|Home | About | Journals | Submit | Contact Us | Français|
Clinical assessment of pupil appearance and pupillary light reflex (PLR) may inform us the integrity of the autonomic nervous system (ANS). Current clinical pupil assessment is limited to qualitative examination, and relies on clinical judgment. Infrared (IR) video pupillography combined with image processing software offer the possibility of recording quantitative parameters. In this study we describe an IR video pupillography set-up intended for human and animal testing. As part of the validation, resting pupil diameter was measured in human subjects using the NeurOptics™ (Irvine, CA, USA) pupillometer, to compare against that measured by our IR video pupillography set–up, and PLR was assessed in guinea pigs. The set-up consisted of a smart phone with a light emitting diode (LED) strobe light (0.2 s light ON, 5 s light OFF cycles) as the stimulus and an IR camera to record pupil kinetics. The consensual response was recorded, and the video recording was processed using a custom MATLAB program. The parameters assessed were resting pupil diameter (D1), constriction velocity (CV), percentage constriction ratio, re-dilation velocity (DV) and percentage re-dilation ratio. We report that the IR video pupillography set-up provided comparable results as the NeurOptics™ pupillometer in human subjects, and was able to detect larger resting pupil size in juvenile male guinea pigs compared to juvenile female guinea pigs. At juvenile age, male guinea pigs also had stronger pupil kinetics for both pupil constriction and dilation. Furthermore, our IR video pupillography set-up was able to detect an age-specific increase in pupil diameter (female guinea pigs only) and reduction in CV (male and female guinea pigs) as animals developed from juvenile (3 months) to adult age (7 months). This technique demonstrated accurate and quantitative assessment of pupil parameters, and may provide the foundation for further development of an integrated system useful for clinical applications.
The pupillary light reflex (PLR) is a unique physiological mechanism that allows the examination of the function of the autonomic nervous system (ANS) gated by a light stimulus (McLeod and Tuck, 1987; Neuhuber and Schrödl, 2011; Kaiser et al., 2014; McDougal and Gamlin, 2015). Pupil assessment has been proposed in the evaluation of cholinergic deficiency syndromes such as Alzheimer’s disease and Parkinson’s disease (Takagi et al., 1999; Granholm et al., 2003; Fotiou et al., 2009; Chang et al., 2014). An abnormal PLR may also be suggestive of degeneration or lesion in the ANS (Arendt et al., 1983; Borson et al., 1989; Scinto et al., 2001). Examples of abnormal pupil reflex resulting from neurological problems includes Adie’s tonic pupil, Argyll Robertson Pupil and Horner’s syndrome. Furthermore, dilated pupils that are unresponsive or sluggish are indicative of medical conditions such as oculomotor nerve palsy, trauma, or inflammation (Caglayan et al., 2013). In a clinical setting PLR is examined by observation of the direct response elicited by light in the ipsilateral eye, and consensual response of the contralateral eye (Kaiser et al., 2014). Other relevant parameters examined include resting pupil size, and the relative afferent pupillary response (Kaiser et al., 2014). These observations are made with the use of a bright pen torch shone into the eyes, rely heavily on the judgment of an experienced clinician, and provide qualitative information only. Commercially available pupillometers are useful clinical tools that measure pupil diameter by first capturing an image of the eye, followed by quantification of the pupil diameter in the static image. However, this type of measurement is a lost opportunity to record the dynamic rate and magnitude of pupil diameter change over time. Similarly, pupil measurements in animals have been described in the literature (Pennesi et al., 1998; Dabisch et al., 2008; Mohan et al., 2013; Ostrin et al., 2014) but the parameters measured were limited to pupil size by the experimental set up or by the method of data analysis. A more informative and quantitative way of examining the pupils may be achieved by infrared (IR) videography, that allows the pupil constriction velocity (CV), re-dilation velocity (DV), pre and post illumination pupil sizes to be calculated (Miki et al., 2008; Mohan et al., 2013; Kaiser et al., 2014).
