Free
Multidisciplinary Ophthalmic Imaging  |   May 2012
Differential Optical Density of Subretinal Spaces
Author Affiliations & Notes
  • Meira Neudorfer
    From the Department of Ophthalmology, Tel Aviv-Sourasky Medical Center, Sackler Faculty of Medicine, Tel Aviv University, Tel Aviv, Israel.
  • Amit Weinberg
    From the Department of Ophthalmology, Tel Aviv-Sourasky Medical Center, Sackler Faculty of Medicine, Tel Aviv University, Tel Aviv, Israel.
  • Anat Loewenstein
    From the Department of Ophthalmology, Tel Aviv-Sourasky Medical Center, Sackler Faculty of Medicine, Tel Aviv University, Tel Aviv, Israel.
  • Adiel Barak
    From the Department of Ophthalmology, Tel Aviv-Sourasky Medical Center, Sackler Faculty of Medicine, Tel Aviv University, Tel Aviv, Israel.
  • Corresponding author: Adiel Barak, Department of Ophthalmology, Tel Aviv-Sourasky Medical Center, 6 Weizmann Street, Tel Aviv 64239, Israel; Telephone: +972-3-6873408; Fax: +972-3-6873480; adielbarak@gmail.com
Investigative Ophthalmology & Visual Science May 2012, Vol.53, 3104-3110. doi:10.1167/iovs.11-8700
  • Views
  • PDF
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Meira Neudorfer, Amit Weinberg, Anat Loewenstein, Adiel Barak; Differential Optical Density of Subretinal Spaces. Invest. Ophthalmol. Vis. Sci. 2012;53(6):3104-3110. doi: 10.1167/iovs.11-8700.

      Download citation file:


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

      ×
  • Supplements
Abstract

Purpose.: We investigated the optical density characteristics of 3subretinal spaces in neovascular age-related macular degeneration (AMD), diabetic retinopathy (DR), rhegmatogenous retinal detachment (RRD), central serous retinopathy (CSR), retinoschisis (RS), and pseudophakic cystoids macular edema (PCME).

Methods.: Patients in whom subretinal fluid (SRF) was detected by optical coherence tomography (OCT), and whose earliest OCT scans showed sufficient SRF for sampling that did not include tissue edges, were chosen for study. The highest quality B-scan containing SRF (as graded by the OCT image acquisition software) was analyzed. Optical density measurements were obtained using ImageJ, an open code Java-based image processing software.

Results.: The diagnoses of the 71 patients who met the inclusion criteria were AMD in 17, DR in 7, RRD in 18, CSR in 17, RS in 8, and PCME in 4. Optical density ratios (ODRs) were calculated as SRF OD divided by vitreous OD. ODRs were significantly higher in patients with AMD, DR, CSR, and PCME than in those with RRD and RS. No significant difference in vitreous reflectivity was detected between the former and latter patients.

Conclusions.: The finding that disease states produce significant changes in optical density ratios calls for further investigation of the possible usefulness of the parameter in differentiating between disease states, determining the outcome of various retinal diseases, and designing therapies aimed at treating the disease by correcting the abnormal density.

Introduction
Subretinal fluid (SRF) accumulates in rhegmatogenous retinal detachment (RRD), retinoschisis (RS), neovascular age-related macular degeneration (AMD), central serous retinopathy (CSR), and many other retinal diseases. The actual nature and composition of the content of the fluids in these pathologies remain unknown. 1 Optical coherence tomography (OCT), which is used widely to evaluate the anatomical and semi-histological cross-sections of the retina, can detect precisely small and large amounts of intraretinal and SRF. 2,3 Unlike subjective anatomical evaluations, it enables quantitative analysis of light reflectivity profiles from the OCT images. One of the important parameters measured by OCT is optical density (OD). OD parameters inside the retina and in accumulations of SRF have been correlated with the type of pathology causing the fluid accumulation. They also were found to be clinically relevant biomarkers in exudative macular disease, and in the response to antivascular endothelial growth factor treatment. Barthelmes et al. reported that hyporeflective spaces in the neuroretina in eyes with exudation associated with diabetic macular edema, retinitis pigmentosa, and central serous choroidopathy had higher reflectivity than the vitreous, and that the cystoid spaces in the maculae of the eyes without exudation (e.g., cone dystrophy and idiopathic perifoveal telangiectasia) had lower reflectivity than the normal vitreous. 4 Ahlers et al. demonstrated that the OD changes in SRF correlated with best corrected visual acuity changes under intra-vitreous ranibizumab therapy for neovascular AMD. 5  
In view of the findings on the relationship between the OD of SRF and various retinal diseases, we undertook a cross-sectional study in which the mean reflectivity of the SRF compartment was measured and expressed as optical density ratios (ODRs) between the SRF and the overlying vitreous located at exactly the same lateral site above the SRF. We followed an approach similar to that of Ahlers et al., who used a Cirrus Spectral domain OCT (software version, 2.0.1.3; Carl Zeiss Meditec, Inc., Dublin, CA). 5 ODRs were compared in patients with neovascular AMD, diabetic retinopathy (DR), RRD, CSR, RS, and pseudophakic cystoid macular edema (PCME). We speculated that analysis of the ODR and its variation in different ocular disease states would differentiate between pathologies in difficult diagnostic cases, and provide clues to the pathogenesis of retinal diseases. 
Methods
Patient Selection and Data Collection
Patients were chosen based on the detection of SRF accumulation on OCT examinations. Full medical records were extracted and reviewed by a senior ophthalmologist (AB, MN) for confirmed diagnosis of neovascular AMD, DR, RRD, CSR, retinal detachment associated with RS, or PCME (Fig. 1A). Exclusion criteria were coexisting ophthalmic pathology, ophthalmic treatments, surgery, or intravitreal injections before the earliest available OCT scan, and any medical condition that might affect ODR results. Demographic data (sex, age, and date of first visit/referral to the clinic) were taken from the patients' medical files. 
Figure 1. 
 
