Development of a Smartphone-based Pupillometer for

Neuro-ophthalmological Diseases Screening

Ana Isabel Sousa

1 a

, Rui Valente Almeida

1 b

, Maria Narciso

1 c

, Fernando Sacilotto Crivellaro

1 d

Carlos Marques Neves

2 e

, Lu

ıs Abeg

ao Pinto

2 f

and Pedro Vieira

1 g

Department of Physics, Faculty of Science and Technology, NOVA University of Lisbon, Caparica Campus,

2829-516 Caparica, Portugal

Faculty of Medicine, University of Lisbon, 1649-028 Lisbon, Portugal

Keywords:

Pupil, Pupillometry, Smartphone, Neuro-ophthalmological Diseases.

Abstract:

Over the last two decades pupillometry gained a renewed interest, due to the discovery of intrinsically pho-

tosensitive retinal ganglion cells (ipRGCs) and their function in pupil light reﬂex (PLR). This technique is

usually used to assess patient’s neurological state and has been researched as a screening tool for neuro-

ophthalmological diseases. Several automated pupillometers have been developed, as they allow a quanti-

tative measure of PLR, but most of them are expensive and not portable, which reduces their possibility to

be a widespread screening tool. Taking advantage of low price and accessible smartphone technology, a

smartphone-based pupillometer was developed in this work. An Android application was developed that al-

lows pupil’s dynamic video recording and its processing for pupil detection. The preliminary tests made to

validate the application and the algorithms have shown that the proposed system is a promising tool for a

simple, inexpensive and portable pupillometry.

1 INTRODUCTION

Pupil light reﬂex (PLR) has been widely used to as-

sess the patient’s consciousness in both qualitative

and quantitative ways. Over the last 20 years, due

to the discovery of intrinsically photosensitive retinal

ganglion cells (ipRGCs) and their function in pupil re-

sponse to light (Hattar et al., 2002; Lucas et al., 2001),

pupillometry gained a new interest. Particularly be-

cause these cells discovery and research showed that

pupil light reﬂex is not only pursued by rods and

cones, but also by ipRGCs, as they are sensitive to the

absorption of blue light (Gamlin et al., 2007). This

renewed interest in pupillometry research also lead

to an increase in its potential to be applied to neuro-

ophthalmological diseases screening and detection,

such as Parkinson (Giza et al., 2011; Wang et al.,

https://orcid.org/0000-0003-2980-4742

https://orcid.org/0000-0002-2269-7094

https://orcid.org/0000-0001-5079-9381

https://orcid.org/0000-0002-7534-9149

https://orcid.org/0000-0002-3842-2466

https://orcid.org/0000-0002-9960-7579

https://orcid.org/0000-0002-3823-1184

2016), Alzheimer (Granholm et al., 2017) or Glau-

coma (Rukmini et al., 2019; Rukmini et al., 2015;

Gracitelli et al., 2014)

Usually known as pupillometry, this technique al-

lows an objective measurement of pupil’s dynamic to

a certain stimulus when automated. Pupillometry is

non-invasive and allows a functional assessment of

the pupil light reﬂex. With ipRGCs discovery, chro-

matic pupillometry also gained an important role as

it allows to study different types of damage to rod,

cones and ipRGCs, measuring pupil responses to red

or blue light stimuli (Rukmini et al., 2019). This

technique using red or blue stimuli has been stud-

ied and applied for the previously mentioned neuro-

ophthalmological diseases screening.

Several types of pupillometers have been devel-

oped over the years based in infrared video acquisi-

tion, ﬁrst established by Loewenstein et al. (Lowen-

stein and Loewenfel, 1958) with the construction of a

photoelectric pupillograph in 1947. Technology im-

provements over the last decades allowed continuous

video recording of the pupil and automatic computer

data analysis, leading to a large upgrade in pupil-

lometry technique. Since Loewenstein et al. (Lowen-

stein and Loewenfel, 1958) work, pupillometry has

Sousa, A., Almeida, R., Narciso, M., Crivellaro, F., Neves, C., Pinto, L. and Vieira, P.

Development of a Smartphone-based Pupillometer for Neuro-ophthalmological Diseases Screening.

