IMAGING TECHNIQUES
The first stereoscopic fundus examination with a plano-convex lens and the slit-lamp biomicroscope was performed by El Bayadi2 in 1953, and the technique gained widespread acceptance in the 1980s with the use of double aspheric lenses. During that decade, digital imaging was an emerging technology, so the most convenient method by which to document changes over time in the optic nerve's appearance was conventional, silver halide-based film photography.
Photographic films are coated with small crystals of light-sensitive, silver halide salts suspended in a transparent gelatin (an emulsion), which is coated on plastic film. When exposed to light, some of the silver halide crystals transform into metallic silver particles. During development and fixation, the metallic particles remain in the gelatin, and the unexposed crystals are removed. Black-and-white photographic film usually has one layer of silver salts. Color films use several emulsion formulations, and more chemical processing steps are needed to leave dyes that represent the color content of the exposed image.
Clinicians routinely image patients' optic nerve heads with 35-mm film in fundus cameras, which create a single monoscopic image or two stereoscopic images, either in black and white or in color. The physician obtains the two stereoscopic fundus images simultaneously with prism-based optics3 (Figure 1) or sequentially in the following manner. The photographer customarily takes the first image as far to the left of the dilated pupil as possible. He obtains the second image with an Allen stereo separator or by moving the camera a minimum of 2 mm to the right to achieve between 15º and 30º of separation between the two images. For the second image, it is permissible to reduce the focus and clarity in order to maintain the stereoscopic effect.4
Figure 1. The practitioner obtains stereophotographs of a glaucomatous optic nerve simultaneously using a prism-based optical system.
ASSESSING OBSERVERS' RELIABILITY
Methods
Reliability is defined as the degree to which errors in measurement are absent from the data. To assess observers as sources of error, one quantifies the data's consistency primarily through two approaches. Within-observer reliability, or intraobserver agreement, relates to how consistent the observer's repeated measurements were and the degree of measuring error in the observational data. Between-observer reliability, or interobserver agreement, refers to the degree to which observers are interchangeable.
The metrics commonly used to assess observational reliability include the Kappa statistic and the intraclass correlation coefficient (ICC). Kappa is directly related to the proportion of agreement achieved by raters beyond chance agreement, and it can be weighted to correct for chance agreement. Kappa values may range from -1.00 to 1.00. Kappa values greater than zero indicate that raters agree more frequently than predicted by chance, whereas values below zero indicate agreement less frequently than predicted by chance. Kappa values of less than 0.60 represent unacceptable reliability, and values greater than 0.80 represent excellent reliability.
Researchers use the ICC less commonly than Kappa. The ICC is related to the proportion of total variance in observations that is attributable to true variation. ICC values range from 0 to 1.00, with 1.00 indicating perfect observer agreement when there is some degree of variation.
Research Findings
Evaluating photographs of the optic nerve typically involves rating the horizontal and vertical cup-to-disc ratios and the neuroretinal rim. A number of studies have compared different imaging techniques and measured the intra- and interobserver agreement among raters of optic disc photographs. Lichter5 asked 16 glaucoma experts to assess 20 eyes of 10 patients. The results showed that the stereoscopic evaluation of cup-to-disc ratios yielded better intra- and interobserver agreement than monoscopic evaluation. A somewhat disquieting finding of the study, however, was the significant interobserver variability among the glaucoma specialists in their evaluation of cup-to-disc ratios.
Caprioli et al6 showed that the detection of structural changes through the qualitative evaluation of sequential stereoscopic optic disc photographs was more accurate than the qualitative evaluation of the nerve fiber layer and manual stereoplanimetry of the disc rim's area. In a study of three eyes, Rosenthal et al7 showed that simultaneous fixed-angle stereophotography was more reproducible than sequential stereophotography.
Tielsch et al8 reported strong intraobserver agreement (K = 0.71) and poor interobserver agreement (K = 0.58) between two expert examiners who evaluated the neuroretinal rims of 200 eyes. Similarly, studies by Abrams et al9 and Varma et al10 as well as one involving the European Glaucoma Prevention Study Group11 verified that, when assessing a disc for glaucomatous damage, intraobserver agreement is generally much stronger than interobserver agreement.