(A) Pathologies studied. (B) Choice of ROIs.
Figure 1. 
 
(A) Pathologies studied. (B) Choice of ROIs.
OCT Scans
The patient's earliest OCT exam (SPECTRALIS, SD-OCT; Heidelberg Engineering, Carlsbad, CA) exhibiting SRF was chosen for subsequent analysis. The OCT image acquisition software scores the quality of each image based on the signal-to-noise ratio, 6 enabling us to select the highest quality B-scan containing SRF. This feature provided an objective and reproducible selection process for the B-scan, and minimized the amount of noise in our calculations. The single B-scan that had the highest image quality score and contained a sufficient volume of SRF to be sampled without including edges was chosen. The OCT scans were exported from the OCT acquisition software as grayscale, compression-free, quality-preserving JPEG images. Image quality and acquisition mode (HR/HS) were recorded. 
The study was approved by the institutional ethics committee, and its protocol complied with the standards of the Declaration of Helsinki. All OCT examinations were performed at the Department of Ophthalmology, Tel Aviv-Sourasky Medical Center, Israel. 
ODR Measurements
OD measurements were obtained using ImageJ 7 (developed by Wayne Rasband, National Institutes of Health, Bethesda, MD; available at http://rsb.info.nih.gov/ij/index.html), an open code Java-based image processing software. Two identical regions of interest (ROIs) of identical shape were chosen (“ROI selection method”), one for the SRF compartment and the other for the vitreous space (Fig. 1B). ROIs were chosen on the same vertical line to avoid errors associated with refraction nonhomogeneities of the various structures (corneal opacifications, cataract, vitreous floaters, or other causes of nonhomogeneous signal intensity at the retinal level). Vertical coordinates of the two ROIs were recorded to take into account the attenuation of light intensity due to passage through the tissue. Quadrangular-shaped ROIs were chosen in a manner that avoids tissue-fluid interfaces. 
In addition to the ROI selection method described above, a second set of OD measurements was performed, this time with the entire SRF and vitreous selected (“entire region selection method”). This method aimed to decrease the subjective component of ROI selection, and offered the possibility to compare the two methods, and determine whether ROI selection is a reasonable alternative to entire region OD measurements. 
A clear advantage of using the vitreous space as a baseline medium for comparing reflectivity profiles is its distinct borders, which enable accurate identification of its contours by different observers. A potential drawback is the noise suppression algorithm applied by the OCT acquisition device to provide a “cleaner” image, which cuts down the lower reflectivity values. We attempted to overcome this drawback by testing two baseline bright media—the retinal pigment epithelium (RPE) and the retinal nerve fiber layer (RNFL), both through ROI OD measurements—in the hopes of reproducing the results obtained by the ODR. ODs were extracted from the measured gray level intensity of the corresponding ROI/entire region selection in the SRF compartment and vitreous space, RPE, and RNFL on a scale of 0 (pure black) to 255 (pure white). The resulting measurements were exported to a database program (Excel; Microsoft, Redmond, WA). Reflectivity ratios consisted of four parameters calculated from the measured ODs: 
1. O D ( S R F R O I ) / O D ( V I T R E O U S R O I ) = O D R 2. O D ( S R F T O T A L ) / O D ( V I T R E O U S T O T A L ) = S- V I T T O T A L 3. O D ( S R F T O T A L ) / O D ( R P E R O I ) = S- R P E 4. O D ( S R F T O T A L ) / O D ( R N F L R O I ) = S- R N F L
 