DOI: 10.5220/0008962600500056

In Proceedings of the 13th International Joint Conference on Biomedical Engineering Systems and Technologies (BIOSTEC 2020) - Volume 1: BIODEVICES, pages 50-56

ISBN: 978-989-758-398-8; ISSN: 2184-4305

Initial light adaptation post-stimulusbaseline Pupil recovery

Start

Stop

post-stimulusbaseline

Start Stop

time

Figure 1: Chromatic pupillometry protocol schema.

been infrared camera based, which allows a proper

image contrast and the environment light does not

have an impact in pupil size. These type of systems,

with infrared cameras, are highly effective and pro-

vide precise measurements of the pupil. However,

they are also expensive, not portable and usually re-

quire some trained operator, which is a restriction for

its widespread use.

Mobile smartphones industry exponential increase

has brought a new interest for medicine applications,

as they overcome those previously referred limita-

tions, such as price, portability and accessibility. As

smartphones are present in everyones daily routines,

they could lead, in the limit, to a self-diagnosis tool

easily widespread over the population. The usage of

smartphones for pupillometry has started in 2013, by

Kim et al. (Kim and Youn, 2013), with the devel-

opment of a smartphone-based infrared video pupil-

lometer. This device included an optical apparatus at-

tached to the camera, with four infrared light emitting

diodes (LEDs), one white LED to work as the stim-

uli, one infrared cut-off ﬁlter and one microcontroller.

Tight occlusion around the eye was also taken care

to protect the eye from environment light, to reduce

its inﬂuence in PLR. All the acquired data was trans-

fered to a laptop and the analysis was made using a

proposed algorithm processed in MATLAB



(Math-

works Inc., Natick, MA).

Another smartphone app has been developed in

2016 by Shin et al. (Shin et al., 2016) used only to ac-

quire ﬁve steady images of the eye in different stages:

one before the ﬂash, another one during the light stim-

uli and the remaining three photos after the ﬂash. The

acquired images were then analyzed by a clinician

and compared the measurements with a penlight mea-

surement. In this research study, no automated neither

computerized algorithm was used to analyze pupil’s

size variations with the stimuli. The results showed

that pupil size measurements with a smartphone ap-

plication were similar to the ones made by a trained

clinician.

More recently, McAnany et al. (McAnany et al.,

2018) have developed an iPhone-based pupillometer,

which uses the rear camera to capture a high speed

video, its ﬂash as a white stimulus and processes real

time measurements. The major difference from this

research study to the other two mentioned is the com-

pletely software based system, without the need to

any external optic apparatus, and the full process be-

ing taken care by the iPhone. The algorithm used to

analyze the acquired data uses a ratio between pupil

diameter and iris diameter. The measurements with

this system were then compared to ones acquired with

an infrared pupillometer and the results were in agree-

ment.

Although these studies show an increased po-

tential in smartphone-based pupillometry, apart from

McAnany et al. (McAnany et al., 2018) system, they

do not provide a low-cost, real-time, accessible and

portable device for pupillometry. The iPhone project

can be considered low cost when comparing to the in-

dustrial pupillometers, but comparing to other smart-

phones is a high range one. It is also important to

notice that the referred studies using smartphones do

not speciﬁcally target chromatic pupillometry, which

has been studied as a proper neuro-ophthalmological

diseases screening method (Rukmini et al., 2019).

The present study aims to describe an all-in-

one smartphone-based chromatic pupillometer using

a medium range Android device. All-in-one indicates

that the smartphone is used to acquire and process the

pupillometry data, running image processing algo-

rithms through the Android application, without the

need of high computing machines. The main goal of

this study was to show Android capability to perform

chromatic pupillometry measures and to run pupil de-

tection algorithms in real-time.

2 METHODS

Given the state-of-the-art of mobile pupillometers,

it is intended to develop a low cost system using a

medium range Android smartphone, with a camera to

allow pupil recording, a ﬂash to work as a stimulus

and with enough capability to run image processing

algorithms. The system should also allow chromatic

pupillometry, using both blue and red stimuli in order

to be used to screen neuro-ophthalmological diseases

according to the recent ﬁndings and protocols.

Development of a Smartphone-based Pupillometer for Neuro-ophthalmological Diseases Screening

The acquisition protocol should start with a period

for the eye to adapt to the environmental light condi-

tions, then the recording period with some initial time

to acquire pupil’s baseline, then a short colored light

stimulus ﬂash followed by a post-stimulus period. Be-

fore a new acquisition a pause should be made for

pupil recovery and the process should then be re-

peated with change of light stimulus color. A repre-

sentation schema of this type of protocol is shown in

Figure 1. Each of the protocol periods duration should

be tested and optimized in future work, particularly

to be applied to neuro-ophthalmological pathologies

such as Alzheimer, Dementia, Glaucoma or Parkin-

son.

