The ability of confocal scanning laser tomography to quantify the subretinal features was investigated. The slope ratios (anterior slope/posterior slope) of the axial intensity profiles were analyzed. The data from normal subjects showed only minimal influence of individual ocular pigmentation. In the eyes with age-related macular degeneration, the light-tissue interactions vary according to the type of retinal features. Three-dimensional information could be obtained from the axial intensity profiles.
©2001 Optical Society of America
Three-dimensional information of the retinal and subretinal layers can provide useful information concerning the sight-threatening exudation in age-related macular degeneration (AMD) (1,2). This work is aimed at characterizing the light-tissue interactions in the deeper layers of the human fundus, so that pathological features of AMD can be detected and quantified with minimal influence from individual differences in ocular pigmentation. The deeper layers of the human fundus are the site of much of the damage in AMD, but these have not been adequately studied in large patient-oriented studies. It is critical to develop robust methods to detect pathological features leading to severe vision loss, including both the deeper layers and the retina (3).
A rapid and noninvasive way to quantify the severity of the exudation is to use confocal scanning laser tomography. A series of confocal images that vary in depth is acquired rapidly, and these are analyzed to provide three-dimensional information. Quantitative analysis requires that the instrument provide an output that is linear with respect to the amount of light returning from the layers of the eye (Fig. 1), rather than merely an image that is improved in contrast by the use of an aperture. The intensity of light returning to the detector at each focal volume, as a function of the axial position, is called the axial profile function (Fig. 2). The most common way to use this three-dimensional information is to assume that the retinal surface lies at the maximum intensity of the axial profile function. In most previous studies, the location in depth of a single maximum intensity was used to map the relative heights of the retinal surface across the retina.
Exudative structures elevate the overlying retina, and confocal scanning laser tomography is used to map this elevation (1,2). Exudation can cause substantial elevation, such as pigment epithelial detachments hundreds of microns thick (1). A pigment epithelial detachment that is unusually high compared with its diameter is associated with poor visual acuity; angiography has confirmed that neovascular growth invaded the retina as well as the choroid. We have shown that this is predictive of severe and widespread loss of vision (3). We have also used a variation of this technique to visualize and quantify small exudative structures that reduce visual acuity: macular cysts imaged as separate structures could not be resolved on photographic fluorescein angiography (2).
The separate images used in the three-dimensional computations contain more information than merely the height of the retinal surface. Confocal tomography has been used with indocyanine green dye to visualize and quantify microvascular features in choroidal melanoma (4), typically with blind deconvolution (5). Hudson and colleagues (6) showed a method to estimate retinal thickness in macular edema associated with diabetes. We have shown that the individual confocal images in a series visualize structures differently, and that the axial profile functions are not always a curve with a single peak (1, 2, 7, 8). Thus, from our work, as well as that from Holmes (5), Hudson (6), and their colleagues, the light-tissue interactions of interest in the pathological fundus are not well-described by the idealized model that a reflection from a large blood vessel behaves as a point source (9). With methods to reduce aberrations of the anterior segment, the available depth information concerning light-tissue interactions of a pathological retina should be even more useful clinically.
The axial profile of light returning from the fundus is based on the optics anterior to the retinal and choroidal layers and the light-tissue interactions within them. The main absorbing tissues in the normal eye are blood and melanin (10). We previously showed that there is little difference between eyes for confocal or scattered light infrared imaging, when the retinal surface is in focus (11). However, much of the pathology in exudative AMD is beneath the retina. The main melanin absorption of the human eye lies beneath the retina in the choroidal and retinal pigment epithelial layers, with the choroid being the main absorbing layer and corresponding to iris color (10). The choriocapillaris and larger choroidal vessels also lie beneath the retina.
We investigated whether there are significant individual differences in the axial profile functions in near infrared for a baseline sample of young, healthy eyes and those of patients with AMD. We determined whether the axial profiles were significantly influenced by ocular pigmentation. Significant melanin absorption in the choroid would cause less light return from these layers in dark eyes as opposed to light eyes; the posterior slope of the axial profile function would be steeper relative to the anterior slope for dark eyes vs. light eyes. The ratio of the anterior to posterior slope would then be smaller for dark eyes than for light eyes. We investigated the confocal scanning laser tomography results for young to middle-age aged subjects with AMD patients. We compared the axial profile functions for several retinal locations containing different features. The effects of absorption, if significant, should have an affect on the axial profile function, from most to least, for these regions of interest: 1) fovea outside of the foveal reflex, 2) temporal retina, 3) nasal retina near the ONH, 4) central portion of largest retinal vessel near the macula, and 5) central reflex from a large retinal vessel (preferably artery) near the ONH. For patients with AMD, we investigated the axial profile functions of exudative features. For example, significant light-tissue interactions in the pathological deeper layers should influence the slope of the axial profile function posterior to the retinal surface.