THE FUTURE OF DIGITAL IMAGING
Digital images can be presented in various ways to provide greater sensitivity in detecting glaucomatous progression. By using systems that “flicker” serial disc images on a computer screen, one method improves the detection of changes in the neuroretinal rim compared with ophthalmoscopic and stereophotographic methods of assessment.12 The Disc Photography Reading Center used in the Early Manifest Glaucoma Trial used flicker chronoscopy, which had previously been shown to sensitively detect small changes in optic disc topography.13
Parkin et al14 compared stereoscopic and monoscopic evaluations of optic disc pairs using flicker chronoscopy and a handheld stereoviewer (Figure 2). The investigators noted similar intraobserver agreement between stereoscopic and monoscopic evaluations of the vertical cup-to-disc ratios. Interobserver agreements showed better agreement for stereoscopic (ICC = 0.84) than for monoscopic evaluations (ICC = 0.71). Morgan et al15 also found an improvement in interobserver agreement when using digital images to estimate the cup-to-disc ratio for stereoscopic assessment (ICC = 0.80) when compared with monoscopic assessment (ICC = 0.74).
Figure 2. The physician uses a handheld stereoviewer to assess digital stereophotographs of the optic disc on a computer monitor.
The advent of charge-coupled display (CCD) cameras allows clinicians to obtain digital images of the optic disc directly instead of relying on scanned 35-mm color slides. When comparing analog and digital imaging methods, it is useful to know that, given the density of silver halide crystals, the theoretical resolution of 35-mm film is approximately 6,000 dots per inch (dpi).16 Scanning a 35-mm slide (0.94 X 1.42 inches) at this resolution would generate an image that is over 144 megapixels in size. Due to interpolation routines in scanning software, such high resolution is rarely necessary.
Scanning at 3,000 dpi will produce a 36-megapixel color image with quality comparable to that of 35-mm film.17 With a digital camera—where each red, green, and blue CCD sensor registers as a separate pixel—the resolution of a 6.1-megapixel digital camera would be equivalent to scanning a 35-mm slide at 1,200 dpi. Khouri et al18 compared simultaneous optic disc stereophotographs obtained from 35-mm film to digital images obtained using a 6.1-megapixel camera. The investigators found no significant differences in agreement regarding the evaluation of cup-to-disc ratio.
STILL THE GOLD STANDARD
Several studies have documented that structural injury of the optic nerve head is one of the earliest indicators of progressive glaucomatous damage and can precede detectable visual field loss.19-21 It is therefore imperative that the imaging techniques practitioners employ permit them to discern subtle changes in the optic disc's appearance over time. The high resolution of 35-mm film stereophotographs has allowed physicians to detect these changes, but the method's shortcomings suggest that serial evaluations are most reliable when performed by the same person.
Digital stereophotography has the advantage of providing instant feedback on the image's quality, and it allows computerized analysis of the image, its electronic storage, and its use in telemedicine.18 Standard CCD cameras, however, provide less-than-optimal resolution compared with 35-mm film. These cameras may require a resolution higher than 6 megapixels in order to detect subtle changes in serial disc images.
One hopes that newly refined measures of quantifying changes in the optic disc and nerve fiber layer by means of confocal scanning laser ophthalmoscopy, ocular coherence tomography, or scanning laser polarimetry will allow the earlier detection of glaucomatous progression. Until longitudinal data validate these devices, stereophotography of the optic disc will remain the gold standard against which these exciting new technologies are measured.
Stuart J. McKinnon, MD, PhD, is Associate Professor at Duke University Eye Center in Durham, North Carolina. He acknowledged no financial interest in the products or companies mentioned herein. Dr. McKinnon may be reached at (919) 684-8424; stuart.mckinnon@duke.edu.
1. American Academy of Ophthalmology. Primary open-angle glaucoma. Preferred Practice Patterns. Limited revision. San Francisco, CA: American Academy of Ophthalmology; 2003.