Essentially, the main goal is to develop a system

with all these characteristics, that could be, in the

future, used in any Android device as a tool spread

through the population for neuro-ophthalmological

diseases early screening.

2.1 System Architecture

The system proposed in this work consists only in a

smartphone that allows acquiring and processing the

pupillometric data. Development was made using a

Nokia 7 Plus (Nokia Corporation, HMD Global, Fin-

land) which is an Android One device, with Android

9 Pie operating system (Android sdk 28). The appli-

cation was developed in Java programming language

using Android Studio (IntelliJ



Platform). Video and

image processing was made using OpenCV library

(Open Source Computer Vision Library) with Java

Native Interface (JNI) framework, which allows Java

to run C or C++ code, being then incorporated in the

Java Android application.

Acquisition

- Camera 2 API

- Rear facing ﬂash control

Image Processing

- Video and Image Processing

- ElSe Algorithm

JAVA

C++

JNI

Figure 2: Proposed system architecture.

For the acquisition part of the application An-

droid’s Camera2 API was used, which provides an in-

terface to individual camera devices available in the

smartphone and proper adjustments of the recording

and image characteristics. In this case, Nokia 7 Plus

has one front camera and two rear cameras. The

rear-cameras are considered as one logical camera by

Camera2 API, which means that it is not possible to

access each of the physical rear cameras, they work as

one, so when changing recording characteristics the

result comes from both physical cameras combined

in one image.

Camera2 API also allows to control rear-facing

ﬂash light, which was used as light stimuli for pupil-

lometry measures. The spectral emission of the

rear-facing ﬂash of the Nokia smartphone was mea-

sured using a spectrometer (AvaSpec - Mini2048CL

- UVI25 by Avantes, Netherlands) and the average

resultant of three acquisitions made is shown in Fig-

ure 3.

Figure 3: Spectral emission characteristics of the rear-

facing camera ﬂash of the Nokia 7 Plus.

To allow chromatic pupillometry, a simple ﬁlter

made with standard grade cellophane paper, with blue

or red colors, was placed in front of the rear-facing

ﬂash. In this way the color of the ﬂash gets ﬁltered to

get blue or red ﬂash lights, whose spectra, which was

also acquired with Avantes spectrometer, are shown

in Figure 4. This low cost and easy solution to get

colored stimuli allows to get the proper wavelength

light stimuli (red and blue), according to literature.

2.2 Video and Image Processing

After video acquisition, the second main part of the

proposed smartphone application is the video and im-

age processing algorithm. Using OpenCV for An-

droid and C++ language, as previously referred, the

acquired video is then processed, converted to frames

and pupil is detected in each frame through a proper

algorithm.

BIODEVICES 2020 - 13th International Conference on Biomedical Electronics and Devices

Figure 4: Spectral emission characteristics of the rear-

facing camera ﬂash of the Nokia 7 Plus with red and blue

ﬁlters.

The acquired video does not contain only the eye,

but also some part of subject’s face, due to the dis-

tance to the smartphone to allow proper image focus.

To overcome this and to reduce non relevant informa-

tion in the image for the application of pupil detec-

tion algorithm, an eye detection algorithm is applied.

OpenCV offers a pre-trained Haar cascade algorithm

for face and eye detection, based in Viola and Jones

Haar cascade object detection algorithm (Viola and

Jones, 2001).

In this work, OpenCV Haar cascade eye detection

was applied to each frame,and was then cropped in the

obtained location. This algorithm sometimes fails and

considers some other part of the image to be an eye;

these detections with smaller size were automatically

discarded.

After eye detection, a process was made as sum-

marized in Figure 5 in order to detect the pupil.

Contrast Limited Adaptive Histogram and ElSe Al-

gorithm are going to be further explained in 2.2.1

and 2.2.2.

2.2.1 Contrast Limited Adaptive Histogram

One of the problems of images acquired with non in-

frared cameras or in non ideal lightning conditions is

the low contrast ratio between iris and pupil, particu-

larly in dark colored iris. To overcome this situation,

a contrast enhancement of the image increases pupil’s

visibility.

Histogram equalization distributes the intensities

on the histogram, leading to an increase in the global

contrast of an image. It is highly efﬁcient and simple,

however it can produce ”washed out” effect or can

destroy the brightness of the image.

