Degradation of Visual Performance With Increasing Levels of Retinal Stray Light

Abstract

Purpose: To quantify the effect of induced stray light on halo size, luminance threshold, and contrast sensitivity.

Methods: Retinal stray light was induced in five healthy subjects using different photographic filters. The stray light induced ranged from levels observed in intraocular lenses (IOLs) with glistenings (low) to cataract level (high). The visual impact was measured for halo size, luminance detection threshold, and contrast sensitivity with and without a glare source.

Results: The amount of retinal stray light induced by the different filters was similar when measured using the psychophysical method and the optical bench method. Low amounts of induced stray light cause the halo size to increase by 21%, the luminance detection threshold to increase by 156%, and contrast sensitivity to decrease by 10% to 21% dependent on spatial frequency and presence of a glare source. The visual impact percentages for high amounts of induced stray light were, respectively, 76%, 2130%, and 30% to 49%. In the presence of a glare source, contrast sensitivity losses were larger and shifted to lower spatial frequencies.

Conclusions: Low levels of retinal stray light can cause significant increases in halo sizes, elevations in luminance detection thresholds, and reductions in contrast sensitivity whether or not a glare source is present.

Some of the most significant visual disturbances affecting cataract patients are caused by retinal stray light.¹ Nevertheless, the visual impact of relatively low levels of stray light can be difficult to quantify, as it does not cause a drop in visual acuity.² The diagnoses of these visual problems are most commonly made using slit-lamp exams of potential sources of stray light. Nowadays, direct psychophysical measurements of stray light levels are also available as commercial instruments.³ However, the visual impact of measured stray light with these instruments is still not well understood. The aim of this study was therefore to quantify the effect of stray light on three measures of visual quality. 

The level of retinal stray light determines the level of disability glare and can be a reason for persistent visual complaints.^1,3 Scatterers in the crystalline lens or micro-vacuoles present in the optic body of an intraocular lens (IOL) are sources that can cause retinal stray light. The resulting disability glare may constitute visual complaints like hazy or blurry vision leading to cataract extraction or IOL explantation.^4,5 The stray light behavior of the eye as a function of age, pigmentation, and angle has been studied extensively,^6–8 and the influence of age can be used to illustrate the effect of stray light in cataract.⁸ Furthermore, the stray light behavior of micro-vacuoles or glistenings has been described in the past for four different types of IOLs.⁹ However, retinal stray light is currently not routinely measured in clinical practice as a way to diagnose cataract or glistenings; the opacification of the crystalline lens or the micro-vacuoles in the optic body are routinely assessed only in a slit-lamp exam. This exam subjectively illustrates the effect of backward scattered light from the opacities, while it is the forward light scattering distributed over the retina that determines the visual impairment. Therefore, there is a need to relate visual complaints of patients and the corresponding slit-lamp exam to understand visual performance measures and clarify the visual impairment caused by retinal stray light. 

In clinical practice, vision is assessed most commonly with visual acuity (VA) and occasionally with contrast sensitivity tests. Cataract patients may have uncompromised VA but may still complain of poor vision.¹⁰ The clinician's ability to accurately diagnose the complaints is thereby limited. Visual acuity and retinal stray light are known to be independent² because they assess different aspects of vision. Visual acuity depends on foveal image quality of high-contrast details and therefore does not change with increasing stray light until excessive levels are reached, whereas retinal stray light determines the level of disability glare as defined by CIE.¹¹ Disability glare can be defined as the contrast reduction of a visual scene due to the retinal veiling luminance induced by a glare source in the field of view. 