One case lacked a clearly discernible RPE and was excluded from the S-RPE calculation. Nine cases were excluded from the S-RNFL calculation due to lack of clearly discernible RNFL. Entire region SRF OD measurements were favored over ROI OD measurements in the calculation of S-RPE and S-RNFL because of their theoretical advantage of being less prone to observer variability. 
Statistical Analysis
Data were analyzed by statistical analysis software (SPSS for Windows, version 17; SPSS Inc., Chicago, IL). Significance was defined as a type alpha error probability <0.05. Natural logarithm of all ODs values and of the four reflectivity profiles was used for the subsequent statistical analysis due to their normal distribution and equality of variances between the groups of pathologies studied (i.e., agreement with the prerequisites for using the F-statistics in the ANOVA model and the t-test, see Supplementary Material). Thus, all ODs and reflectivity ratios were used henceforth in their natural logarithmic form unless stated otherwise. 
Paired t-tests were used to compare between the ROI selection method, and the entire region selection method in the SRF and compartment vitreous space. Pearson correlation coefficients (R) for the results of the two measurement methods in the SRF compartment and the vitreous space were calculated as well. An ANOVA model was generated to test for significant differences in ln(ODR) and ln(S-VIT TOTAL ) values between the different pathologies studied. The post-hoc Tukey B test was used to search for significant differences in ln(ODR) and ln(S-VIT TOTAL ) values between groups of pathologies. Independent sample t-tests were executed to exclude significant differences in the baseline media (vitreous, RPE, and RNFL) ODs, which might account for the observed differences in ln(ODR) ln(S-VIT TOTAL ) values between the groups of pathologies found on the post-hoc analysis. Reflectivity ratios in addition to ODR and S-VIT TOTAL (i.e., S-RPE and S-RNFL) were calculated and tested as to whether their respective natural logarithms reproduce the same significant differences between groups of pathologies as the ln(ODR) and ln(S-VIT TOTAL ) shown using independent sample t-tests. Observer agreement for ROI selection in the SRF compartment, vitreous space, RPE, and RNFL was tested with three different observers. Observer agreement is important especially for the RPE and RNFL ODs, since these media are measured using only the ROI selection method. 
A multiple linear regression then was used to take account of the effect certain confounders (i.e., age, image quality, and vertical distance between the SRF ROI and the vitreous ROI, whose respective natural logarithms were correlated linearly with ln[ODR] values; as well as age and image quality, whose respective natural logarithms were correlated linearly with ln[S-VIT TOTAL ] values) might have had on ln(ODR) and ln(S-VIT TOTAL ) measurements results. Significant differences were sought again between the groups of pathologies already found on the post-hoc analysis through the use of intermediate variable in the multiple linear regression. 
Results
The study population of 71 patients consisted of 32 males and 39 females, mean age 59.82 ± 19.947 years. Of the patients 17 were diagnosed as having AMD, 7 DR, 18 RRD, 17 CSR, 8 RS, and 4 PCME. The sample means and SD for the other variables in our analysis were as follows: image quality 26.93 ± 4.615 (on a scale from 0–40), vertical distance between SRF and vitreous ROIs 537.48 ± 133.206 μm, and time interval since the first visit/date of referral to the clinic 1363.57 ± 970.459 days. Table 1 summarizes these parameters according to diagnoses. A particularly vexing problem is the possible association of the baseline medium (vitreous, RPE, and RNFL) OD with the age of the patient population. Scatters plots are shown in Figure 2 for the OD of the baseline media used in this analysis versus the age of the study population. Pearson correlation coefficients for the various baseline media ODs versus the age of the study population are shown to be weak and insignificant (Table 2). Table 3 summarizes the mean reflectivity ratios and SDs for each of the pathologies. This section presents final results of the statistical analysis and corresponding P values. Detailed results of the statistical tests appear in the Supplementary Material
Table 1.  
 
Means and SD for the Different Parameters Studied (Age, Image Quality, Vertical Distance between the SRF ROI and the Vitreous ROI, and Duration of Follow-Up in the Clinic) according to Diagnosis
Table 1.  
 
Means and SD for the Different Parameters Studied (Age, Image Quality, Vertical Distance between the SRF ROI and the Vitreous ROI, and Duration of Follow-Up in the Clinic) according to Diagnosis
Diagnosis Statistics Age Image Quality Vertical Distance between ROIs (μ) Follow-up
AMD Mean 81.69 26.25 478.92 1665.44
N 16 16 16 16
SD 9.631 4.139 64.165 1035.256
DR Mean 58.29 27.29 545.66 2258.00
N 7 7 7 7
SD 13.400 4.030 149.182 1579.355
RRD Mean 55.72 25.94 603.27 1008.00
N 18 18 18 14
SD 18.745 4.318 174.599 413.773
CSR Mean 49.47 29.18 479.35 1204.00
N 17 17 17 15
SD 14.371 3.206 89.796 893.874
RS Mean 44.56 24.89 565.14 827.83
N 9 9 9 6
SD 19.191 7.474 123.037 162.130
PCME Mean 71.75 28.50 646.08 1195.33
N 4 4 4 3
SD 15.305 4.041 105.973 672.533
Total Mean 59.82 26.93 537.48 1363.57
N 71 71 71 61
SD 19.947 4.615 133.206 970.459
Figure 2. 
 
Scatter plots of the different baseline media versus age, vitreous OD measured in the ROI selection method (A), vitreous OD measured in the entire region selection method (B), RPE OD (C), and RNFL OD (D).
Figure 2. 
 