Adaptive Histogram Equalization allows locally

enhancement of the contrast, as it perform histogram

equalization in different sections of the image redis-

tributing the lightness values of the image. One of the

VideoAcquisition

GetVideoFrames

ConvertRGBtoGrayScale

ApplyCLAHE

ApplyElSeAlgorithm

Pupildetected

For each frame

EyeDetectionand

ImageCrop

Figure 5: Flowchart of the image processing algorithms.

main problems of this technique is the high compu-

tational complexity, not being favorable for real-time

applications.

An extended case of adaptive histogram equaliza-

tion is Contrast Limited Adaptive Histogram Equal-

ization (CLAHE), which performs an histogram clip-

ping at some threshold and redistributes the image us-

ing the maximum values. It has a lower computational

complexity and prevents over-ampliﬁcation of noise

signals.

According to Hassan et al. (Hassan et al., 2017),

CLAHE algorithm outperforms a simple histogram

equalization or an adaptive histogram equalization for

iris recognition. Taking these results into considera-

tion, in this work CLAHE was tested and applied to

video frames before running the pupil detection algo-

rithm, using the OpenCV CLAHE function with cut

limit = 4.

Development of a Smartphone-based Pupillometer for Neuro-ophthalmological Diseases Screening

2.2.2 Pupil Detection Algorithm

After image acquisition and eye detection, the con-

cern rests in pupil detection algorithms as one of the

main parts in automated pupillometry systems. It is

important to clarify that pupil detection refers to ﬁnd-

ing its center and size (area or diameter) in the image,

either in pixels or converted to some unit of measure.

There are some difﬁculties regarding achieving this,

that can go from low contrast images, blur, illumina-

tion issues and many others.

With the increasing interest in automated pupil-

lometry has also increased the need to have better and

more precise pupil detection algorithms. Particularly,

as the applications of pupillometry are being studied

to be in real-world environments and not only under

laboratory and controlled conditions.

One of those is ElSe algorithm, developed by Fuhl

et al. (Fuhl et al., 2015), based on ellipse evaluation

of a ﬁltered edge image thought to be applied in real-

world scenarios, such as in-door environments or dur-

ing driving. In a state-of-the-art review published in

2016 (Fuhl et al., 2016) that compared several pupil

detection algorithms behavior has considered ElSe al-

gorithm as a gold standard for pupil detection. This is

an open source algorithm which is the one chosen to

be used in the system proposed in this work, due to

its easy access and it is targeted for real world envi-

ronments, which is what having a smartphone-based

pupillometer pursues.

The input is a gray scale image and in a very sum-

marized way the algorithm tries to ﬁnd an ellipse that

could most likely be the pupil. First applies a Canny

ﬁlter to have the image edges, then they are ﬁltered

using straightening patterns, the straight lines are re-

moved and the best ellipse is selected through least

square ellipse ﬁtting. The ﬁnal step is the ellipse eval-

uation, excluding those that are unlikely to be pupils.

If this ﬁrst process fails there is a second approach that

the algorithm tries through coarse positioning. This

second analysis is made by downscaling and convolv-

ing the image with two different ﬁlters: a surface dif-

ference ﬁlter and a mean ﬁlter. The results of both

convolutions are multiplied and the maximum value

is the starting point to be reﬁned. The surrounding

pixels of this point is veriﬁed, and the new pupil po-

sition is the center of mass of the pixels under this

threshold.

Even though ElSe algorithm was developed to

real-world scenarios it is important to notice that it

was tested and validated in datasets acquired with in-

frared cameras, which in most of the cases have a very

evident pupil. In the present work, this algorithm was

applied to images acquired using Nokia 7 Plus smart-

phone camera, without any optical apparatus, as a pre-

liminary test and validation of the proposed solution.

Some of Fuhl et al. (Fuhl et al., 2015) datasets pos-

sess eye images where the pupil is hidden, sideways

or even more far-fetched scenarios. In this prelimi-

nary study, ElSe algorithm was only applied to im-

ages where the pupil is normal, with subject looking

straight forward and the pupil is not occluded by eye-

lashes for example.

Figure 6: Original video frame in gray scale with and with-

out CLAHE with respective histograms.

3 PRELIMINARY RESULTS

First, pupil detection algorithm running in the An-

droid application was validate with eye images

datasets published by Fuhl et al. (Fuhl et al., 2015).

As expected, pupil center and size were the same as

Fuhl research group has labeled. This simple test

was just to guarantee that ElSe algorithm was prop-

erly running in the Android application, considering

the system architecture and the linkage between pro-

gramming languages.

Preliminary tests were made using the developed

smartphone application to acquire videos of the eye

and apply the processing algorithms to get the pupil.

