Novel mathematical algorithm for pupillometric data analysis

https://doi.org/10.1016/j.cmpb.2013.08.008Get rights and content

Abstract

Pupillometry is used clinically to evaluate retinal and optic nerve function by measuring pupillary response to light stimuli. We have developed a mathematical algorithm to automate and expedite the analysis of non-filtered, non-calculated pupillometric data obtained from mouse pupillary light reflex recordings, obtained from dynamic pupillary diameter recordings following exposure of varying light intensities. The non-filtered, non-calculated pupillometric data is filtered through a low pass finite impulse response (FIR) filter. Thresholding is used to remove data caused by eye blinking, loss of pupil tracking, and/or head movement. Twelve physiologically relevant parameters were extracted from the collected data: (1) baseline diameter, (2) minimum diameter, (3) response amplitude, (4) re-dilation amplitude, (5) percent of baseline diameter, (6) response time, (7) re-dilation time, (8) average constriction velocity, (9) average re-dilation velocity, (10) maximum constriction velocity, (11) maximum re-dilation velocity, and (12) onset latency. No significant differences were noted between parameters derived from algorithm calculated values and manually derived results (p  0.05). This mathematical algorithm will expedite endpoint data derivation and eliminate human error in the manual calculation of pupillometric parameters from non-filtered, non-calculated pupillometric values. Subsequently, these values can be used as reference metrics for characterizing the natural history of retinal disease. Furthermore, it will be instrumental in the assessment of functional visual recovery in humans and pre-clinical models of retinal degeneration and optic nerve disease following pharmacological or gene-based therapies.

Introduction

Pupillometry is a non-invasive imaging technique that is used to assess pupillary light reflexes [1]. The pupil is innervated solely by the autonomic nervous system, allowing for assessment of visual function with respect to the central and peripheral autonomic systems [2]. Clinically, pupillometry is used as an objective means of evaluating the integrity of the retinal-cortical vision pathways. Currently, pupillometry is used as an outcome measurement in clinical trials following gene replacement therapy for retinal degeneration, indicating the restoration of pupillary function as a direct indication of improved retinal function [3], [4], [5]. However, only recently has commercially made recording equipment been available for pupillometric evaluation involving laboratory animals, such as mice (Neuroptics Inc., San Clemente, CA). Pupillometry can be an important evaluative parameter for various retinal and neurodegenerative disorders [6], [7]. This method for pupillometric data collection utilizes a dual-camera system and infrared illumination to record pupillary responses to light stimuli. The Neuroptics pupillometer tracks pupillary diameter through the differences in threshold detection, which refers to the differences in hue of the dark pupil and lighter iris tissue. When light stimuli are applied, pupillary constriction results, manifesting as a decrease in the pupillary diameter followed by a period of pupillary recovery prior to a subsequent stimulus. Constriction of the pupil is initiated after light-sensing photoreceptor cells in the retina are stimulated and an electrical impulse is relayed through bipolar, horizontal, amacrine, and ganglion cells, and ultimately conducted through the optic nerve. The signal then travels via the optic tract to synapse in the ipsilateral lateral geniculate nucleus of the thalamus and the ipsilateral pretectal region of the rostral midbrain. The latter set of afferent neurons synapse bilaterally at the Edinger-Westphal nuclei of the midbrain. These neurons contribute to pre-ganglionic parasympathetic efferent fibers in the oculomotor nerve (cranial nerve III) which projects back towards the brain and synapse in the ciliary ganglion. When stimulated, the post-ganglionic short ciliary nerves act to constrict the pupil. The pupillary light reflex involves multiple synaptic connections and is maintained only if all components of the retina and optic nerve are intact [8]. Here we demonstrate how a mathematical algorithm for pupillometric data analysis can be used to expedite, automate and standardize pupillometry data analysis and establish normative pupillometry parameters that are highly accurate for mouse models and humans. Normative values can ultimately be used to assess efficacy of pharmacologic and gene therapy treatments of retinal or neurodegenerative diseases.

Section snippets

Experimental design

Mouse pupillometric measurements were used to develop the algorithm. The Neuroptics Pupillometer (San Clemente, CA, USA) utilizes a dual-camera system and infrared wavelengths to record pupillary responses to light stimuli. The pupillometer tracks pupil diameter through threshold detection. The Neuroptics software sampled at 15 Hz frequency. The data used to develop the program involved a pulse duration of 100 ms and a relaxation time of 9900 ms. Wild-type C57Bl6 mice were dark-adapted for a

Evaluation of algorithm

The calculated parameters from the novel algorithm were compared to manually-derived values from the raw data obtained from the pupillometry apparatus. The comparison of the derived parameters from the two sources was conducted using the paired Students t-test, which yielded non-significant p-values for all parameters (n = 40) (Table 1).

The comparison of the data obtained by the custom algorithm and the manual method demonstrate that the novel algorithm is accurate and expeditious in calculating

Discussion

The p-values (p  0.05) demonstrate that there is no significant difference between the parameter values generated by the novel algorithm as compared to the parameter values calculated manually. Thus the described algorithm can be used to reliably extract a variety of parameters from raw pupillometric data generated by the Neuroptics Pupillometer. The algorithm can be used to establish normative values of pupillary response for numerous mouse strains, both wild type and transgenic models. The

Conclusion

Since pupillometry is beset with background and subject-derived aberrations, this algorithm can reliably remove data altered by eye blinking, loss of pupil tracking, and/or head movement. Based on our findings, the optimal threshold for removal of aberrant data is a value of k between one and two (one and two standard deviations) when artifacts are visually present. However, a threshold system also allows the experimenter to control the sensitivity and/or specificity of the algorithm. Accurate

Acknowledgments

The authors would like to thank Dr. Kenneth Foster, Professor of Bioengineering, School of Engineering and Applied Science, University of Pennsylvania, for his advice concerning filter choice and design. The authors would also like to thank Dr. Gui-Shuang Ying for his advice concerning statistical analysis. This work was supported by the Foundation Fighting Blindness which sponsors the CHOP-Penn Pediatric Center for Retinal Degenerations, Research to Prevent Blindness (JB and departmental

References (14)

There are more references available in the full text version of this article.

Cited by (7)

  • Effects of stimulus size, eccentricity, luminance, and attention on pupillary light response examined by concentric stimulus

    2020, Vision Research
    Citation Excerpt :

    Time 2 (T2) indicates the timing of maximum constriction acceleration. Pupillary latency (PL), which is defined as the onset of pupil constriction in reaction to the onset of stimulus, is calculated by the difference between T1 and T2 (Bergamin & Kardon, 2003; Canver et al., 2014). Time 3 (T3) indicates the timing of maximum constriction velocity, and the velocity at T3 is regarded as maximum constriction velocity (MCV).

  • Impaired pupillary control in “schizophrenia-like” WISKET rats

    2018, Autonomic Neuroscience: Basic and Clinical
  • Estimation of operators' fatigue using optical methods for determination of pupil activity

    2015, International Journal of Occupational Medicine and Environmental Health
View all citing articles on Scopus
1

These authors contributed equal work.

View full text