In recent years it has become increasingly common to utilize smart devices in ocular assessment due to their portability, ease of access, and versatility in the adjustment of color and brightness display. Currently smart phones are used to assess color vision, astigmatism, pupil size, macula integrity by the Amsler grid test, and fundus photography (Busis, 2010; Bastawrous, 2012; Bastawrous et al., 2012). Hence, an IR videography set-up coupled with a smart phone application seems an accessible approach in the assessment of the PLR in the clinical setting (Fotiou et al., 2000; Busis, 2010; Giza et al., 2011; Mohan et al., 2013). Such quantitative method is also more informative than the conventional technique using a pen torch, as the information could be used for future analysis and comparison.
We describe in this article the development of a stepwise protocol for an IR video pupillography set-up. This study was inspired by our previously published review article (Chang et al., 2014), which described abnormal pupil reflex in patients with Alzheimer’s disease. Our goal is to optimize an IR video pupillography set-up, which will be used to accurately assess pupil reflex in human and animal models of Alzheimer’s disease. We took into consideration the methods of original research articles that assessed human (Fotiou et al., 2000, 2009) and rodent pupils (Taylor et al., 2008; Mohan et al., 2013) to help us determine the illumination sources, video recording system and dark-adaptation period. To validate our IR video pupillography set-up, resting pupil diameter was measured in human subjects, and PLR was studied in guinea pigs, using a protocol that allows quantitative evaluation of the constriction and re-dilation response over time.
This part of the study was conducted with approval from the University of Auckland Human Ethics Committee (reference number 010966) and in accordance with the Code of Ethics of the World Medical Association (Declaration of Helsinki). Written informed consent was obtained for experimentation with human subjects.
This part of the study was conducted according to the University of Auckland Guideline for the experimental use of animals in research approved by the University of Auckland animal research committee (reference number 001138). Guinea pigs (male n = 6, female n = 5) were 12–28 weeks old. The guinea pigs were bred in open enclosures and in a social group of five or more animals. Water and food was provided ad libidum. The light intensity of the breeding room was 1388 watt per steradian for the male enclosure and 1010 watts per steradian in the female enclosure.
A custom MATLAB program (Mathworks; Natick, MA, USA) was used to define the pupil area by a best-fit ellipse, formed by the points where the highest contrast were detected using the starburst edge detection toolbox in MATLAB. The corneal reflection in the video was removed to allow the software to correctly locate the border of the pupil. The pupil center of the first frame was manually selected and the selected points were traced in the successive frame. The horizontal palpebral aperture measured for each guinea pig was used to set the scale in the custom MATLAB program (Figure (Figure1D).1D). A least square ellipse was fitted to the feature points and, 1D median filter with a window size of 10 frames was applied to the raw data to remove sampling noise. The custom MATLAB program outputs a spreadsheet with each frame number corresponding to a pupil diameter value. We assessed the following parameters, using the following equations:
The parameters were calculated using the following equations:
Parameters were statistically analyzed using one-way analysis of variance (ANOVA) of the means of two independent groups for the influence of one categorical unrelated variable (PLR parameters) on one dependent variable (age, gender). p < 0.05 was considered statistically significant.
We intended to verify our pupillography set-up by comparing the pupil diameter measurements obtained with a commercially available pupillometer. The horizontal visible iris diameter (measured by the slit lamp, Figure Figure1C)1C) of each participant was used as a scale bar for image analysis of the IR pupillography video. Repeated measurements of the resting pupil diameter using the NeurOpticsTM pupillometer was comparable with the values obtained with our IR video pupillography system in three subjects (Figure (Figure2).2). The results showed no significant differences between measurements using the NeurOpticsTM pupillometer vs. our IR video pupillography (One way ANOVA, p > 0.05).
The smart device light emitting diode (LED) light source was placed 5 cm away from the animal’s eye. The intensity of the flash was measured to be 2300 watt per steradian at 5 cm by the ILT1700 radiometer.
