Abstract
Purpose.:
We investigated effects of pupil shifts, occurring with changes in luminance and accommodation stimuli, on refraction components and higher-order aberrations.
Methods.:
Participants were young and older groups (n = 20; 22 ± 2 years; age range, 18–25 years; and n = 19, 49 ± 4 years, 45–58 years, respectively). Aberrations/refractions at 4- and 3-mm diameters were compared between centered and decentered pupils for low (background, 0.01 cd/m2, 0 diopters [D]), and high (6100 cd/m2, 4 or 6 D) stimuli. Decentration was the difference between pupil centers for low and high stimuli. Clinical important changes with decentration were: M at ±0.50 or ±0.25 D, J 180 and J 45 at ±0.25 or ±0.125 D, HORMS at ±0.05 μm, C(3, 1) at ±0.05 μm, and C(4, 0) at ±0.05 μm.
Results.:
Because of small pupil shifts in most participants (mean 0.26 mm), there were few important changes in most refraction components and higher-order aberration terms. However, M changed by >0.25 D for a third of participants with 4-mm pupils. When determining refractions from second to sixth order aberration coefficients, the more stringent criteria gave 76/534 (14%) possible important changes. Some participants had large pupil shifts with considerable aberration changes. Comparisons at the high stimulus were possible for only 11 participants because of small pupils. When refractions were determined from second order aberration coefficients only, only 35 (7%) had important changes for the more stringent criteria.
Conclusions.:
Usually pupil shifts with changes in stimulus conditions have little influence on aberrations, but they can with high shifts. The number of aberrations orders that are considered as contributing to refraction influences the proportion of cases that might be considered clinically important.
The study complied with the tenets of the Declaration of Helsinki and was approved by the University Human Research Ethics Committee. Experimental methods were described in detail by Mathur et al.
1 Participants were staff and students of Queensland University of Technology in good general and ocular health, with best corrected visual acuities ≥ 6/6, spherical equivalent refractions > −3 diopters (D), and cylinder ≤ 0.75 D. There were 20 young participants (mean age, 22 ± 2 years; spherical equivalent, −1.5 ± 0.9 D) and 19 older participants (mean age, 49 ± 4 years; spherical equivalent, −1.8 ± 1.6 D). Pupil images and aberrations were determined with a modified COAS-HD Hartmann-Shack aberrometer (Wavefront Sciences, Inc., Albuquerque, NM, USA) with room lights off and the nontested eyes occluded. A matrix of stimulation conditions were used in which there were 4 luminance levels between 0.01 cd/m
2 (level 1) and 6100 cd/m
2 (level 4), of a 12.5° × 11.0° background, and up to 4 accommodation stimulus levels (0, 2, 4, and 6 D) provided by moving the internal target. Three measurements were taken for each luminance-accommodation stimulus combination. Accommodation stimuli were increased until the participant reported that the target could no longer be made clear, up to a maximum of 6 D. Eye images were analyzed using an algorithm that estimated
x,
y coordinates of the pupil center relative to the limbus center. Nasal and superior pupil center positions were taken as positive.
To determine uncertainty associated with determining pupil centers, two images at a randomly selected luminance/accommodation combination were analyzed for each of five young and five older participants. Each image was analyzed three times and the absolute distance of the pupil center from the limbus center was obtained. The standard deviations of the three analyses and the absolute difference between the averages for the two images were determined. Across all participant/image combinations, the mean of the standard deviations of three analyses was 0.04 ± 0.03 mm. Across all participants, the mean absolute differences between first and second images was 0.03 ± 0.03 mm. This indicated that pupil shifts determined of 0.05 mm or more are meaningful.
We did comparisons at conditions giving the largest and smallest pupil sizes. The largest pupil sizes occurred for the luminance level 1–0 D accommodation condition, henceforth referred to as the “low stimulus condition.” For most of the younger group the smallest pupil size was determined for the luminance level 4–6 D accommodation condition, and for the majority of the older group the smallest pupil size was determined for the luminance level 4–4 D accommodation condition. Some people could not make the target of the aberrometer appear clear at these accommodation stimuli, and in these cases the determinations were made for 4 D stimulus in one case for the younger group and for 2 D stimulus in eight cases for the older group. The high luminance and high accommodation combination will be referred to as the “high stimulus condition.”