We tested 10 normal subjects, 5 with blue or light green eyes and 5 with dark brown eyes. We selected young subjects to provide baseline data for two reasons. First, these subjects were young enough to precede the loss of ocular pigmentation with age (25–41yr), providing a baseline against which changes due to aging or pathology can be measured. Second, poorer optical quality in older eyes might lead to such broad axial profile functions that a potential artifact of ocular pigmentation would be quantifiable only with a large number of subjects. We compared these results to the data of 6 patients with AMD (age 73–80yr). All subjects signed an informed consent form and were tested with a protocol approved by our Institutional Review Board.
2.2 Scanning laser tomography
The TopSS (Laser Diagnostic Technologies, San Diego, CA) was used to acquire a series of 32 images of the macula in 0.9 sec, using 790 nm illumination. The field of view was 20×20 deg, and the scan depths were nominally 2 or 3 mm. Large eye movements are readily visible, indicating to the operator to retake a series. Two image series were used for each subject.
2.3 Data analysis
From the raw data, each image series was separated using our software. Next, the effects of the remaining small eye movements occurring within the 0.9 sec of image acquisition were quantified by using retinal landmarks, such as blood vessel crossings. Potential errors were in the transverse direction, within measurement error. Compensation was made to each image so that the location on the retina of each sample remained constant over the series and between series. Samples were taken from each image for each region of interest above, using sub-samples of 8×8 pixels except on the narrowlumen of vessels. All obvious artifacts unrelated to the tissues or pathology of interest, such as retinal vessels, were avoided for purposes of the strictest test for artifacts due to ocular pigmentation. The mean of individual samples for each region of interest, as well as the multiple samples per se, were computed.
We examined parameters of the axial profile functions: anterior slope and posterior slope from 50% height to 90% height, and goodness of fit. The slope and goodness of fit calculations were made by interactively selecting the above points from each curve and computing a linear regression based on a least squares fit, using Statview (SAS, Cary, NC). The average of each anterior and posterior slope for each region of interest was computed from two series. For data collected within one series for a single eye, which used the same illumination and gain values, any anterior of posterior slopes can be compared for a region of interest. Slopes may be compared for relatively normal as opposed to pathological regions in the eye of a patient with AMD. For comparisons between different eyes or among patients, the individual slopes are affected by the gain settings, which need to be set according to the pathology of a given eye to obtain a suitable dynamic range to characterize the light-tissue interactions. A steeper gain of an axial profile function can result from a greater return of light from a given structure or a higher gain setting, while a shallower slope can result from a lesser return of light or a lower gain setting. One parameter that is not influenced by gain changes is the slope ratio, i.e. the ratio of anterior to posterior slopes, as long as the instrument output is a linear function of light returning from the eye. We performed a calibration to determine if the instrument behaves in a linear manner
2.4 Gain calibration
We calibrated the gain settings of the TopSS using a model eye. We used two combinations of neutral density filters that allowed a low and a high gain at similar gray scale outputs. Using the same region of interest on the target in the model eye, we obtained a sample for each image. On these samples, we performed a linear regression, using a least squares fit, of the output at high gain as a function of the output at low gain for the corresponding image numbers in the two series. The result confirmed a linear relation between target intensity and output, described as y=ax+b. The slope of the function, a, varies as a constant with the actual gain setting used, as expected (Fig. 1). The small offset difference, b, has little effect on the computations from the axial profile functions, since only the top 50% of the values are used. The linear fit is excellent.
In patient studies, gain must be optimized according to the main pathological features: fluid-filled lesions are dark, but atrophy and hard exudates are bright. Thus, we chose parameters that were insensitive to gain: slope ratio, defined as anterior slope/posterior slope, as well as the correlation coefficient r for the linear portion of the slopes. For analysis in the same image series, slope may be used as well as slope ratio.
3. Results and Discussion
3.1Individual variation of subretinal pigmentation
There are individual differences in the shapes of the axial profile functions, but no statistically significant difference for any parameter as a function of eye color (ANOVA, P=0.80) (Figs. 2, 3, 4). If melanin absorption in the layers beneath the retina were an important factor, then the posterior slopes would be steeper for subjects with dark eyes, and the slope ratios would have been significantly less. The subjects with dark eyes did not have significantly better or worse fits of the linear portions of the axial profiles, nor did they have lower slope ratios for the axial profile functions in the center of large retinal vessels. These findings together indicate that there were not large differences in the optics of our sample of dark vs. light eyes that masked the influence of melanin. There was also no effect of melanin on the slope ratios of two different areas: the fovea vs. a major retinal vessel near the optic nerve head, with the former having ocular pigmentation directly beneath the retinal layers, which are thin compared with other regions, and the latter with the main reflection anterior to both blood and melanin (paired-t, P=0.35). Thus, this method is insensitive to the artifact of normal ocular pigmentation, which varies greatly in the deeper layers that are the site of the main exudative pathology in AMD.
