Conversion of Lens Slit Lamp Photographs into Physical Light-Scattering Units

Abstract

purpose. To derive from lens slit lamp photographs by means of densitometry the physically defined quantity for light scattering (the Rayleigh ratio) and to expand the use of the Lens Opacity Classification System (LOCS III) to include clear lenses and also to calibrate the LOCS III Nuclear Opacity (NO) score in physical terms.

methods. Series of slit lamp photographs were taken from 38 eyes from 29 subjects (age range 18 to 84 years old) including cataracts, for 0.1- and 0.2-mm slit width, using 200 ASA and 1600 ASA film speed (Kodak professional; Eastman Kodak, Rochester, NY) and different flash settings with a Topcon SL-6E (12 slit/speed/flash combinations; Paramus, NJ). Additionally 19 eyes were photographed with a Zeiss 40 SL/P (8 slit/speed/flash combinations; Carl Zeiss, Thornwood, NY). A calibrated suspension of latex spheres also was photographed at the same 20 conditions. Densitometry was performed on the nuclear area of all photographs including the LOCS III standards, using a photometrically corrected photocell. Slit width and flash intensity settings were photometrically calibrated. All eyes and the suspension were digitally “photographed” with the EAS-1000 (Nidek, Gamagori, Japan) Scheimpflug system.

results. For each eye and the suspension, the series of 20 or 12 densities, corresponding to a range of about 1 log unit in the amount of light used, proved to follow closely a course common to all eyes (the two film characteristics), apart from a shift in the amount of light (because of the differences in light back scattering).

conclusions. From normal slit lamp photographs, the physical quantity for light (back) scattering can be derived using transformation graphs derived in this study. The LOCS III NO score also can be used for clear lenses and translated into physical units. In this way, slit lamp photography can be used better for more precise studies, provided some minimal calibration of the photograph slit lamp.

Slit lamp observation, direct or photographic, is still the major way to assess the condition of the optical media of the patient’s eye. Although the slit lamp is unsurpassed for visualization of lens details, it has long been realized that the observation cannot be used directly for comparative research, especially in epidemiologic studies. Cataract classification systems were developed, i.e., methods to transform the subjective slit lamp observation to semi-quantitative data. ^{1 2 3 4} Slit lamp observation rests on light that is scattered (or reflected) backwards. This backscattered light can also be measured with photosensitive devices giving objective and quantitative data. ^{5 6 7 8 9 10 11 12} However, these instruments have not (yet) pushed aside the slit lamp biomicroscope. 

All the above methods use the backscattered light, but results are presented in more or less arbitrary units, different between the different methods. A physical definition of light scattering is given by the so-called Rayleigh ratio R. It would be valuable if such a well-defined and generally applicable unit could be used easily in lens research. This would aid, e.g., in the comparison between epidemiologic studies or for in vivo versus in vitro studies. Because of the dominance of the normal slit lamp, we studied whether this instrument can be used to estimate R in a simple way. 

Two approaches were evaluated, photographic and direct observation at the slit lamp. Photographic information can be transformed to quantitative data by densitometry. With careful calibration of the photographic procedure and the use of a backscattering standard, a direct relationship can be established between photograph density and R. However, to eventually avoid the need for densitometry, a transformation from subjective cataract score to R also was developed. As the scoring system, LOCS III for nuclear opacity (NO) was used. ⁴ In this system, a direct slit lamp picture or a photograph is compared by eye to five standard photographs with different grades of nuclear opacity, numbered 1 to 5. The densities of these five photographs also were measured and transformed to R. The NO score of a patient’s lens, e.g., 3.6 (one decimal place by visual interpolation), can then be transformed to R. Both transformation techniques, densitometry-based as well as LOCS III–based, were tested against independent assessments of R on the same lenses. 

