Pupil size dynamics during dark adaptation, in the presence of fixation on a target

Pupil size can directly affect the amount of light that reaches retinal structures and is returned by them in full-pupil, double-pass measuring technology, such as retinal birefringence scanners (Hunter et al., 2004), scanning laser ophthalmoscopes (Webb and Hughes, 1981), scanning laser polarimeters (Weinreb et al., 1995), optical coherence tomography (OCT) devices (Huang et al., 1991) etc. Since retinal illuminance is proportional to the area of the entrance pupil (Atchison and Smith, 2000), the signals obtained by the above mentioned devices from the light reflected from the retina strongly depend on pupil size. A number of investigators have reported the relationship between luminance and pupil size. (Watson and Yellott, 2012) Yet these publications use a steady state (after adaptation), ignoring pupil dynamics, and do not mention the influence of accommodation. Numerous publications describe also the acute “lights-off” effect and the slower dark adaptation (Loewenfeld, 1993), but there appears to be very little information on how pupil size changes in the first several minutes after ambient lights are turned off and while the subject is fixating on a target, which are usually the conditions when double-pass systems are used. It is known that both fixation and accommodative effort cause pupil constriction, thus eliminating some peripherally entering rays and masking high-order monochromatic aberrations. But the interplay between this phenomenon, known as accommodative pupillary constriction (Wolffsohn et al., 2006; Charman and Radhakrishnan, 2009), and the competing, or rather counteracting pupil dilation due to dark adaptation, is not well studied. The goal of this study was to determine the extent of pupil size changes, and to potentially find a time window during which the pupil size is maximal, to allow best conditions for obtaining information from the retina at maximum signal-to-noise ratio.

Figure 1. Optical setup used to measure the pupil size.

We studied 5 test subjects, age 28-60, all properly consented. After a period of 10 minutes of room-light adaptation, the subjects were asked to fixate on a white-light partially accommodative target (a red dot with a white border, 3x1.5 mm), optically 33 cm from the eye. The target was front-on illuminated constantly by a faint electric bulb, providing background luminance in the area of the target of about 1.10^-2 cd/m², just enough to enable the test subject to fixate. The ambient illumination was turned down immediately after initiation of the recording, from 27 cd/m² to about 2.10^-3 cd/m². Pupil diameter was measured under monocular conditions (with one eye occluded) by means of an eye tracking apparatus (Ramey et al., 2008) using video-oculography and comprising an infrared-sensitive USB video camera (240x320 pixel resolution; Web Digital Camera, Hong Kong) equipped with a 12 mm fixed-focal-length lens (Figure 1). Near- infrared illumination of the pupil was provided by an infrared light emitting diode (OD-50L, 880 nm; Opto Diode Corp., Inc., Newbury Park, CA). The camera was connected to a desktop computer that controlled video frame capture using custom acquisition software written in MATLAB (MathWorks, Inc., Natick, MA). We used image acquisition with a frame rate of 5 fps, with continuous recording. The recorded eye’s image sequences were analyzed off-line. Pupils were approximated with circles, and their diameters were calculated with commercial eye tracking software (IRIS; Chronos Vision, Berlin, Germany). Pupil detection uses edge detection and the Hough transform (Duda and Hart, 1972; Ballard, 1981) to identify a circle in a parameterized space. Blinks were detected as abrupt drops of more than 30% in pupil diameter, lasting for 200-400 ms), and were replaced by the preceding value. Pupil area was calculated based on the diameter measured from each frame. In order to compare pupil behavior across test subjects, and possibly derive a general trend, the pupil area traces were normalized:

(1)

Where A(t) is the area measured in time, A(0) is the baseline value at the initial moment when the light was turned off, and A_n(t) is the normalized area.

Results

Figure 1. Pupil dynamics during dark adaptation Upper trace: without accommodation; lower trace: with accommodation on a target. Pupil area in mm², time in seconds.