ElSe algorithm was tested in frames with and with-

out CLAHE. An example of the same frame with and

without CLAHE is shown in Figure 6 with the re-

spective histograms.

The image processing algorithms proposed in this

work for each frame are exempliﬁed in Figures 7

and 8 with a frame from a video acquired with the

developed smartphone application.

BIODEVICES 2020 - 13th International Conference on Biomedical Electronics and Devices

Table 1: Pupil parameters mean values obtained for 41 images with and without CLAHE. Unit of measure: pixels.

Pupil parameter Original Image Image with CLAHE

Center (124 ± 6, 106 ± 5) (124 ± 6, 106 ± 5)

Height 20 ± 8 21 ± 7

Width 19 ± 7 19 ± 6

Angle 108 ± 32 105 ± 36

The pupil detection algorithm with and without

CLAHE was also applied to 41 frames from the same

acquisition made with the same person. The average

results for these 41 images are summarized in Table 1

for both CLAHE and non CLAHE algorithms. From

this test it is possible to verify that the results are sim-

ilar in terms of center, height and width, being the

angle the most different value, in average, and with

higher standard deviation.

Figure 7: Image Process Schema. a) Frame from the

video acquired using the smartphone-bases pupillometer

with Nokia 7 Plus; b) Cropped image around the eye ob-

tained through eye detection algorithm; c) Eye image with

pupil detected.

Figure 8: Image Process Schema with CLAHE. a) Frame

from the video acquired using the smartphone-bases pupil-

lometer with Nokia 7 Plus; b) Cropped image around the

eye obtained through eye detection algorithm; c) Eye image

with CLAHE and pupil detected.

With the images acquired in these preliminary

tests this algorithm seems promising to deal with im-

ages acquired with this smartphone camera without

any optical apparatus and further tests should be made

to better validate the proposed algorithm.

4 CONCLUSION

Using low-cost, portable and accessible technology

for medical applications, particularly for screening

and monitoring diseases, is gaining interest and mar-

ket all over the world. The usage of a smartphone for

this purpose is a smart and easy way to spread early

screening and make it more available and accessible

to everyone and all over the world. This work pro-

poses a solution for pupillometry measurements with

a smartphone, which overcomes the main problems

with the existing pupillometers to spread this tech-

nique into medical and screening usage. The pre-

liminary tests made with the proposed prototype in-

dicate a great potential of this solution, particularly

due to its low price, easy accessibility and portability.

Another advantage of the proposed system is that it

only needs the smartphone, diminishing the requisites

of high range technology and apparatus. To perform

chromatic pupillometry it needs a standard grade cel-

lophane paper, which, in the future in a commercial

solution, can be available as a kit to complement the

smartphone application.

Further work is to improve the algorithms for

pupil detection and make more validation tests. It

is also relevant to test this solution in different light

conditions and make some adjustments in the camera

recording characteristics in order to get a more efﬁ-

cient and precise pupillometer application.

After these improvements and algorithm val-

idations, the next step should be validating

this smartphone-based pupillometer for neuro-

ophthalmological diseases screening, using the

colored ﬁlters to apply colored stimuli allowing to do

chromatic pupillometry. With these colored stimuli,

should be tested different acquisition protocols

that could early screen pathologies as Alzheimer,

Dementia, Glaucoma or Parkinson in a portable and

accessible way.

In general, a smartphone-based pupillometer

seems to be the future of pupillometry to lower the

gap between academic research and clinical applica-

tion. The preliminary tests made with the proposed

system show its potential to be used as a pupillome-

ter and, in the future, to screen and monitor neuro-

ophthalmological diseases.

ACKNOWLEDGEMENTS

This work is funded by National Funds through FCT

- Portuguese Foundation for Science and Technology

Development of a Smartphone-based Pupillometer for Neuro-ophthalmological Diseases Screening

and Compta S.A. under the PhD grant with reference

PD/BDE/135002/2017. A special acknowledgment to

Compta S.A. team for all the support given.

REFERENCES

Fuhl, W., Santini, T. C., Kuebler, T., and Kasneci, E. (2015).

ElSe: Ellipse Selection for Robust Pupil Detection in

Real-World Environments.

Fuhl, W., Tonsen, M., Bulling, A., and Kasneci, E. (2016).

Pupil detection for head-mounted eye tracking in the

wild: an evaluation of the state of the art. Machine

Vision and Applications, 27(8):1275–1288.