Scatter plots of the different baseline media versus age, vitreous OD measured in the ROI selection method (A), vitreous OD measured in the entire region selection method (B), RPE OD (C), and RNFL OD (D).
Table 2.  
 
Pearson Correlation Coefficients for the Various Baseline Media OD Used in the Analysis versus Age of the Study Population
Table 2.  
 
Pearson Correlation Coefficients for the Various Baseline Media OD Used in the Analysis versus Age of the Study Population
Vitreous_ROI OD Vitreous_TOTAL OD RPE OD RNFL OD
Age Pearson correlation −0.016 −0.087 0.133 −0.107
Sig. (2-tailed) 0.892 0.473 0.272 0.406
N 71 71 70 62
Table 3.  
 
Descriptive Statistics for the Different Reflectivity Ratios Studied according to Diagnosis
Table 3.  
 
Descriptive Statistics for the Different Reflectivity Ratios Studied according to Diagnosis
Reflectivity Ratio Diagnosis N Mean SD SE 95% Confidence Interval for Mean Minimum Maximum
Lower Bound Upper Bound
Ln(ODR) AMD 16 0.2932 0.44846 0.11212 0.0542 0.5321 −0.41 1.30
DR 7 0.3380 0.38076 0.14391 −0.0141 0.6902 −0.35 0.85
RRD 18 −0.4705 0.31993 0.07541 −0.6296 −0.3114 −1.06 0.07
CSR 17 0.2856 0.40867 0.09912 0.0755 0.4957 −0.66 0.88
RS 9 −0.3858 0.23911 0.07970 −0.5696 −0.2020 −0.83 −0.09
PCME 4 0.3130 0.15181 0.07591 0.0715 0.5546 0.19 0.51
Total 71 0.0172 0.50826 0.06032 −0.1031 0.1375 −1.06 1.30
Ln(S-VITTOTAL) AMD 16 0.0722 0.34144 0.08536 −0.1098 0.2541 −0.56 0.70
DR 7 0.1827 0.49827 0.18833 −0.2781 0.6435 −0.79 0.60
RRD 18 −0.6765 0.31244 0.07364 −0.8319 −0.5211 −1.11 −0.10
CSR 17 0.0505 0.37423 0.09076 −0.1419 0.2429 −0.85 0.51
RS 9 −0.6472 0.25563 0.08521 −0.8437 −0.4507 −1.24 −0.39
PCME 4 0.1601 0.40918 0.20459 −0.4910 0.8112 −0.35 0.63
Total 71 −0.1982 0.50492 0.05992 −0.3177 −0.0787 −1.24 0.70
Ln(S-RPE) AMD 16 −2.0400 0.30485 0.07621 −2.2024 −1.8775 −2.58 −1.49
DR 7 −1.8778 0.18721 0.07076 −2.0510 −1.7047 −1.99 −1.47
RRD 18 −2.4566 0.61843 0.14577 −2.7641 −2.1490 −3.94 −1.60
CSR 17 −2.3865 0.45667 0.11076 −2.6213 −2.1517 −3.21 −1.70
RS 8 −2.4152 0.86336 0.30524 −3.1370 −1.6934 −3.80 −1.40
PCME 4 −2.0866 0.37156 0.18578 −2.6778 −1.4953 −2.61 −1.74
Total 70 −2.2606 0.54043 0.06459 −2.3894 −2.1317 −3.94 −1.40
Ln(S-RNFL) AMD 15 −2.2470 0.32240 0.08324 −2.4255 −2.0685 −2.83 −1.80
DR 7 −2.0368 0.28013 0.10588 −2.2959 −1.7777 −2.43 −1.54
RRD 16 −2.9832 0.51951 0.12988 −3.2600 −2.7064 −3.93 −2.16
CSR 17 −2.5562 0.53195 0.12902 −2.8297 −2.2827 −3.38 −1.69
RS 3 −3.2830 0.47467 0.27405 −4.4622 −2.1039 −3.83 −2.99
PCME 4 −2.2915 0.33326 0.16663 −2.8218 −1.7612 −2.77 −2.00
Total 62 −2.5510 0.56379 0.07160 −2.6942 −2.4079 −3.93 −1.54
A point of particular interest is the direct comparison of the two methods of measurements used for obtaining ln(OD) means in the SRF compartment and the vitreous space: ROI selection method versus entire region selection method. Obviously, entire region selection averages over larger image area (as can be seen in Table 4) and, thus, is less susceptible to local variations in OD. In addition, the contour of the entire region is readily identifiable and theoretically less prone to observer variability compared to ROI selection. However, in the ROI selection method, placing the ROIs in the exact same horizontal coordinates corrects for possible refraction nonhomogeneities. Drawing ROIs also is less time consuming than traversing the entire region contour. The lack of significance of the differences in SRF mean ln(OD) values between the ROI and the entire region selection methods was demonstrated by the paired t-test (P = 0.233). However, the vitreous ROI mean ln(OD) was significantly different from the vitreous entire region mean ln(OD) (P = 0.001), throwing into question the ability of vitreous ROI to represent truly the vitreous entire region characteristics. The significance of the difference in measured vitreous ln(OD) means between the two selection methods, as opposed to the nonsignificance of the difference in measured SRF ln(OD) means between the two measurement methods, might be due to the vitreous' larger size compared to the SRF compartment. This would mean that when identical size ROIs are chosen in the vitreous space and the SRF compartment, the relative portion the vitreous ROI occupies in the entire vitreous is smaller than the relative portion the identical size SRF ROI occupies in the SRF compartment, making the former less representative of the entire vitreous. Pearson correlations between ln(OD) values obtained by the entire region selection method and the ROI selection method were 0.875 and 0.943 for the vitreous and SRF, respectively (P < 0.001 for both). 
Table 4.  
 