Several studies found decreased contrast sensitivity for all, or part of, the spatial frequencies reported in the presence of cataract, also without a glare source.^10,12–16 The level of decrease in contrast may depend on the cataract morphology investigated: cortical, nuclear, or posterior subcapsular.^12–15 Similarly, five studies have shown decreased contrast sensitivity for the higher spatial frequencies when glistenings are present in the body of the IOL optic,^17–21 and two studies were nonconclusive.^22,23 Despite these studies, general understanding of the implication of these relatively low levels of retinal stray light on visual function is limited and uncertain even in cases when retinal stray light is measured and quantified.^4,23–25 Therefore, in this study, an improved contrast sensitivity test, with and without the presence of a glare source, and a luminance detection threshold test are used to create quantifiable visual performance measures of the impact of retinal stray light. Halo size was chosen as another visual performance measure because its clinical significance is intuitively understandable. In daily life, even low amounts of scatter may cause significant visual problems if glare sources are present. In this investigation, we systematically induced low degrees of scatter, typical of the levels found for glistenings and moderate degrees of cataracts, and measured the visual effects with and without the presence of glare sources. The objective of this study was to determine the visual consequences of scatter by relating varying levels of stray light to three measures of the quality of vision: halo size, luminance threshold, and contrast sensitivity. 

Methods

In this investigation, measurements were performed on five right eyes of five healthy subjects between the ages of 28 and 53 years at the Royal Institute of Technology, Stockholm. Four subjects were emmetropic, and one habitually wore contact lenses (−2.50 diopters [D]). The study conformed to the tenets of the Declaration of Helsinki and was approved by the regional ethics committee in Stockholm, Sweden (2013/1433-31/1). Written informed consent was obtained prior to the start of the study. 

The scatter parameter s as function of angle θ is defined as s(θ) = PSF(θ) * θ², where PSF is the point spread function.⁷ The scatter parameter is an important and visually relevant measure of performance for angles larger than 1°, where the PSF is dominated by ocular scatter sources like cataract or glistenings. This part of the PSF can also be described by photometric quantities as the ratio between veiling luminance on the retina and illuminance of the glare source at the pupil plane.⁷ 

In this experiment, three photographic filters, Black Pro-Mist (BPM) filters BPM¼, BPM1, and BPM3 (The Tiffen Company, Hauppauge, NY, USA), were chosen to induce stray light ranging from s = 3 deg²/sr to s = 30 deg²/sr, representing glistening and cataract cases.^2,9,26 The filters were characterized using an optical bench–based technique that has been described in detail previously.²⁷ This optical bench method measures the stray light parameter, s, as a function of angle θ. Furthermore, retinal stray light at an average angle of 7° was measured using the compensation comparison method implemented in the C-quant (Oculus, Wetzlar, Germany).³ In addition to the standard setup, a C-quant was modified by extending the distance of the eyepiece so that the stray light could be measured for an average angle of 2.5°. The C-quant provides in vivo measures of the logarithm of the stray light parameter log(s). 

Halo Size

The halo size was measured using the Rostock Glare Perimeter.²⁸ In the Rostock Glare Perimeter, a small square marker is detected as it moves outwardly from a central glare source. The subject fixates on the glare source and indicates when the marker becomes visible, which will reproduce the psychophysical halo radius using the method of adjustment. The square marker has a side length of 0.1° and a luminance of 25 cd/m², and the subject sits 3 m away. The glare source has an illuminance of 0.4 lux at the eye, and the procedure is repeated three times in 12 different meridians. The perceived retinal halo is constructed by connecting all measurement outcomes from each meridian and averaging the halo radius R to determine log(R) for each subject and each stray light level. 

Luminance Threshold

The luminance detection threshold was determined using a modified Rostock Glare Perimeter with a novel procedure developed for this study. In this procedure, a two-alternative forced-choice method was used to determine the detection threshold at 2.5° angular distance to the glare source. The task of the subject was to indicate the location of the 0.1° square stimulus, which appeared for 0.5 second above or below the glare source. The luminance threshold of the marker was determined by an adaptive Bayesian algorithm, and 50 trials were used to find the luminance level at which the subjects would indicate the correct location of the marker with 75% probability. The procedure took less than 2 minutes. The subject was located 2.5 m from the central glare source, which had an illuminance of 0.75 lux at the eye. The procedure was repeated three times for each subject and condition, and the logarithm of the luminance, log (L), at the threshold was used as the outcome measure. 