Gamlin, P. D., McDougal, D. H., Pokorny, J., Smith, V. C.,

Yau, K. W., and Dacey, D. M. (2007). Human

and macaque pupil responses driven by melanopsin-

containing retinal ganglion cells. Vision Research,

47(7):946–954.

Giza, E., Fotiou, D., Bostantjopoulou, S., Katsarou, Z., and

Karlovasitou, A. (2011). Pupil light reﬂex in Parkin-

son’s disease: Evaluation with pupillometry. Interna-

tional Journal of Neuroscience, 121(1):37–43.

Gracitelli, C. P. B., Duque-Chica, G. L., Moura, A. L.,

Nagy, B. V., de Melo, G. R., Roizenblatt, M., Borba,

P. D., Teixeira, S. H., Ventura, D. F., and Paranhos, A.

(2014). A Positive Association Between Intrinsically

Photosensitive Retinal Ganglion Cells and Retinal

Nerve Fiber Layer Thinning in Glaucoma. Investiga-

tive Ophthalmology & Visual Science, 55(12):7997–

8005.

Granholm, E. L., Panizzon, M. S., Elman, J. A., Jak, A. J.,

Hauger, R. L., Bondi, M. W., Lyons, M. J., Franz,

C. E., and Kremen, W. S. (2017). Pupillary Responses

as a Biomarker of Early Risk for Alzheimer’s Disease.

Journal of Alzheimer’s disease : JAD, 56(4):1419–

1428.

Hassan, R., Kasim, S., Jafery, W. A. Z. W. C., and Shah,

Z. A. (2017). Image enhancement technique at differ-

ent distance for Iris recognition. International Journal

on Advanced Science, Engineering and Information

Technology, 7(4-2 Special Issue):1510–1515.

Hattar, S., Liao, H. W., Takao, M., Berson, D. M., and Yau,

K. W. (2002). Melanopsin-containing retinal ganglion

cells: Architecture, projections, and intrinsic photo-

sensitivity. Science, 295(5557):1065–1070.

Kim, T. H. and Youn, J. I. (2013). Development of a

smartphone-based pupillometer. Journal of the Op-

tical Society of Korea, 17(3):249–254.

Lowenstein, O. and Loewenfel, I. E. (1958). Electronic

Pupillography; A New Instrument and Some Clini-

cal Applications. A.M.A. Archives of Ophthalmology,

59(3):352.

Lucas, R. J., Douglas, R. H., and Foster, R. G. (2001).

Characterization of an ocular photopigment capable

of driving pupillary constriction in mice. Nature Neu-

roscience, 4(6):621–626.

McAnany, J. J., Smith, B. M., Garland, A., and Kagen, S. L.

(2018). IPhone-based pupillometry: A novel approach

for assessing the pupillary light reﬂex. Optometry and

Vision Science, 95(10):953–958.

Rukmini, A. V., Milea, D., Baskaran, M., How, A. C., Per-

era, S. A., Aung, T., and Gooley, J. J. (2015). Pupillary

Responses to High-Irradiance Blue Light Correlate

with Glaucoma Severity. Ophthalmology, 122(9[1]

A. V. Rukmini et al., “Pupillary Responses to High-

Irradiance Blue Light Correlate with Glaucoma Sever-

ity,” Ophthalmology, vol. 122, no. 9, pp. 1777–1785,

2015.):1777–1785.

Rukmini, A. V., Milea, D., and Gooley, J. J. (2019). Chro-

matic pupillometry methods for assessing photorecep-

tor health in retinal and optic nerve diseases. Frontiers

in Neurology, 10(FEB):1–20.

Shin, Y. D., Bae, J. H., Kwon, E. J., Kim, H. T., Lee, T.-S.,

and Choi, Y. J. (2016). Assessment of pupillary light

reﬂex using a smartphone application. Experimental

and Therapeutic Medicine, 12(2):720–724.

Viola, P. and Jones, M. (2001). Rapid object detection us-

ing a boosted cascade of simple features. In Proceed-

ings of the 2001 IEEE Computer Society Conference

on Computer Vision and Pattern Recognition. CVPR

2001, volume 1, pages I–511–I–518. IEEE Comput.

Soc.

Wang, C. A., McInnis, H., Brien, D. C., Pari, G., and

Munoz, D. P. (2016). Disruption of pupil size mod-

ulation correlates with voluntary motor preparation

deﬁcits in Parkinson’s disease. Neuropsychologia,

80:176–184.

BIODEVICES 2020 - 13th International Conference on Biomedical Electronics and Devices