2. El Bayadi G. New method of slit lamp microophthalmoscopy. Br J Ophthalmol. 1953;37:625-628.
3. Kottler MS, Rosenthal AR, Drance SM, et al. Twin-prism stereo-separator. Can J Ophthalmol. 1976;11:78-81.
4. Allen L. Ocular fundus photography. Am J Ophthalmol. 1964;57:13-28.
5. Lichter P. Variability of expert observers in evaluating the optic disc. Trans Am Ophthalmol Soc. 1976;74:532-572.
6. Caprioli J, Prum B, Zeyen T. Comparison of methods to evaluate the optic nerve head and nerve fiber layer for glaucomatous change. Am J Ophthalmol. 1996;121:659-667.
7. Rosenthal AR, Kottler MS, Donaldson DD, et al. Comparative reproducibility of the digital photogrammetric procedure utilizing three methods of stereophotography. Invest Ophthalmol Vis Sci. 1977;16:54-60.
8. Tielsch JM, Katz J, Quigley HA, et al. Intraobserver and interobserver agreement in measurement of optic disc characteristics. Ophthalmology. 1988;95:350-356.
9. Abrams LS, Scott IU, Spaeth GL, et al. Agreement among optometrists, ophthalmologists, and residents in evaluating the optic disc for glaucoma. Ophthalmology. 1994;101:1662-1667.
10. Varma R, Steinmann WC, Scott IU. Expert agreement in evaluating the optic disc for glaucoma. Ophthalmology. 1992;99:215-221.
11. Zeyen T, Miglior S, Pfeiffer N, et al; European Glaucoma Prevention Study Group. Reproducibility of evaluation of optic disc change for glaucoma with stereo optic disc photographs. Ophthalmology. 2003;110:340-344.
12. Shuttleworth GN, Khong CH, Diamond JP. A new digital optic disc stereo camera: intraobserver and interobserver repeatability of optic disc measurements. Br J Ophthalmol. 2000;84:403-407.
13. Heijl A, Bengtsson B. Diagnosis of early glaucoma with flicker comparisons of serial disc photographs. Invest Ophthalmol Vis Sci. 1989;30:2376-2384.
14. Parkin B, Shuttleworth G, Costen M, et al. A comparison of stereoscopic and monoscopic evaluation of optic disc topography using a digital optic disc stereo camera. Br J Ophthalmol. 2001;85:1347-1351.
15. Morgan JE, Sheen NJ, North RV, et al. Digital imaging of the optic nerve head: monoscopic and stereoscopic analysis. Br J Ophthalmol. 2005;89:879-884.
16. Accurate image manipulation for desktop publishing. Silver Halide Resolution. Available at: http://www.aim-dtp.net/aim/technology/silver_halide/index.htm. Accessed October 25, 2005.
17. R. N. Clark's Photography. Image detail. Available at: http://www.users.qwest.net/~rnclark/scandetail.htm. Accessed: October 6, 2005.
18. Khouri AS, Szirth BC, Realini T, Fechtner RD. Digital images correlate well with film for simultaneous stereophotography of the optic nerve in glaucoma. Paper presented at: The ARVO Annual Meeting; May 5, 2005; Fort Lauderdale, FL.
19. Quigley HA, Addicks EM, Green WR. Optic nerve damage in human glaucoma. III. Quantitative correlation of nerve fiber loss and visual field defect in glaucoma, ischemic neuropathy, papilledema, and toxic neuropathy. Arch Ophthalmol. 1982;100:135-146.
20. Sommer A, Katz J, Quigley HA, et al. Clinically detectable nerve fiber atrophy precedes the onset of glaucomatous field loss. Arch Ophthalmol. 1991;109:77-83.
21. Kass MA, Heuer DK, Higginbotham EJ, et al. The Ocular Hypertension Treatment Study: a randomized trial determines that topical ocular hypotensive medication delays or prevents the onset of primary open-angle glaucoma. Arch Ophthalmol. 2002;120:701-713.