Methods

Series of slit lamp photographs were taken in two groups: (1) from 19 eyes from 19 subjects (age range 18 to 64 years old), each photograph once repeated, and (2) from 19 eyes from 10 subjects (age range 55 to 84 years old, from a cataract clinic, some with nuclear cataract), without repetitions. For all eyes, 12 photographs were made with 0.1- and 0.2-mm slit width, using 200 ASA and 1600 ASA film speed (Kodak professional; Eastman Kodak, Rochester, NY), and different flash settings (see Table 1 ) with a Topcon SL-6E (12 slit/speed/flash combinations; Topcon, Paramus, NJ). Additionally, the first group of 19 eyes was photographed with a Zeiss 40 SL/P, also with 0.1- and 0.2-mm slit width, using 200 ASA and 1600 ASA film speed (8 slit/speed/flash combinations; Carl Zeiss, Thornwood, NY; see Table 1 ). The light sources were the standard xenon-filled flash tubes, used in these instruments, with no color filters interposed. A calibrated suspension of latex spheres was photographed at the same 20 conditions. The suspension was contained in an artificial eye, the “cornea” consisting of a zero-diopter, 8-mm-radius contact lens. Readings were taken at 2 mm behind the“ cornea.” Photographic magnification was 2.2 for the Topcon and 1.8 for the Zeiss camera. This difference causes in itself a difference in film exposure of a factor (2.2/1.8)² = 1.5 (0.17 log units) higher exposure, corresponding to lower magnification (the Zeiss camera). The films were developed by a certified professional photograph laboratory (Q-laboratory). Standard procedures in such a laboratory include developing a trial film every morning and evening and checking it densitometrically, and once a week Kodak checks the film. The photographic procedures were according to and including the set of conditions of the LOCS system (200 ASA, 0.2 mm, etc.). 

Densitometry on the nuclear area of all photographs including the LOCS III standards (positive color transparencies of the standard 2.4 × 3.6-cm size) was performed using a photometrically corrected photocell. A circular diaphragm in front of the photocell was adjusted to cover the full depth of the nucleus, defined according to other densitometric studies. ^{4 13} Slit width and flash intensity settings were calibrated with the help of a photometrically corrected photocell, placed at the same position as the patient’s eye, which was large enough to collect all light. All eyes and the suspension also were digitally “photographed” with the EAS-1000 Scheimpflug system (Nidek, Gamagori, Japan). 

As physical unit for the light-scattering property of scattering materials, the so-called Rayleigh ratio is used:  

\[\mathrm{Rayleigh\ ratio}{=}R({\theta}){=}I({\theta}){\div}E{_\ast}V,\ \mathrm{or\ }I({\theta}){=}R({\theta}){_\ast}E{_\ast}V,\]

with I(θ) the amount of light scattered per unit solid angle in W/steradian by a volume of scatterer V in m³ illuminated by incident light E in W/m². In the present study, I(θ), where θ = 135°, is proportional to the amount of light collected by the reception aperture of the slit lamp microscope, part of which is projected on the photographic film. The resulting light density on the film is proportional to 1/magnification. ² I(θ) is proportional on the other hand to E, which is controlled by the flash setting on the slit lamp. In fact, instead of I(θ) and E, their integrals over the full flash duration determine film exposure. I(θ) and film exposure depend moreover on slit width, because V does. In fact, because of limited depth of focus of the slit projection system, E is not constant over lens depth. However, because slit width is small, only the total E over slit width (in the focal plane E × slit width), in combination with the surface of the illuminated cross section instead of V, need be considered. The maximal flash was larger with the Zeiss instrument, but the reception aperture was smaller (larger depth of focus), resulting in about equal film exposure, apart from the magnification effect (factor 1.5). 

For the suspension of latex spheres in water, at a concentration of 5 × 10⁻⁵ weight fraction and a sphere radius of 98 nm, the Rayleigh ratio for unpolarized light at 135° R(135) = 1.0 m⁻¹ Sr⁻¹, more or less constant between 450 and 650 nm. This was measured separately using a radiometric setup described earlier. ^{14 15 16} On the basis of the mentioned parameter values and a refractive index of about 1.58, slightly depending on wavelength, the experimental result was found to be in keeping with well-established theory, using the so called Rayleigh–Gans approximation. ¹⁷ It must be noted that these spheres are intermediary between the well-known small particle Rayleigh domain (scattering proportional to wavelength⁻⁴) and the large particle domain with deep local minima in scattering as a function of angle, first appearing on the small wavelength side. For the present suspension, e.g., at precisely 500 nm R(135) = 0.97 and at 600 nm R(135) = 1.03. Throughout this article, Briggsian logarithms (base 10) are used. 