As pupil sizes were small at the high stimulus condition, we decided to do analyses for 4.0 and 3.0 mm diameter pupils. For the low stimulus condition, we determined the aberrations at these pupil sizes when the data were centered and when they were rereferenced to the pupil center of the high stimulus condition for each participant. For the high stimulus condition, we determined the aberrations when the data were centered and when they were rereferenced to the pupil center of the low stimulus. The rereferencing for the low stimulus condition is particularly relevant as the low stimulus condition, with relatively large pupils, is the one that usually is used to determine aberrations at smaller pupil sizes.
Aberrations up to the sixth order were determined from the positions of the spots (black spots in
Fig. 2) in the Hartmann–Shack images using custom software. For analyzing in the centered case, a subset of spots (blue spots overlaid over black spots) was used that matched the pupil size of interest (blue ring). For analyzing in the decentered case, another subset of spots (red spots overlaid on blue and black spots) was selected around the new reference point to match the pupil size of interest (red ring).
Often, analyses were not valid at the high stimulus condition, because the size at which analyses were made (3.0 or 4.0 mm) was not attained either by the natural pupil or by the effective pupil when the effect of decentration was studied. To explain the pupil size limitations further, we present two examples. Firstly, the pupil size might be 3.5 mm compared to the reference pupil size of 4.0 mm. Secondly, say the pupil size is 3.5 mm compared to the reference pupil size of 3.0 mm, and a decentration of 0.3 mm is required (see
Fig. 3). The effective pupil size = 3.5 − 2*0.3 = 2.9 mm. When the effective pupil size was smaller than the reference pupil size, we extrapolated the aberrations of the former to match the latter using our algorithm for this purpose; this was considered to be valid in two cases where the effective pupil sizes were ≤ 0.1 mm smaller than the reference pupil size.
Aberrations were referenced to 550 nm. In the results, we show changes in mean spherical equivalents (
M), regular astigmatism (
J 180), oblique astigmatism (
J 45), higher-order root–mean squared aberrations (
HORMS), horizontal coma coefficients (
C[3, 1]), and spherical aberration coefficients (
C[4, 0]). We have used the ANSI/ISO system of specifying aberration coefficients.
14 M,
J 180, and
J 45 were determined by combining second to sixth order aberration coefficients and by considering only the second order aberration coefficients.
Results are shown in
Figure 4 (young group, 4-mm pupil),
Figure 5 (young group, 3-mm pupil),
Figure 6 (older group, 4-mm pupil), and
Figure 7 (older group, 3-mm pupil). Scales have been chosen that include the maximum and minimum values across all group, pupil size, and stimulus combinations. We have chosen the following changes to represent clinically important changes in refraction upon decentration:
M at ±0.50 or ±0.25 D,
J 180 and
J 45 at ±0.25 or ±0.125 D, and
HORMS,
C(3, 1), and
C(4, 0) at ±0.05 μm. The more stringent refraction criteria were selected as they match prescription intervals. These limits are given in the figures by dotted lines. As mentioned earlier, for many cases, the high stimulus results were invalid because pupil sizes were too small.
All but one participant had valid data, whether centered or decentered, for the low stimulus condition and both pupil sizes. Only one participant had valid data for the high stimulus condition and 4-mm pupils. Several participants had valid data for the high stimulus condition and 3-mm pupils, but only six young and five older participants had valid results for centered and decentered conditions, while three young and eight older participants had valid results for only the centered condition.
Using M as ±0.50 D with J 180 and J 45 as ±0.25 D gives only 33 important changes, while using the more stringent criteria of M as ±0.25 D and J 180 and J 45 as ±0.125 D gives only 76 important changes. This is of 534 possible cases with valid comparisons; that is, where effective pupil size meets the reference size for centered and decentered situations. However, within the important changes were some interesting cases.
