Design of a Mobile Application for Eye Signs Screening

A. R. Silva

, P. Gomes

, D. Pereira

, S. Guimar˜aes

and A. Castro

Instituto Superior de Engenharia do Porto, Porto, Portugal

Instituto de Telecomunicac¸˜oes, Universidade do Porto, Porto, Portugal

Faculdade de Medicina da Universidade do Porto and Instituto de Telecomunicac¸˜oes, Porto, Portugal

Hospital de Braga, Braga, Portugal

Faculdade de Engenharia da Universidade do Porto, Porto, Portugal

Keywords:

Eye Pathologies, Mobile Application, Protocol, Red Reﬂex.

Abstract:

Advances in technology make mobile phones very attractive to everyone, specially smartphones, with a large

number of applications. In this paper we describe the necessary features of a mobile application for eye

signs screening, describing some of the ocular pathologies that can be detected with a photograph using a

smartphone, and the data collection protocol necessary to obtain this data and later process it. A ﬁrst version

of the developed application and some results are also presented. This application is simple to use, since the

main target are the parents, who can use it at home as a tool to trace the visual health of their children, given

that an ophthalmologist follow-up is scarce or nonexistent in many places.

1 INTRODUCTION

Currently, the use of mobile applications has expe-

rienced a substantial growth, becoming part of day-

to-day life of many people. With the increasing use

of smartphones several applications in different ar-

eas are being developed, including the medical ﬁeld

(J. Chhablani, 2012), (Bethke, 2010). This can make

the difference achieving an early diagnosis and, con-

sequently, a higher probability that the disease does

not progress to an advanced stage, also in remote

places, or with lack of experts (Gallagher, 2013), (To-

day, 2011). Ophthalmic and optometric organizations

recommend a visit to an eye care professional every 1-

2 years to have a comprehensiveeye exam (Vela et al.,

2012). In this paper we propose to develop a mobile

application for certain ocular pathologies detection

with adequate image collection protocol. The ”Red

Reﬂex Test” (RRT), also named ”Br¨uckner Test’, was

considered as an effective form of blindness preven-

tion and early diagnosis of serious diseases such as

retinoblastoma, Coats’ disease, anisometropia or stra-

bismus (P.J. Magalhes, 2009), (AAOP et al., 2008),

(Amitava et al., 2010).

The motivation for this work arose from several news

around the world about parents who post photos of

their children on social networks, with some friends

noticing that something may be wrong with their vi-

sion (Mail, 2007), (Tvi24, 2014), (Stampler, 2014),

(JornalNot´ıcias, 2014), (Anne, 2013). A correct diag-

nosis was made when parents went to an ophthalmol-

ogist, who actually detected an eye problem/disease,

what led us to the development of an application,

which will be presented next.

2 EYE PATHOLOGIES AND

OPHTHAMOLOGIC TESTS

In order to design a tool for eye signs screening in a

smartphone, it is important to understand the differ-

ent clinical tests and eye pathologies, since we need

to clarify the technological needs, and deﬁne an ad-

equate data collection protocol. After reviewing this

information, a clinical test feasible for the application

in a smartphone was designed. The test and corre-

sponding most common pathologies will be presented

in the following sections. This application is meant to

be used in detecting common eye pathologies of chil-

dren, and therefore the data collection protocol must

be simple to be practical in this context.

519

R. Silva A., Gomes P., Pereira D., Guimarães S. and Castro A..

Design of a Mobile Application for Eye Signs Screening.

DOI: 10.5220/0005256405190524

In Proceedings of the International Conference on Health Informatics (HEALTHINF-2015), pages 519-524

ISBN: 978-989-758-068-0

 2015 SCITEPRESS (Science and Technology Publications, Lda.)

2.1 Eye Pathologies

A brief overview of the pathologies detectable using

the camera of a smartphone will be given in this sec-

tion. The features that may be useful in the detection

of these pathologies will also be described.