Contrast Sensitivity

The complete contrast sensitivity function (CSF) was measured in 100 trials with the quick CSF method²⁹ using a calibrated cathode ray tube screen displaying Gabor gratings with a mean luminance of 48 cd/m². The CSF was also measured with a glare source placed in the horizontal nasal field at 2.5° and 7°, respectively, with an illuminance of 12 lux at the eye. A range of contrast sensitivity levels was determined starting from a spatial frequency of 1 cycle per degree (cyc/deg), and the area under the logarithm of the contrast sensitivity curve AULCSF, using the logarithm of the spatial frequency, was determined for each subject and stray light level. The procedure took less than 4 minutes for one condition. 

Analysis

For each subject, the induced stray light of the filters as measured with the C-quant was calculated by subtracting the stray light level without any filter from the stray light level with each filter. The average measured degree of induced stray light for all subjects was compared to the optical bench measurement result for each filter of the linear stray light parameters for 2.5° and 7°. The stray light levels were then correlated with each visual outcome for the angles of 2.5° and 7°. Because the same subject was measured with different filters, the within-subject data points are interdependent. Therefore, the relationship between all stray light outcomes and visual performance outcomes was fit using the partial least squares (PLS) method Modde 5 (Umetrics AB, Umea, Sweden) to isolate the effect of stray light from any subject-dependent effect. The confidence level was set to 95%. 

Results were compared for a typical baseline stray light value, which is 1.1 log(s) for a 60-year-old healthy or a typical pseudophakic subject.² To determine the stray light effect on all measured visual performance outcomes caused by glistening (low) and cataract (high), the additional induced stray light levels are enlarged by 0.15 log(s) for low amounts and by 0.40 log(s) for high amounts. 

Results

The stray light characteristics measured on the optical bench were comparable to the average induced stray light levels as measured with the extended and standard C-quant in the five subjects. The photographic filters induce stray light levels ranging from 3 to 30 deg²/sr and are therefore able to simulate the stray light effects caused by glistenings in IOLs and by cataract cases. 

A night driving scene as perceived by a typical pseudophakic case was used to simulate the visual impact for a low, medium, and high amount of additional induced stray light. The headlights were assumed to be glare sources giving an illuminance of 0.6 lux with an average background luminance of 1 cd/m². The images themselves have a size of 20° (Supplementary Movie). The pedestrians are less visible because of the increasing halo of the oncoming car. 

The PLS fit showed that for all visual test outcomes there were no fundamental interactions between the observed stray light levels and the subjects tested. This also explains why the linear regression yielded correlation results similar to those obtained with the PLS method. The reported slopes were calculated using the standard linear regression method. 

Halo Size

The measured halo size was highly correlated with the level of stray light at both 2.5° and 7° (R² = 0.75 and R² = 0.79, respectively) (Fig. 1). The halo size increased with a slope of 0.55 log units for one log(s) unit increase in stray light at 2.5°. For one log(s) unit increase in stray light at 7°, the halo radius increased with 0.61 log units. This means that the halo size measurement is approximately half as sensitive as the stray light measurement regardless of the angle for which the stray light is determined. Recalculated, this means that for a 0.15 log(s) stray light increase at 2.5°, the halo radius increased by 21%, and for a 0.40 log(s) stray light increase at 7°, the halo radius increased by 76%. 

Figure 1

 

Halo size as function of stray light level (left) at 2.5°, and (right) at 7° for five subjects. AS, average slope.

Luminance Threshold

The measured luminance threshold as a function of stray light level at 2.5° and 7° had a high correlation (R² = 0.73 and R² = 0.71, respectively) (Fig. 2). The luminance threshold increased with a slope of 2.72 log units for one log(s) unit increase in stray light at 2.5°. For one log(s) unit increase in stray light at 7°, the luminance threshold increased with a slope of 3.37 log units. The luminance threshold measurement is therefore almost four times as sensitive as the stray light measurements at 7°. For a 0.15 log(s) stray light increase at 2.5°, the luminance threshold increased by 156%; and for a 0.4 log(s) stray light increase at 7°, the luminance threshold increased by 2130%. This illustrates the detrimental effect of cataract by showing that there is a more than 20-fold loss in luminance threshold for a moderate cataract case. 