Areas (Measured in Pixels) Selected for the SRF and Vitreous OD Measurements in the ROI Selection and Entire Region Selection Methods
Table 4.  
 
Areas (Measured in Pixels) Selected for the SRF and Vitreous OD Measurements in the ROI Selection and Entire Region Selection Methods
Selection N Minimum Maximum Mean SD
Vitreous ROI area 71 105 2484 826.00 522.512
Vitreous total area 71 684 332,829 110,944.52 57,960.991
SRF ROI area 71 105 2484 826.94 522.532
SRF total area 71 210 221,304 21,569.11 42,980.305
Observer agreement in ROI selection was tested by calculating Pearson correlation coefficients for the results of three different observers' measurements in the SRF compartment, that is the vitreous space, RPE and RNFL. The averages of the Pearson correlation coefficients were 0.949 (all P < 0.01) for the SRF compartment ROI measurements, 0.903 (all P < 0.05) for vitreous space ROI OD measurements, 0.077 (none significant) for RPE ROI OD measurements, and 0.926 (all P < 0.05) for RNFL ROI OD measurements. 
The ANOVA model revealed significant variation in ln(ODR) and ln(S-VIT TOTAL ) between the different pathologies studied (ln[ODR]: P < 0.001, ln[S-VIT TOTAL ]: P < 0.001). Post-hoc analysis segmented the population study into two classes of pathologies based on ln(ODR) and ln(S-VIT TOTAL ) values (the same division for both ln[ODR] and ln[S-VIT TOTAL ] in the post-hoc tests). Class A consisted of AMD, DR, CSR, and PCME. Class B consisted of RRD and RD. Figure 3 shows confidence intervals for ln(ODR) values according to diagnosis (a clear demarcation is seen to exist between ln[ODR] values of the two classes, see Supplementary Material, for more details of the ANOVA and post-hoc analysis). Obviously, we would not expect any significant differences in the vitreous fluid reflectivity or other baseline medium reflectivity (RPE, RNFL) between the two classes found on the post-hoc analysis; otherwise the observed difference might be attributed to differences in the reflectivity of the baseline medium itself. To make certain that the aforementioned differences in ln(ODR) and ln(S-VIT TOTAL ) values were not simply the result of differences in vitreous fluid reflectivity, an independent samples t-test was conducted comparing vitreous reflectivity values between the classes in the ROI (P = 0.15) and the entire region (P = 0.216) selection method measurements. A similar comparison between the RNFL ln(OD) measurement in the two classes also yielded no significant differences (P = 0.964). However, significant differences were found between the RPE ln(OD) in the two classes (P < 0.001), precluding its use as an alternative baseline for the vitreous space (for more details of the various t-tests see Supplementary Material, Table 2
Figure 3. 
 
Confidence intervals for ln(ODR) values according to diagnosis.
Figure 3. 
 