The guinea pig video recording was analyzed by the custom MATLAB program, which detected the greatest positive and negative change in illumination. The positive peaks would correlate to the frame numbers when the strobe light turned ON, and the negative troughs correlate to the frame numbers when the strobe light turned OFF (Figure (Figure3).3). Our results showed the ON duration was 210.00 ± 22.50 ms, and the OFF duration was 5018.52 ± 17.57 ms.
The pupillometry set-up and data output from the MATLAB program yielded the following graph, which shows how pupil diameter changes over the light ON and OFF cycles during video recording (Figure (Figure4).4). Further data output may be obtained from the graph to calculate the resting pupil diameter, constriction and DV, % constriction and re-dilation ratio as described in stepwise procedures. During protocol development these parameter were assessed at two time points in male and female guinea pigs’ development (at 3 months and 7 months old).
Male juvenile guinea pigs had larger resting pupil size than female guinea pigs (M = 7.42 ± 0.20 mm vs. F = 5.43 ± 0.32 mm; One way ANOVA, p < 0.05; Table Table1).1). In terms of PLR kinetics, all parameters except the average constriction ratio (AvCR) were significantly different between male and female juvenile guinea pigs. Juvenile males had quicker PLR recovery kinetics, greater CV than females (M = 2.02 ± 0.18 mm/s vs. F = 1.13 ± 0.10 mm/s, p < 0.01) and greater DV (M = 0.41 ± 0.05 mm vs. F = 0.22 ± 0.02 mm, p < 0.001). Juvenile males had both quicker pupil CV and DV suggestive of a more readily mobile state of both the sphincter and dilator pupillae on stimulation of a PLR (Table (Table1).1). In adult female and male guinea pigs, the AvCR (p < 0.05) and average re-dilation ratio (AvDR; p < 0.001) were significantly greater in females.
There were no significant differences in pupil size between juvenile and adult males; however, juvenile female guinea pigs had significantly smaller pupil size than adults (One way ANOVA, p < 0.05; Figure Figure5A).5A). In terms of PLR kinetics, CV significantly decreased in adult compared with juveniles in both females and males guinea pigs (Figure (Figure5B).5B). AvCR was larger in the adult female group compared with juvenile (p < 0.05; Figure Figure5C),5C), which means adult female pupils were constricting to a lesser extent than juvenile females. This may be suggestive of an age-related decline in parasympathetic input as the mobility of sphincter pupillae was acting more slowly, and constricted to a lesser extent in the adult animals, and that the sympathetic input to the dilator pupillae muscle was relatively more unopposed. Such age-related effect in pupil constriction parameters were not observed in male guinea pigs. DV significantly decreased in males (p < 0.001), but was not significant in females (p < 0.1; Figure Figure5D5D).
The experimental set-up and analysis described in our study is more informative than the ones described in other animal studies (Pennesi et al., 1998; Dabisch et al., 2008; Taylor et al., 2008; Mohan et al., 2013; Ostrin et al., 2014) as we have measured dynamic pupil change over time rather than pupil diameter only. The video pupillography was recorded at a frame rate (30 fps) similar to other animal studies that had sampling rate of every 33 ms, and was sufficient for calculating CV, and DV. Velocity offered more information about the state of the neurons along the visual pathway, as it is a direct indicator of the integrity and strength of the neural signals (Dabisch et al., 2008; Mohan et al., 2013). Our results in female guinea pigs were consistent with that reported by Howlett and McFadden (2007) which described an increase in pupil diameter from 2 days to 200 days of age. Gender-specific pupil data, and parameters such as CV and DV were unavailable in guinea pigs—further investigation is required to confirm the gender and age-specific findings in our anticipated and representative results.