Figure 2

 

Luminance threshold as function of stray light level (left) at 2.5°, and (right) at 7° for five subjects. AS, average slope.

Contrast Sensitivity

For all measured conditions, AULCSF was highly correlated with induced stray light for each subject, but the baseline level varied between individuals. This trend is displayed by the average slope of AULCSF of all five individuals as function of stray light level after correcting for the interindividual offsets. For one log(s) unit increase in stray light at 2.5°, AULCSF without a glare source and with a glare source located at 2.5° decreased by 0.48 to 0.87 log units, respectively. Correcting for the interindividual offsets yielded high correlations (R² = 0.70 and R² = 0.83, respectively) (Fig. 3). For one log(s) unit increase in stray light at 7°, AULCSF without and with a glare source located at 7° decreased by 0.59 and 0.67 log units, respectively, with a high degree of correlation (R² = 0.78 and R² = 0.64, respectively) (Fig. 4). The retinal stray light effect on contrast sensitivity (CS) was largest for a glare source located at 2.5°. 

Figure 3

 

Area of contrast sensitivity (AULCSF) as function of stray light level at 2.5° for (left) without glare source and (right) with glare source at 2.5° for five subjects. AS, average slope.

Figure 4

 

Area of contrast sensitivity (AULCSF) as function of stray light level at 7° for (left) without glare source and (right) with glare source at 7° for five subjects. AS, average slope.

The average effects of a low amount of induced stray light and a high amount of induced stray light on the CS as a function of spatial frequency follow the same trends; however, the degree of severity differs (Fig. 5). The effect of increased stray light on CS without a glare source is largest for spatial frequencies between 7.5 and 25 cyc/deg. When stray light is elevated by the level associated with a low amount of induced stray light, the CS for this spatial frequency range decreases between 14% and 20%; and when stray light is elevated by a high amount of induced stray light, the CS for this spatial frequency range decreases between 30% and 42%. The effect of stray light on CS with a glare source present is largest for spatial frequencies between 1.5 and 10 cyc/deg. When stray light is elevated by the level associated with a low amount, the CS for this spatial frequency range decreases between 10% and 21%; and when stray light is elevated by a high amount, the CS for this spatial frequency range decreases between 34% and 49%. Contrast sensitivity was reduced by a maximum of 20% at 15 cyc/deg for a low amount of induced stray light and 42% at 18 cyc/deg for a high amount of induced stray light. These percentages were 21% at 3.5 cyc/deg and 49% at 3 cyc/deg when contrast was measured in the presence of a glare source. The presence of a glare source further decreases the CS when stray light levels are elevated. Additionally, the presence of the glare source shifts the CS decrease to the lower spatial frequency range. 

Figure 5

 

Average baseline contrast sensitivity (solid blue line), contrast sensitivity for low amount of induced stray light (dotted gray line), and contrast sensitivity for high amount of stray light (dashed red line). The left graph shows contrast sensitivities without glare source and the right with a glare source present at 2.5°.