Confidence intervals for ln(ODR) values according to diagnosis.
Our expectation to calculate the ln(S-RNFL) and reproduce the significant differences observed between the two classes found that the post-hoc analysis done for ln(ODR) and ln(S-VIT TOTAL ) did, indeed, yield significant differences in ln(S-RNFL) between the two classes (P < 0.001), supporting the diagnostic role of the reflectivity ratios. Possible confounding effects were examined by a multiple linear regression model with ln(ODR) values corrected for the natural logarithms of the independent parameters listed above (age, image quality, and vertical distance between the ROIs). These parameters, taken as natural logarithms, had significant linear correlations with ODR, that is ln(age): R = 0.329, P = 0.005; ln(image quality): R = 0.242, P = 0.042, and ln(vertical distance between the ROIs): R = 0.274, P = 0.021. After correcting for the influences of these confounding variables on ln(ODR), a significant difference persisted between the classes (for details of the linear regression see Supplementary Material). 
The same procedure was repeated for ln(S-VIT TOTAL ) and ln(S-RNFL), with the independent parameters this time being age and image quality for ln(S-VIT TOTAL ), and age for ln(S-RNFL) (vertical distance between ROIs is irrelevant since SRF OD was measured in the entire region). Pearson correlation coefficients were R = 0.351, P = 0.003 for ln(S-VIT TOTAL ) – ln(age); R = 0.257, P = 0.03 for ln(image quality), and R = 0.466, P < 0.001 for ln(S-RNFL) and ln(age). Again, after correcting for these confounding influences, a significant difference persisted in ln(S-VIT TOTAL ) and ln(S-RNFL) values between the two classes. 
Discussion
This study on OD characteristics of subretinal spaces in six retinal pathologies revealed a significantly lower ln(ODR) and ln(S-VIT TOTAL ) among patients with RD and RS compared to patients with AMD, DR, CSR, and PCME. No significant difference in vitreous ln(OD) was found between the two classes in the entire region or ROI selection method. Although the OD as measured on OCT does not give the composition of the SRF, a higher OD signal is an indication of increased refractivity of the tissue caused by increased density of particles (mostly proteins) inside the tissue or fluids being scanned, whereas a low OD signal reflects a relatively clear fluid composition with fewer particles. 8,9  
To our knowledge, and based on a careful search of PubMed.com (1988–2011, http://www.ncbi.nlm.nih.gov/pubmed), this is the first attempt to describe the ODs of the SRF resulting from retinal pathologies other than RD and RS. What we know about the composition of SRF comes from analysis of subretinal composition in cases of RRD and RS. The lack of data on the composition of SRF in the other conditions probably is due to the inherently small volume of SRF, and the lack of means for collecting them. These obstacles are overcome by high-resolution spectral domain OCT, a technology that offers the ability to examine even a small volume of SRF, and to compare its properties between disease states. This is not the case in RRD, where a large volume of fluid accumulates beneath the retina and is easy to collect for examination during surgery. The SRF in RRD is composed mostly of proteins and lipids. The most abundant protein is albumin, with the protein concentration varying as a function of the duration of the RD. 10 SRF lipid concentration in RRD varies from 0.1 to 2.4 mg/mL, 11 and the main lipid components of SRF on chromatography are phospholipids, esterified cholesterol, and triglycerides. There are very few fatty acids. The source of the high lipid concentration is thought to be the destruction of the photoreceptor cells. Lipids are the main components of photoreceptor external segments, and those cells are degraded during RRD, 12 and release lipid component to the SRF. 
The origin of SRF long has been a matter of debate. In RRD, they are now believed to originate from three sources: the vitreous, 13 serum, 14 and retina. 15 To our knowledge, the origin of SRF in other retinal conditions has not been investigated to date. The presence of proteins and acute phase reactants in SRF of eyes with pathologies, like AMD and DR, probably is secondary to breakdown of the blood-retinal barrier. The SRF in RD and RS has no exudative or inflammatory components. 
Our findings of different disease etiologies in different ODRs are interesting, but their clinical relevance still is unknown. Our current study did not look into the possibility of whether changes in OD within specific disease states correlate with prognosis, as was done with neovascular AMD in an earlier study by Ahlers et al. 5 More data on the development and clinical course of diseases and changes in OD are needed to determine the relationship between altered OD and prognosis. 
The only difference in OD that emerged from our study was between two classes of retinal pathologies (RD and RS versus AMD, DR, CSR, and PCME), not within the individual disease classes. While Ahlers et al. showed highly significant differences in OD measurements between CSR and AMD, 5 we can speculate only as to why we detected no such differences in our patients. One reason might be the different OCT instruments used: the Cirrus spectral domain OCT in the study of Ahlers et al. 5 versus the SPECTRALIS spectral domain OCT in our study. 
Our study has clear advantages over previous works in this field. The use of an alternative bright light baseline medium, RPE or RNFL, overcame the problem that the very low signal is cut off when a mapping function is used to suppress noise and produce a clearer image. Without knowing the exact algorithm used for these ends in advance, it is impossible to know its possible effect on the ODR values, and whether it actually will have a consistent effect from one case to the next. RPE was rejected due to its being observer-dependent and significantly different between the two classes, making it unsuitable as a baseline medium. The RNFL, on the other hand, lacks these drawbacks and is a valid baseline medium. The significant differences reproduced in ln(S-RNFL) between the two classes found in the post-hoc analysis strengthen our confidence in the use of reflectivity ratios as diagnostic markers. 
Other advantages of our study include its taking into account the influence of the vertical distance between the SRF ROI and the vitreous ROI by attenuation of light intensity as it passes through the tissue; the use of multiple regression analysis to correct for other confounders (age, image quality), with persistence of significant differences in ln(ODR) and ln(S-VIT TOTAL ) values between the classes found on the post-hoc analysis; and the ability to compare directly the entire region measurement method with the ROI selection method. A drawback of the study is the small size of the sample for each retinal pathology that precludes drawing significant conclusions about differences between the individual pathologies within the classes found on the post-hoc analysis (e.g., significant differences between AMD, DR, CSR, and PCME within Class A). 
In conclusion, the finding of significant differences in OD in different retinal disease states warrants further investigation into the usefulness of OD for differential diagnosis between retinal pathologies, and the possibility of its assisting in predicting their outcome. 
Supplementary Materials
References
Quintyn JC Brasseur G . Subretinal fluid in primary rhegmatogenous retinal detachment: physiopathology and composition`. Surv Ophthalmol . 2004;49:96–108. [CrossRef] [PubMed]
Ko TH Fujimoto JG Schuman JS Comparison of ultrahigh- and standard-resolution optical coherence tomography for imaging macular pathology`. Ophthalmology . 2005;112:1922.e1–1922.e15. [CrossRef]
Srinivasan VJ Wojtkowski M Witkin AJ High-definition and 3-dimensional imaging of macular pathologies with high-speed ultrahigh-resolution optical coherence tomography`. Ophthalmology . 2006;113:2054.e1–2054.e14. [CrossRef]
Barthelmes D Sutter FK Gillies MC . Differential optical densities of intraretinal spaces`. Invest Ophthalmol Vis Sci . 2008;49:3529–3534. [CrossRef] [PubMed]
Ahlers C Golbaz I Einwallner E Identification of optical density ratios in subretinal fluid as a clinically relevant biomarker in exudative macular disease`. Invest Ophthalmol Vis Sci . 2009;50:3417–3424. [CrossRef] [PubMed]
Spectralis Quickguide, Software Version 5.0. Heidelberg, Germany;Heidelberg Engineering GmbH; 2008. 2008:7 11 Available at: http://www.heidelbergengineering.com/wp-content/uploads/Spectralis_QuickGuide.pdf. Accessed April 28, 2012
Abramoff MD Magelhaes PJ Ram SJ . Image processing with ImageJ`. Biophotonics Int . 2004;7:36–42.
Dubois A Vabre L Boccara A Beaurepaire E . High-resolution full-field optical coherence tomography with a Linnik microscope`. Appl Optics . 2002;41:805–812. [CrossRef]
Nakano A . Studies on the subretinal fluid. Report I. Refractometric and paper-electrophoretic study on the subretinal fluid of spontaneous retinal detachment. Jpn J Ophthalmol . 1961;3:23–28.
Berrod JP Kayl P Rozot P Raspiller A . Proteins in the subretinal fluid`. Eur J Ophthalmol . 1993;3:132–137. [PubMed]
Feman SS Andrews JS James M Leonard-Martin T . Human biochemical components in retinoschisis and detachment`. Arch Ophthalmol . 1981;99:475–477. [CrossRef] [PubMed]
Machemer R . Experimental retinal detachment in the owl monkey. II. Histology of retina and pigment epithelium. Am J Ophthalmol . 1968;66:396–410. [CrossRef] [PubMed]
Cooper WC Halbert SP Manski WJ . Immunochemical analysis of vitreous and subretinal fluid`. Invest Ophthalmol . 1963;2:369–377. [PubMed]
Dorello U . Electrophoretic studies of the protein content of the subretinal fluid in idiopathic retinal detachment`. Am J Ophthalmol . 1956;41:564.
Hara S Ishiguro S Hayasaka S Mizuno K . Immunoreactive opsin content in subretinal fluid from patients with rhegmatogenous retinal detachments`. Arch Ophthalmol . 1987;105:260–263. [CrossRef] [PubMed]
Footnotes
2  These authors contributed equally to the work presented here and should therefore be regarded as equivalent authors.
Footnotes
 Disclosure: M. Neudorfer, None; A. Weinberg, None; A. Loewenstein, None; A. Barak, None
Figure 1. 
 