Results

Figures 1a nd 2 show examples of the relation between log film transmittance (= minus photograph density) and relative film illuminance for a 19-year-old subject and a 64-year-old subject, respectively. To clarify these figures, it must be noted that film characteristics generally are described by log(film transmittance) = function[log(I _film)]. For any camera, film illuminance I _film = k · R · I _source, where I _source is the light intensity incident on the sample, R is the back scattering strength of the sample, and k is a constant depending on the camera and its settings. So, log(film transmittance) = function[ log(k) + log(R) + log(I _source)]. Thus, plots of log(film transmittance) versus [log(k) + log(I _source)] for samples having different scattering strengths should have the same shape, but they are shifted along the horizontal axis according to their log(R) values. The dependence on k and I _source is combined in “relative film illuminance,” which is plotted along the horizontal axis and is calculated using the above calibration factors for the two instruments (a factor 1.5 difference due to magnification), the two slit widths (factor 2 difference), and the seven different flash settings. All these factors were combined to arrive at a value for film illuminance relative to the condition (Topcon, 200-W flash, 0.2-mm slit width). The curves for the two subjects show the same shape, only shifted horizontally because the nucleus of the older lens scatters a larger proportion of the incident light backwards. 

Figure 3 shows the data of all subjects superimposed, after individual adjustment for the horizontal shift, based on the least squares criterion. In compliance with the correspondence between Figures 1 and 2 , this figure shows that the curves for all subjects follow a common course, viz. the two film characteristics. The circles give the data for the latex spheres suspension with known Rayleigh ratio R. The horizontal axis was chosen such that the data point (Topcon, 200-W flash, 0.2-mm slit width) lies at the correct R. The drawn lines are four-parameter fits for the four-parameter logistic literature model for film characteristics. ¹⁸ For 200 ASA: log film transmittance = −3.38 + 3.27/(1 + 10^{−1.25(logR−0.01)}). For 1600 ASA: log film transmittance = −3.01 + 2.90/(1 + 10^{−1.52(logR+0.57)}). These parameters were adjusted simultaneously with the above-mentioned horizontal shifts (one for each lens), on the basis of the least squares criterion. It must be noted that the film characteristics themselves could have been estimated independently, using only the latex suspension or other samples of backscattering material. Instead, we chose to use the available dataset, because one may expect that using much data gives better estimates for the film characteristics. 

Also indicated in Figure 3 are the findings for log film transmittance of the LOCS III standards NO1 through NO6. For research that uses precisely the LOCS III settings (200 ASA, 0.2-mm slit width, etc.), the following are the respective log R values for the NO standards that can be read from this figure: 0.00, 0.37, 0.49, 0.62, 0.77, and 0.82. If one would only change the use of film to 1600 ASA, the values are, respectively, −0.65, −0.33, −0.22, −0.11, 0.01, and 0.06. A warning must be placed here that the LOCS grades presented in this article no longer directly grade the respective lenses. To be more precise, the LOCS system grades images of the lenses. Because in the original LOCS system the images are produced identically to the way that the LOCS standards themselves are produced, there is a one-to-one relationship between grades and lenses. In the present study, the images may be produced in different ways to accommodate a broader range of applications, especially a broader range of intrinsic lens-scattering efficiencies. So, the LOCS grades must be interpreted, depending on the way the images are produced. 

Figure 3 could be used directly to derive R from log film transmittance of a slit lamp photograph made with the Topcon at 200-W flash and 0.2-mm slit width. Because it is more customary to have the independent variable (log film transmittance) horizontally, in Figure 4 the x- and y-axis are exchanged, but the same model lines are drawn. For other flash or slit width settings, a correction for the difference in light content must be made, which is a simple shift because of the logarithmic R axes. Figure 4 gives as an example the line for 1600 ASA, 0.1-mm, and 200 W. Figure 4 is the calibration graph in case log film transmittance (optical density) is available. 