Discussion

The aim of this study was to develop a better understanding of the functional effects of increased retinal stray light caused by various visual disruptions on visual performance. We found a significant and consistent impact of stray light on all three visual tests for all subjects. The stray light effects induced by glistenings are predominantly distributed around two retinal peaks located at an angle of 2.5° and 15°.⁹ This means that the standard C-quant is not the most suitable instrument to detect the effects of glistenings, as it operates with a glare source located at an average angle of 7°. The extended C-quant used in this study, which performs stray light measurements at an average angle of 2.5°, is therefore a better instrument for measuring these effects. Another method used to measure the stray light from glistenings is the Scheimpflug technique,^24,25 which quantifies backward light scatter. However, the perception by the patient depends on the forward (retinal) light scatter. Forward stray light levels can be as much as 300 times larger than that of the measured backward scatter due to the fact that glistenings induce Mie scatter.⁹ The double-pass technique has the potential to estimate the forward scatter. However, the currently available double-pass method provides a scatter index that is defined only within the central angular range of 0.3°,³⁰ and was therefore not suitable for this study. A new development in the double-pass technique shows the potential to capture a field of 8°.³¹ For cataract cases the forward scatter between 1° and 30° is uniformly distributed over the retina,²⁶ and the ratio between forward scatter and backward scatter in cataract has been shown to be on average a factor of approximately 2.3,³² which is much less than that shown for glistenings due to the smaller size (0.7 μm) of the responsible scatterers in cataract.³³ 

The results of this study show that increased retinal stray light causes decreased visual performance for all of the tests performed, even for stray light levels shown to be typical of those induced by glistenings in IOLs. The relevance of the chosen visual performance tests is supported by the settings for the level of illuminance at the eye and the level of the stimulus luminance. Increased halo size has been reported to be correlated to forward light scatter.³⁴ Halo size measurement provides a psychophysical test of an otherwise highly subjective factor.³⁵ The luminance threshold test is a detection task intended to capture the impact of stray light in low-light environments, where both glare source and stimuli are of low strength. Degraded performance for this type of visual task may represent a risk factor under low-light conditions such as night driving. The luminance threshold test was found to be the most sensitive visual performance test executed, and shows the dramatic effect that stray light may have under these conditions. The measure illustrates, for example, that a 20-fold increase in luminance may be necessary under low-light conditions in order to detect a pedestrian crossing the street in the presence of an oncoming car. In light of this fact, the complaints of some cataract patients who have adequate VA may be better understood. 

Finally, the CS test measures visual capability in an environment with higher luminance, where resolution tasks are required. Contrast sensitivity with and without glare are also relevant factors in driving.³⁶ Contrast sensitivity losses of 14% to 20%, as induced by low amounts of retinal stray light, have been predicted.⁹ This demonstrates the clinical relevance of small elevated levels of retinal stray light. The retinal stray light effect on CS was largest with a glare source present at 2.5°, causing a veiling luminance deteriorating the contrast of the image on the retina. The effect for a glare source located at 7° is less due to the limited size of the halo radius. Individual variation in the relationship found between CS and stray light, as was measured in this study, is expected among the subjects, as age and neural factors also contribute to the CSF. Decreased CS has been reported in a number of IOL studies with and without glistening,^17–21 and was of the same order of magnitude as measured in our study with the BPM¼ filter. The quick CSF method employed in this study allows for a more sensitive measurement of the CS than the 40% between consecutive levels used in standard contrast vision tests. This may explain why contrast loss has not always been concluded in IOL studies on glistenings. Contrast sensitivity loss with a higher magnitude was found for some types of cataracts.^10,12–20 In particular, these studies showed that the CS loss was largest for the cataract type posterior subcapsular opacity and less for nuclear and cortical opacities. These higher amounts of contrast losses measured in the past are of the same order of magnitude as measured in our study with the BPM3 filter. Although our study induced stray light extraocularly, measuring visual effects with filters is a fair approximation to measurements with scatterings in the eye. 

In conclusion, retinal stray light correlates strongly with the outcomes of some methods used to measure visual function. Levels of retinal stray light as induced by glistenings and cataract have a measurable and significant impact on visual function. 

Acknowledgments

Supported by the Swedish Research Council (621-2011-4094) and EUREKA Grant INT 111017. 

Disclosure: M. van der Mooren, Abbott Medical Optics Groningen BV (E); R. Rosén, Abbott Medical Optics Groningen BV (E); L. Franssen, Abbott Medical Optics Groningen BV (E); L. Lundström, None; P. Piers, Abbott Medical Optics Groningen BV (E)