(A) Pathologies studied. (B) Choice of ROIs.
Figure 1. 
 
(A) Pathologies studied. (B) Choice of ROIs.
Figure 2. 
 
Scatter plots of the different baseline media versus age, vitreous OD measured in the ROI selection method (A), vitreous OD measured in the entire region selection method (B), RPE OD (C), and RNFL OD (D).
Figure 2. 
 
Scatter plots of the different baseline media versus age, vitreous OD measured in the ROI selection method (A), vitreous OD measured in the entire region selection method (B), RPE OD (C), and RNFL OD (D).
Figure 3. 
 
Confidence intervals for ln(ODR) values according to diagnosis.
Figure 3. 
 
Confidence intervals for ln(ODR) values according to diagnosis.
Table 1.  
 
Means and SD for the Different Parameters Studied (Age, Image Quality, Vertical Distance between the SRF ROI and the Vitreous ROI, and Duration of Follow-Up in the Clinic) according to Diagnosis
Table 1.  
 
Means and SD for the Different Parameters Studied (Age, Image Quality, Vertical Distance between the SRF ROI and the Vitreous ROI, and Duration of Follow-Up in the Clinic) according to Diagnosis
Diagnosis Statistics Age Image Quality Vertical Distance between ROIs (μ) Follow-up
AMD Mean 81.69 26.25 478.92 1665.44
N 16 16 16 16
SD 9.631 4.139 64.165 1035.256
DR Mean 58.29 27.29 545.66 2258.00
N 7 7 7 7
SD 13.400 4.030 149.182 1579.355
RRD Mean 55.72 25.94 603.27 1008.00
N 18 18 18 14
SD 18.745 4.318 174.599 413.773
CSR Mean 49.47 29.18 479.35 1204.00
N 17 17 17 15
SD 14.371 3.206 89.796 893.874
RS Mean 44.56 24.89 565.14 827.83
N 9 9 9 6
SD 19.191 7.474 123.037 162.130
PCME Mean 71.75 28.50 646.08 1195.33
N 4 4 4 3
SD 15.305 4.041 105.973 672.533
Total Mean 59.82 26.93 537.48 1363.57
N 71 71 71 61
SD 19.947 4.615 133.206 970.459
Table 2.  
 