If, however, scoring with the LOCS III system is used, another calibration graph is needed (Fig. 5) . Also, this graph is derived from Figure 3 . For each of the 6 NO standards the intersections of log film transmittance with the two film characteristics gives the R values for 200-W flash and 0.2-mm slit width. These values are plotted in Figure 5 . For other flash or slit width settings, the film characteristics also must be shifted. Figure 5 gives as example the line for 1600 ASA, 0.1-mm, and 200 W. Each potential calibration line in Figure 5 can be derived graphically from Figure 3 (if needed, after the appropriate horizontal shift) or numerically, using the above formulas for the two film characteristics and the values for log film transmittance of the LOCS III NO standards: −1.76, −0.97, −0.76, −0.58, −0.43, and −0.39. 

Discussion

The present study was designed to translate slit lamp photographs to physical light-scattering units, the Rayleigh ratio R. Densitometry on the nucleus was performed, but, of course, the relation between the densitometric values and R holds equally for other parts of the lens. It must be noted, however, that the observed intensity of the backscattered light can be influenced by light absorption in the lens. For younger lenses, this is of little consequence, because the lens pigments absorb predominantly below 500 nm, whereas the photometric and visual sensitivity used in this study lies predominantly above 500 nm. In older lenses, pigments may accumulate to such amounts that backscattered light from deeper layers, especially from the posterior pole, may be weakened significantly. The observed intensity of the backscattered light also can be influenced by scattering of both the incident light or the scattered light itself. These effects are, however, unimportant because only minute amounts are scattered over larger angles. Only a few percent is scattered over more than 10°. ^{19 14}  

The outcomes of both presented methods were verified (Figs. 6 7) . Figure 6 shows a result of the densitometric approach for the same subjects as used to construct Figure 3 ; Scheimpflug images also were made with an EAS-1000 instrument. This instrument was calibrated, using the same suspension of latex spheres, in terms of Rayleigh ratio. Figure 6 shows that the correspondence is close, considering that different planes are cut through the lens. The Scheimpflug instrument cuts the lens over its optical axis, and the slit lamp biomicroscope cuts the lens obliquely at 45° from its optical axis. 

Figure 7 shows verification of the approach using LOCS III NO scoring. For this figure, data were used from a study on light scattering by lenses extracted from donor eyes. ^{14 15 16} Twelve lenses were photographed using 1600 ASA film speed, because these were age-normal lenses (except one lens with nuclear cataract), and scored using the LOCS III NO system. The main issue was to measure and interpret the light-scattering properties in forward as well as backward directions at wavelengths 400, 500, 602, and 700 nm. Figure 7 shows horizontally log Rayleigh ratio of the nucleus for 602 and 500 nm at 140° (40° backwards). Vertically log Rayleigh ratio is plotted as read from the present Figure 5 using the NO scores given earlier. ¹⁴ Note that the present approach involves implicitly averaging over the middle of the visual spectrum, i.e., 555 ± 50 nm. Indeed, Figure 7 shows that the values found are between those for 500 and 602 nm. The ideal correspondence is represented by the two lines, which are positioned according to the well-known wavelength⁻⁴ dependence of light scattering by small particles. It recently was found that this type of scattering dominates in human lenses in the relevant angular domain (around 40° or 45° backward scattering), especially in the nucleus. ¹⁶ The variability in Figure 7 is larger than in Figure 6 . This can be expected because the comparison of Figure 7 involves not only different cuts through the lenses, as in Figure 6 , but also other differences. NO scoring involves a subjective judgment. Moreover, the data are from a completely independent study, performed on donor lenses. Important in this study was that direct measurements on wavelength-resolved Rayleigh ratios were performed. In view of all the differences, the correspondence shown in Figure 7 seems satisfactory. 

In conclusion, normal slit lamp observation and photography can be used to derive the physically defined measure for light scattering, the Rayleigh ratio, making it more suitable for comparative studies. With Figures 4 and 5 and eventually the formulas for the two film characteristics and the density values for the LOCS III NO standards, the conversion can be made directly in case proper standard procedures are followed, such as defined in the LOCS III standard. To include a wider range of applications, the LOCS III standard was extended to include using the 200-W flash tube, professional Kodak film of 200 or 1600 ASA, well-defined slit widths (0.2 or 0.1 mm), and illumination at 45° from the optical axis, along which photography is performed. Some minimal calibration of the slit lamp is highly advisable though. A good way to calibrate is to photograph with one’s standard settings the latex suspension defined in the Methods section. The densitometric value of the latex spheres’ photograph, using a photometric photocell, fixes the position along the horizontal axis of the curves of Figure 3 (vertically in Fig. 4 ). In other aspects, these curves remain the same. It would be most accurate if the photographs of the patients also are measured densitometrically in the same way. Otherwise, one would proceed to derive one’s own curves of Figure 5 and to use the normal LOCS III NO scores. 