PLR is part of a routine examination for the integrity of the nervous system; it evaluates the state of the autonomic response to light, allowing us to see how efficiently the sphincter pupillae reacts, and how well the dilator pupillae counteracts the sphincter pupillae post illumination. Current pupil studies in animals utilize good quality commercial IR video cameras that are capable of recording at up to 60 fps (Taylor et al., 2008; Mohan et al., 2013). The temporal frequencies of the video recording protocol across animal studies are however highly variable, with Taylor et al.’s (2008) study recording at much lower temporal resolution—1 frame every 2 s. The lack of dynamic pupillary data (continuous measure of rate and magnitude of change) in animal studies is a limitation in existing literature, which we investigated in our study. The inclusion of pupil kinetic in our system—the rate of pupil change during the process of constriction and re-dilation is an innovation in pupil reaction studies in animals. Pupil studies in humans often take more kinetic parameters into account, such as PLR latency, CV, acceleration etc. Fotiou et al. (2009) concluded in their study that maximum pupil CV and acceleration were the best predicators in separating subjects with cholinergic deficiency from normal subjects, while baseline pupil diameter and minimum pupil diameter after pupil constriction to light were not significantly different from controls. This implies that the rate of pupil diameter change is a more relevant measure than pupil diameter when evaluating PLR and iris innervation. Fotiou et al.’s (2009) IR video camera recorded at up to 263 fps, and had high quality light sources, one IR source to illuminate the participant’s face (made of a 32 LED array), and one light stimulus stimulating the pupil with diffuse white light flashes (Taylor et al., 2008; Mohan et al., 2013). This set-up is superior to those described in previous animal studies due to its high sampling rate, good quality videos with high contrast due to the bright IR LED illumination, and a bright light stimulus. However, future studies with video pupillography taken at frame rates comparable with human studies will give us insight in how high the frame rate might need to be to achieve optimal pupil mobility measures in animals.
There are limitations to our techniques, and the light source in particular can be improved for future development. One potential drawback of our light stimulus for initiating PLR is that the white LED light of the smart device was a point source, so accurate positioning of the light was necessary to ensure the same amount of light reached the eye across all animals. A LED diffuser lens fitted in front of the smart device LED will ensure that the stimulus can illuminate the pupil more evenly, even if the position of the stimulus alters during experiment. A brighter IR illumination system would allow better illumination of the guinea pig pupil for a higher-contrast video, which would enable our custom MATLAB software to better discriminate pupil margin for data analysis. The IR illumination could also be integrated with the camera—i.e., an array of approximately 20 IR LEDs can be soldered onto an electrical board and fitted around the camera lens by a collar.
The IR video pupillography technique described in this study is effective, accessible and easy to assemble. It allows measurement of pupil diameter as well as the dynamic rate and magnitude of pupil change over time. We analyzed the PLR response in normal guinea pigs using parameters such as CV and DV, that are more comparable with human data (Fotiou et al., 2009), and may assist in transitional research looking at both animal models and human participants. Our IR video pupillography set-up can be applied to clinical research in human, as well as in animal models of Parkinson disease and Alzheimer’s disease, that are known to have cholinergic deficits. PLR is becoming an increasingly popular tool in neurological and eye research, contributing to the examination of the ANS, and the retina and optic nerve of the eye. The experimental set-up described in this study may provide a foundation for further development of a more integrated system, which can be used in research as well as in ophthalmological assessments in the clinical setting.
LY-LC, MLA and JT designed the study. JT and TYQ designed the MATLAB protocols. JMB proofread and discussed the article. LY-LC, MLA and TYQ conducted the experiments and analyzed the data. LY-LC and MLA prepared the manuscript.
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
This work was supported by The Paul Dunlop Memorial Research Scholarship through the New Zealand Association of Optometrists; the Dick Roberts Community Trust through the Neurological Foundation of New Zealand (1229 SPG). We appreciate Bruce Tetley, Adrian Sallis and Richard Warren’s help with preparation of equipment and selected figures. We are also grateful for the unnamed participants in this study, and Mr. Shashi Patel for consenting to the use of his picture in the demonstration of the procedure.