Pearson Correlation Coefficients for the Various Baseline Media OD Used in the Analysis versus Age of the Study Population
Table 2.  
 
Pearson Correlation Coefficients for the Various Baseline Media OD Used in the Analysis versus Age of the Study Population
Vitreous_ROI OD Vitreous_TOTAL OD RPE OD RNFL OD
Age Pearson correlation −0.016 −0.087 0.133 −0.107
Sig. (2-tailed) 0.892 0.473 0.272 0.406
N 71 71 70 62
Table 3.  
 
Descriptive Statistics for the Different Reflectivity Ratios Studied according to Diagnosis
Table 3.  
 
Descriptive Statistics for the Different Reflectivity Ratios Studied according to Diagnosis
Reflectivity Ratio Diagnosis N Mean SD SE 95% Confidence Interval for Mean Minimum Maximum
Lower Bound Upper Bound
Ln(ODR) AMD 16 0.2932 0.44846 0.11212 0.0542 0.5321 −0.41 1.30
DR 7 0.3380 0.38076 0.14391 −0.0141 0.6902 −0.35 0.85
RRD 18 −0.4705 0.31993 0.07541 −0.6296 −0.3114 −1.06 0.07
CSR 17 0.2856 0.40867 0.09912 0.0755 0.4957 −0.66 0.88
RS 9 −0.3858 0.23911 0.07970 −0.5696 −0.2020 −0.83 −0.09
PCME 4 0.3130 0.15181 0.07591 0.0715 0.5546 0.19 0.51
Total 71 0.0172 0.50826 0.06032 −0.1031 0.1375 −1.06 1.30
Ln(S-VITTOTAL) AMD 16 0.0722 0.34144 0.08536 −0.1098 0.2541 −0.56 0.70
DR 7 0.1827 0.49827 0.18833 −0.2781 0.6435 −0.79 0.60
RRD 18 −0.6765 0.31244 0.07364 −0.8319 −0.5211 −1.11 −0.10
CSR 17 0.0505 0.37423 0.09076 −0.1419 0.2429 −0.85 0.51
RS 9 −0.6472 0.25563 0.08521 −0.8437 −0.4507 −1.24 −0.39
PCME 4 0.1601 0.40918 0.20459 −0.4910 0.8112 −0.35 0.63
Total 71 −0.1982 0.50492 0.05992 −0.3177 −0.0787 −1.24 0.70
Ln(S-RPE) AMD 16 −2.0400 0.30485 0.07621 −2.2024 −1.8775 −2.58 −1.49
DR 7 −1.8778 0.18721 0.07076 −2.0510 −1.7047 −1.99 −1.47
RRD 18 −2.4566 0.61843 0.14577 −2.7641 −2.1490 −3.94 −1.60
CSR 17 −2.3865 0.45667 0.11076 −2.6213 −2.1517 −3.21 −1.70
RS 8 −2.4152 0.86336 0.30524 −3.1370 −1.6934 −3.80 −1.40
PCME 4 −2.0866 0.37156 0.18578 −2.6778 −1.4953 −2.61 −1.74
Total 70 −2.2606 0.54043 0.06459 −2.3894 −2.1317 −3.94 −1.40
Ln(S-RNFL) AMD 15 −2.2470 0.32240 0.08324 −2.4255 −2.0685 −2.83 −1.80
DR 7 −2.0368 0.28013 0.10588 −2.2959 −1.7777 −2.43 −1.54
RRD 16 −2.9832 0.51951 0.12988 −3.2600 −2.7064 −3.93 −2.16
CSR 17 −2.5562 0.53195 0.12902 −2.8297 −2.2827 −3.38 −1.69
RS 3 −3.2830 0.47467 0.27405 −4.4622 −2.1039 −3.83 −2.99
PCME 4 −2.2915 0.33326 0.16663 −2.8218 −1.7612 −2.77 −2.00
Total 62 −2.5510 0.56379 0.07160 −2.6942 −2.4079 −3.93 −1.54
Table 4.  
 
Areas (Measured in Pixels) Selected for the SRF and Vitreous OD Measurements in the ROI Selection and Entire Region Selection Methods
Table 4.  
 
Areas (Measured in Pixels) Selected for the SRF and Vitreous OD Measurements in the ROI Selection and Entire Region Selection Methods
Selection N Minimum Maximum Mean SD
Vitreous ROI area 71 105 2484 826.00 522.512
Vitreous total area 71 684 332,829 110,944.52 57,960.991
SRF ROI area 71 105 2484 826.94 522.532
SRF total area 71 210 221,304 21,569.11 42,980.305
×
×

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.

×