 3.
 
Grayscale OCT images illustrating the effect of ND filters on OCT image acquisition. (A) Native image. (BD) OCT images recorded in the same test subject using ND filters with various optical densities. (B) OD 0.1. (C) OD 0.4. (D) OD 0.6. Increasing the optical density of the ND filter results in stronger signal attenuation. At an OD of 0.6 (D), no intraretinal structures are visible except for the photoreceptor signal and RPE.
Figure 3.
 
Grayscale OCT images illustrating the effect of ND filters on OCT image acquisition. (A) Native image. (BD) OCT images recorded in the same test subject using ND filters with various optical densities. (B) OD 0.1. (C) OD 0.4. (D) OD 0.6. Increasing the optical density of the ND filter results in stronger signal attenuation. At an OD of 0.6 (D), no intraretinal structures are visible except for the photoreceptor signal and RPE.
Figure 4.
 
(A) Decrease of reflectivity of several peaks depending on the filter used in jpg images. There is a linear decrease of reflectivity for every peak. Peaks with a higher initial reflectivity show a more pronounced decrease than peaks with low initial reflectivity. (B) Same analysis in LRP from raw data.
Figure 4.
 
(A) Decrease of reflectivity of several peaks depending on the filter used in jpg images. There is a linear decrease of reflectivity for every peak. Peaks with a higher initial reflectivity show a more pronounced decrease than peaks with low initial reflectivity. (B) Same analysis in LRP from raw data.
Figure 5.
 
(A) Curve representing the correction factor for every initial reflectivity in LRP from jpg images. (B) Same curve for LRP from raw data.
Figure 5.
 
(A) Curve representing the correction factor for every initial reflectivity in LRP from jpg images. (B) Same curve for LRP from raw data.
Figure 6.
 
Example of an LRP correction (A) and a whole image correction (BD). (A, dashed line) LRP from the fovea from an unfiltered OCT scan. Dotted curve: LRP from an OCT scan with an ND11 filter in the same location. Correcting the ND11 dotted curve using the correction parameters shown in Figure 4results in the solid line curve, which is equivalent to the initial curve. Differences in the two curves are induced by small motion artifacts. Although the intraretinal curve fits the original curve well, noise (region of the vitreous) shows a little higher reflectivity than the original image. For better visualization, only the part of the LRP illustrating the peaks generated by the neuroretinal layers is shown. (B) Unfiltered OCT scan. (C) OCT scan from the same retinal location in the same eye using an ND filter (OD 0.4). (D) Applying the correction results in a corrected OCT image.
Figure 6.
 
Example of an LRP correction (A) and a whole image correction (BD). (A, dashed line) LRP from the fovea from an unfiltered OCT scan. Dotted curve: LRP from an OCT scan with an ND11 filter in the same location. Correcting the ND11 dotted curve using the correction parameters shown in Figure 4results in the solid line curve, which is equivalent to the initial curve. Differences in the two curves are induced by small motion artifacts. Although the intraretinal curve fits the original curve well, noise (region of the vitreous) shows a little higher reflectivity than the original image. For better visualization, only the part of the LRP illustrating the peaks generated by the neuroretinal layers is shown. (B) Unfiltered OCT scan. (C) OCT scan from the same retinal location in the same eye using an ND filter (OD 0.4). (D) Applying the correction results in a corrected OCT image.
Figure 7.
 
Comparison of LRP from raw data (red curve) and from a jpg data (black curve). (A) Central fovea and (B) peripheral macula depict two examples of LRPs from native OCT scans of one test subject. There are only minor differences that are not statistically significant. (CF) Averaged LRPs calculated from data of all patients of the same retinal location. Average values are shown as a thick line, the 95% confidence intervals as a shaded area. Red displays data of compressed images (jpg data), and black or gray shows raw data. (C) Central fovea and (D) peripheral macula show averaged LRPs of native OCT scans calculated from data of all patients in the same retinal location. There are no statistical differences between these curves. (E, F) Same LRPs of OCT scans with a 0.4 filter from the same retinal location as demonstrated in (C, D). A decrease in reflectivity can be detected. There are only minor differences that are not significant. For better visualization, only the part of the LRP illustrating the peaks generated by the neuroretinal layers is shown.
Figure 7.
 
Comparison of LRP from raw data (red curve) and from a jpg data (black curve). (A) Central fovea and (B) peripheral macula depict two examples of LRPs from native OCT scans of one test subject. There are only minor differences that are not statistically significant. (CF) Averaged LRPs calculated from data of all patients of the same retinal location. Average values are shown as a thick line, the 95% confidence intervals as a shaded area. Red displays data of compressed images (jpg data), and black or gray shows raw data. (C) Central fovea and (D) peripheral macula show averaged LRPs of native OCT scans calculated from data of all patients in the same retinal location. There are no statistical differences between these curves. (E, F) Same LRPs of OCT scans with a 0.4 filter from the same retinal location as demonstrated in (C, D). A decrease in reflectivity can be detected. There are only minor differences that are not significant. For better visualization, only the part of the LRP illustrating the peaks generated by the neuroretinal layers is shown.
×
×

This PDF is available to Subscribers Only

Sign in or purchase a subscription to access this content. ×

You must be signed into an individual account to use this feature.

×