For the younger group with 4-mm pupils and the low stimulus condition, and using the more stringent criteria for the refraction components, 25 cases had important changes (
Fig. 4). These were spread mainly between Δ
M (nine cases) and Δ
HORMS (five cases) The most noticeable change was Δ
M = −1.1 D for participant 1, for which there was a decentration of 0.25 mm (large filled circle in
Fig. 4). This participant also had Δ
J 45 −0.3 D.
For the younger group with 4-mm pupils and the high stimulus condition, valid comparisons were possible for only one subject for which no significant changes were found (
Fig. 4).
For the younger group with 3-mm pupils and the low stimulus condition, and using the tighter tolerances for the refraction terms, only 10 cases had important changes (
Fig. 5). The most noticeable change was Δ
M = −1.2 D for young participant 1 (large filled circle), similar to this person's result for the 4-mm pupil.
For the younger group with 3-mm pupils and the high stimulus condition, valid comparisons at the high stimulus condition were possible for six subjects (
Fig. 5), among which there were 13 cases with important changes. The most notable changes were for participant 17 (large open circle), which included Δ
M at +0.6 D, Δ
J 180 at −1.0 D, Δ
J 45 at +0.5 D,
HORMS at −0.20 μm, and
C(3,1) at −0.28 μm, which were the largest changes occurring for these aberrations. This participant had a particularly large pupil decentration of 0.33 mm. Interestingly, aberration changes for the low stimulus condition for this subject were small.
For the older group with 4-mm pupils and the low stimulus condition, and using the more stringent criteria for the refraction terms, seven cases had important changes, three of which involved Δ
J 180 (
Fig. 6). For the older group with 3-mm pupils and the low stimulus condition, and using the more stringent criteria for the refraction terms, 14 cases had important changes, most involving the astigmatism terms (
Fig. 7).
For the older group with 4-mm pupils and the high stimulus condition, valid comparisons were not possible. For the older group with 3-mm pupils and the high stimulus condition, valid comparisons at the high stimulus condition were possible for five subjects (
Fig. 7), among which there were seven cases with important changes.
Because of the small pupil centration changes with change in stimulus condition (from low luminance/low accommodation stimulus to high luminance/high accommodation stimulus) for most participants, there were mainly small changes in refraction and higher-order aberrations that would be considered to have no clinical importance. However, for approximately a third of participants with 4-mm pupils, mean spherical refraction changed by more than 0.25 D. There were a few cases where changes in refraction and/or higher order aberrations were considerable.
When determining refractions from second to sixth order aberration terms, the majority of important shifts, 48 of the 76 across both pupil sizes and using the more stringent refraction criteria, occurred for the younger participants. Given the small number of subjects for whom comparisons could be made at the high stimulus condition, the proportion of important shifts under this condition (20/66) was considerable, possibly reflecting considerable higher-order aberrations for some of the centered pupils at this level.
When refractions were determined from second order aberration terms only, the number of clinically important changes in refraction and higher order aberrations was only 37, with only eight values with the 3-mm pupil. Determining refraction using only second order terms probably is better than using more orders at smaller pupil sizes, as higher order terms can be rather “noisy” and have undue influence on refraction.
This study has implications for clinical refraction using aberrometers. Occasionally, pupil sizes might be considerably different during aberrometer refraction and subjective refraction, with accompanying different pupil centers, such as when the subjective measurements are under photopic conditions and the aberrometer measurements are under mesopic lighting conditions. This may give important refraction “errors” with aberrometry, which will be influenced by the number of orders of aberrations considered as contributing to refraction.
The authors thank Julia Gehrmann for technical assistance.
Supported by Australian Research Council (ARC; Canaberra, Australia) Discovery Grant DP110102018 (DAA), ARC Linkage Grant LP100100575 (DAA), and by Carl Zeiss Vision.
Disclosure: D.A. Atchison, None; A. Mathur, None