 Reprint requests: Thomas J. T. P. van den Berg, AMC/Dept. Med. Physics, PO Box 22660, 1100 DD Amsterdam, The Netherlands.

 Supported by the Technology Foundation (STW).

 Submitted for publication October 13, 1998; revised February 10, 1999; accepted March 25, 1999.

 Proprietary interest category: N.

 

Table 1.

 

Parameters of the Photographs Made

Film Speed (ASA):	Topcon SL-6E							Zeiss 40 SL/P
		200			1600			200			1600
Slit width (mm):	0.1	0.2	0.1	0.2	0.1	0.2	0.1	0.2
Flash setting
2	−	−	−	+	−	−	−	+
3	−	+	+	+	−	+	+	+
4	+	+	+	+	+	+	+	+
5	+	+	+	+	∗	∗	∗	∗

*  Not present at this instrument.

Figure 1.

 

Densitometric values using a photometrically corrected photocell of the nucleus in slit lamp photographs from a 19-year-old subject for different slit/flash combinations. The amount of light delivered by the different slit/flash combinations was photometrically calibrated (horizontal axis). The codes denote slit/slit lamp/flash, e.g., 2t4 denotes 0.2-mm slit/Topcon/flash no. 4. The symbols are the results of a fit of a function describing the film characteristic.

Figure 2.

 

Densitometric values using a photometrically corrected photocell of the nucleus in slit lamp photographs from a 64-year-old subject for different slit/flash combinations. The amount of light delivered by the different slit/flash combinations was photometrically calibrated (horizontal axis). The codes denote slit/slit lamp/flash, e.g., 2t4 denotes 0.2-mm slit/Topcon/flash no. 4. The symbols are the results of a fit of a function describing the film characteristic.

Figure 3.

 

The combined data for all subjects (exemplified in Figs. 1 2 ) show a common course (the two film characteristics). The 20 plotted values for a suspension of latex spheres (circles) are calibrated to yield absolute values for the horizontal axis (Rayleigh ratios in m⁻¹ Sr⁻¹). Continuous lines show the fitted functions for the two film characteristics. To the right, the densities are indicated of the LOCS III NO standards NO1 through NO6 (−1.76, −0.97,− 0.76, −0.58, −0.43, and −0.39, respectively).

Figure 4.

 

Calibration curves to derive log Raleigh ratio (vertical) from film density (horizontal). The curves are slightly different for the two film speeds. For either film speed, the curves for different flash/slit width settings differ only in vertical position. In the example given for 1600 ASA, the 0.1-mm curve is plotted at precisely a factor of 2 (0.3 log unit) higher than the 0.2-mm curve.

Figure 5.

 

Calibration curves to derive log Raleigh ratio (vertical) from the LOCS III NO score (horizontal). The curves are slightly different for the two film speeds. For either film speed, the curves for different flash/slit width settings differ only in vertical position. In the example given for 1600 ASA, the 0.1-mm curve is plotted at precisely a factor of 2 (0.3 log unit) higher than the 0.2-mm curve.

Figure 6.

 

Comparison between log Rayleigh ratios obtained respectively with the present photographic/densitometric method and an independent measurement using a calibrated Scheimpflug system. One eye of 19 subjects was photographed with a Zeiss (stars) and a Topcon (open squares) slit lamp. Nineteen eyes of 10 older subjects were photographed with a Topcon slit lamp (filled squares).

Figure 7.

 

Comparison between log Rayleigh ratios obtained with the present photographic/LOCS III–based scoring system and an independent physical measurement at 500 nm (open squares) and at 602 nm (stars), respectively. The continuous lines represent the ideal case.