Figure 1, upper trace, shows the non-normalized trace from one subject as pupil area vs. time, plotted over 6 minutes (360 s) after the lights were turned off, with the subject not accommodating. Figure 1, lower trace, shows the same type of curve from the same subject, now accommodating. Figure 2 shows the normalized traces of all subjects studied.

Figure 2. Normalized pupil area of all subjects studied. The time is in seconds. The dashed line shows the exponential fit of the averaged curve (please note that the Y-axis starts from 0.5).

In addition to the individual traces, the average trace is also shown (thick gray line). The dilation reaches its maximum, with area 60% above the baseline level, at a time of about 70 s. Then the average normalized curve starts descending exponentially toward the baseline.

The normalized traces were then approximated in MATLAB using a nonlinear least-squares regression fit with the following model function:

(2)

where the time t is in seconds. For each fit it was assured that the estimated coefficients fell into the 95% confidence interval using the Jacobian of function (2), returned by the fit. The coefficients a₁ and a₂ from the individual traces, as well as from the average trace, are shown in table 1.

Table 1. Estimated coefficients for the exponential fit for the individual traces and for the averaged trace

	Individual traces						Coefficients for averaged trace
	Subject 1	Subject 2	Subject 3	Subject 4	Subject 5	Averaged coefficients
a1	0.0287	0.0714	0.0199	0.0174	0.0325	0.0340	0.0310
a2	0.0058	0.0037	0.0153	0.0032	0.0042	0.0064	0.0059

There is minimal difference between the last two columns of the table, indicating that averaging the estimates of the individual approximations yields nearly the same results as the approximation on the averaged trace. The estimated curve according to equation (2) is plotted on figure 2 as a dashed black line.

Discussion

Although this study has not investigated specific clinical patient groups, it has shown that there is a definite pattern in the change of pupil size during dark adaptation and in the presence of an accommodative effort. However, there were marked inter-subject variations in pupil size progression over time, as can be seen on figure 3. We think that the first and foremost cause for this was the different level of accommodation provided by the different subjects. One likely reason for this was the different ability to accommodate, which is age dependent, and probably to some extent the use of an imperfect target, which was small but probably lacking enough detail. But there is also the direct effect of age on pupil size. One study showed that pupil diameter increases slightly across age groups between 1 and 19 years(MacLachlan and Howland, 2002) while other studies have reported that pupil size becomes smaller in an almost linear manner with increasing age.(Winn et al., 1994; Koch et al., 1991) Moreover, the rate of change with age is fastest at lower luminances, as is the present case. Yet, since our study deals with relative changes with respect to a light-adapted baseline, we observed a clear pattern of a relative fast initial increase, and then slower decrease in pupil size.

With the limited number of subjects, this study is merely a proof of concept. Investigating the presence or absence of accommodation, inter-subject variability, age-related variability, and day-to-day variability of pupil size dynamics, by means of analysis of variance, is expected to shed more light on the phenomenon studied, and will most likely lead to more precise criteria for the optimal timing of retinal scanning during dark adaptation. Of interest, a study by Bradley and coworkers showed that gender and iris color have no significant effect on the dark-adapted pupil diameter.(Bradley et al., 2010)

The mechanisms involved in the “lights-off” response are mainly the parasympathetic relaxation and sympathetic activation (Loewenfeld, 1993) causing dilation. The mechanism involved in accommodative pupillary constriction is quite different, involving changes in the accommodative state via the convergence-accommodation mechanism. The extent of influence of each of these mechanisms, and hence the location of the maximum pupillary size found by us, might well be influenced by the variable factors mentioned above, which warrants further investigation.

Algorithms may be developed for adjusting the coefficients of the exponential fit a₁ and a₂ in accordance with valid variability factors, so that the software in retinal scanning instrumentation may suggest the best possible time window for acquiring data with maximum signal-to-noise ratio.