3.2 Data from AMD
The data for patients with AMD showed a significant difference of the slope ratio of region of interest (Fig. 5). The light-tissue interactions vary according to the type of pathology or structure, e.g. retinal vessel vs. CNV (paired-t, P=0.048). This indicates that confocal scanning laser tomography provides not only information concerning the retinal surface, including the elevation of the retina over exudation, but also information about the deeper layers. This agrees with our previous finding of multiple peaks consistent with the relative heights of pathological features in AMD (7).
New computational methods or correction of anterior segment optics could be combined with infrared imaging approaches to provide information about the important pathological sites beneath the retina in AMD, building on techniques that have been mainly limited to the anterior layers of the retina. As confocal scanning laser tomography does not require the injection of dye and is rapid, it could provide an inexpensive means of either early detection or more frequent follow-up of patients following treatment for exudation.
There has been a serious public health problem created by the long intervals between allowable reimbursements and the low rates at which diagnostic testing is reimbursed. The effects of some treatments have been studied by limited diagnostic means and at intervals longer than those known to be optimal for detection of recurrence of exudation or consequences of failure to treat adequately (12,13). A method that is rapid, safe, painless, and inexpensive may improve the management of patients with AMD. Noninvasive techniques may benefit the patients needing closer management.
EYO7624 and EY12178
References and links
1. C. Kunze, A. E. Elsner, E. Beausencourt, L. Moraes, M. E. Hartnett, and C. L. Trempe, “Spatial extent of pigment epithelial detachments in age-related macular degeneration,” Ophthalmology 106, 1830–40 (1999). [CrossRef] [PubMed]
2. E. Beausencourt, A. Remky, A. E. Elsner, M. E. Hartnett, and C. L. Trempe, “Infrared scanning laser tomography of macular cysts,” Ophthalmol. 107, 375–85 (2000). [CrossRef]
3. M. E. Hartnett, J. Weiter, G. Staurenghi, and A. E. Elsner, “Deep retinal vascular anomalous complexes in advanced age-related macular degeneration,” Ophthalmol. 103, 2042–2053 (1996).
4. A. J. Mueller, W. R. Freeman, R. Folberg, D. U. Bartsch, A. Scheider, U. Schaller, and A. Kampik, “Evaluation of microvascularization pattern visibility in human choroidal melanomas: comparison of confocal fluorescein with indocyanine green angiography,” Graefes Arch. Clin. Exp. Ophthalmol. 237, 448–56 (1999). [CrossRef] [PubMed]
6. C. Hudson, J. G. Flanagan, G. S. Turner, and D. McLeod, “Scanning laser tomography Z profile signal width as an objective index of macular retinal thickening,” Br. J. Ophthalmol. 82, 121–30 (1998). [CrossRef] [PubMed]
7. M. Miura, A. E. Elsner, E. Beausencourt, C. Kunze, K. Lashkari, M. E. Hartnett, C. L. Trempe, and T. Hirose, “Cross-sectional analysis of age-related macular degeneration with confocal scanning laser tomography,” Invest. Ophthalmol. Vis. Sci. 41, S162, Abstract Nr: 839 (2000).
8. A. E. Elsner, M. Miura, S. A. Burns, E. Beausencourt, C. Kunze, L. Kelley, J. Walker, G. P. Wing, P. Raskauskas, D. Fletcher, Q. Zhou, and A. W. Dreher, “Multiply scattered light tomography and confocal imaging: detecting neovascularization in age-related macular degeneration,” Opt. Express 7, 95–106 (2000), http://www.opticsexpress.org/oearchive/source/22805.htm. [CrossRef] [PubMed]
9. A. W. Dreher, J. F. Bille, and R. N. Weinreb, “Active optical depth resolution improvement of the laser tomographic scanner,” Appl. Optics 28, 804–808 (1989). [CrossRef]
10. F. C. Delori and K. P. Pflibsen, “Spectral reflectance of the human ocular fundus,” Appl. Optics 28, 1061–1077 (1989). [CrossRef]
12. Macular Photocoagulation Study Group, “Five-year follow-up of fellow eyes of patients with age-related macular degeneration and unilateral extrafoveal choroidal neovascularization,” Arch. Ophthalmol.111, 1189–99 (1993). [PubMed]
13. Macular Photocoagulation Study Group, “Krypton laser photocoagulation for idiopathic neovascular lesions. Results of a randomized clinical trial,” Arch. Ophthalmol.108, 832–37 (1990). [PubMed]