Conclusion

We observed a certain variance between the plots, most likely attributable to a different level of accommodation attempt for the different subjects. Yet, when accommodation attempt was present, the pupil size followed a specific pattern – a sudden increase, followed by a relatively flat peak, then an exponential decay toward the baseline. Based on the signal traces in figure 2, it can be concluded that measurements between 27 s and 110 s are likely to be performed at a pupil area at least 50% larger than the baseline. The pupil size appears to be maximal at about 60 s after “lights off”. This should be taken into consideration when optimizing the time window for measurements on retinal structures with whole-pupil, double-pass systems, when subjects are fixating on a target. As shown in Table 1, in all subjects the coefficient a_1, characterizing the initial rate of pupil change, is significantly larger than coefficient a₂, which describes the slower exponential decay after reaching the maximum. This implies that it is important, after dimming the ambient light, to wait for at least 30 s before starting measurement. The optimal time window for the measurements, according to these results, is during the second minute after dimming the light.

Most of the above material has been published also in the following paper:

Gramatikov, BI, Irsch, K, and Guyton, D; "Optimal timing of retinal scanning during dark adaptation, in the presence of fixation on a target: the role of pupil size dynamics. Journal of Biomedical Optics, 2014, 19(10), 106014. doi:10.1117/1.JBO.19.10.106014.

http://biomedicaloptics.spiedigitallibrary.org/article.aspx?articleid=1921066%20&journalid=93

References

Atchison D A and Smith G 2000 Optics of the Human Eye Butterworth-Heinemann: Oxford

Ballard D H 1981 Generalizing the Hough transform to detect arbitrary shapes Pattern Recognition, Elsevier 13 111–22

Bradley J C, Bentley K C, Mughal A I, Bodhireddy H, Young R S and Brown S M 2010 The effect of gender and iris color on the dark-adapted pupil diameter J Ocul Pharmacol Ther 26 335-40

Charman W N and Radhakrishnan H 2009 Accommodation, pupil diameter and myopia Ophthalmic Physiol Opt 29 72-9

Duda R O and Hart P E 1972 Use of the Hough Transformation to Detect Lines and Curves in Pictures. Comm. ACM 15 11-5

Huang D, Swanson E A, Lin C P, Schuman J S, Stinson W G, Chang W, Hee M R, Flotte T, Gregory K, Puliafito C A and Fujimoto J G 1991 Optical coherence tomography Science 254 1178- 81

Hunter D G, Nassif D S, Piskun N V, Winsor R, Gramatikov B I and Guyton D L 2004 Pediatric Vision Screener 1: instrument design and operation. Journal of Biomedical Optics 9 1363-8

Koch D D, Samuelson S W, Haft E A and Merin L M 1991 Pupillary size and responsiveness. Implications for selection of a bifocal intraocular lens Ophthalmology 98 1030-5

Loewenfeld I E 1993 The Pupil: Anatomy, Physiology, and Clinical Applications Iowa State University Press and Wayne State University Press: Detroit

MacLachlan C and Howland H C 2002 Normal values and standard deviations for pupil diameter and interpupillary distance in subjects aged 1 month to 19 years Ophthalmic Physiol Opt 22 175-82

Ramey N A, Ying H S, Irsch K, Mullenbroich M C, Vaswani R and Guyton D L 2008 A novel haploscopic viewing apparatus with a three-axis eye tracker J AAPOS 12 498-503

Watson A B and Yellott J I 2012 A unified formula for light-adapted pupil size J Vis 12 12

Webb R H and Hughes G W 1981 Scanning laser ophthalmoscope IEEE Trans Biomed Eng 28 488-92

Weinreb R N, Shakiba S and Zangwill L 1995 Scanning laser polarimetry to measure the nerve fiber layer of normal and glaucomatous eyes Am J Ophthalmol 119 627-36

Winn B, Whitaker D, Elliott D B and Phillips N J 1994 Factors affecting light-adapted pupil size in normal human subjects Invest Ophthalmol Vis Sci 35 1132-7

Wolffsohn J S, Ukai K and Gilmartin B 2006 Dynamic measurement of accommodation and pupil size using the portable Grand Seiko FR-5000 autorefractor Optom Vis Sci 83 306-10