2.1.1 Retinoblastoma

Retinoblastoma is a tumor that originates from im-

mature retinoblasts in the neural retina and it is the

most frequent intraocular tumor in children, being re-

sponsible for 11% of cancer cases in the ﬁrst year of

life. The most common signs of retinoblastoma are:

leukocoria, strabismus and inﬂammatory signs (Ra-

masubramanian and Shields, 2012). Leukocoria is

an abnormal (white) reﬂex when a direct light source

reaches the eye. Because retina is highly vascular-

ized and has red-orange color the light reﬂected to-

ward the eye must have a reddish appearance, but in

this disease there is a change in some internal struc-

ture of the eyeball so this reﬂex is absent, and there

is usually a whitish reﬂection. An algorithm may be

envisioned to detect the white spot in the eye of the

children, by taking a series of photos. The specialists

recommend parents to take photos once a month, in

a room with soft light and without red-eye reduction,

to check for a normal red-eye reﬂex in children under

six years of age (Fund, 2010). Coats’ disease, as well

as leukocoria, may be detected in photographs taken

with ﬂash, but in this case the eye presents a yellowish

color (J. Liebmann-Smith, 2010).

2.1.2 Amblyopia

Amblyopia is another visual impairment that may be

detected with the RRT or Br¨uckner test, which is very

helpful to detect amblyogenic disorders (Gr¨af, 2010).

This disease is normally called ”lazy eye” because

the major symptom is a decrease in visual acuity in

a normal eye (Miller et al., 1995). Causes of ambly-

opia, which is a major cause of vision loss in chil-

dren and even adults, include strabismus, refractive

errors and ocular misalignment (Miller et al., 1995),

(Alhammouri et al., 2011), that is why it is very im-

portant to detect amblyogenic factors as early as pos-

sible. Like strabismus, amblyopia may be detected

in a photo with ﬂash light, if there is a change in red

reﬂex symmetry.

2.1.3 Strabismus

Strabismus is an ocular misalignment (TheRoyalCol-

legeofOphthalmologists, 2012), where one eye is ei-

ther constantly or intermittently not directed towards

the same point as the other eye when the patient at-

tempts to ﬁxate an object. It usually develops during

childhood but can occur at any age. Most cases be-

gin before 6 years of age, being the peak age of onset

around 3 years (Rutstein et al., 2011). The exams that

are made by an ophthalmologist usually include the

”Hirschberg test”, the ”Cross-over”, and the ”Cover-

uncover” test, which reveal latent strabismus (Angart,

2014). The Hirschberg test may be used to determine

the type and degree of strabismus using a small pen-

light and directing it toward the eyes. When the pa-

tient is looking at the light, a reﬂection can be seen

on the front surface of the pupil. The reﬂected point

of light will reveal the type and degree of strabismus

(Medical, 2011). The cover-uncover test may reveal

latent strabismus, ”heterophoria”. A cover (occluder)

is placed brieﬂy before the eye that appears to ﬁxate

while the patient looks at a small object (E.C. Cam-

pos, 2002). Finally, the Cross-over test may reveal

latent as well as manifest strabismus. As the name

suggests in this test, an occluder is transferred from

one eye to the other without interval (E.C. Campos,

2002). This brings the idea to develop features in the

application that can be helpful in the diagnosis of stra-

bismus. Such as the other disorders referred before,

the application will take a photograph with ﬂash, and

then, with an algorithm automatically process the im-

age to detect and evaluate the local of the reﬂex in the

pupil, like in the Hirschberg test. After the tests are

done, an algorithm may read and calculate the devia-

tion of the eye relatively to the center of the image.

2.2 Red Reﬂex Test

The RRT is a screening exam for detecting ocular

abnormalities using an ophthalmoscope, that emits

light with the appropriate intensity over the pupil of

the newborn, even without the use of previous drops

(P.J. Magalhes, 2009). This test is performed to eva-

lute the eye condition, helping in the diagnosis of the

described pathologies, because the reﬂection of the

incident light produces a reddish and continuous color

on healthy eyes, meaning that the main internal struc-

tures of the eye (cornea, iris, pupil, lens and retina)

are transparent, allowing the light normally to reach

the retina. In the presence of an abnormality that

obstructs the arrival of light to the retina and its re-

ﬂection, the light reﬂex undergoes changes that inter-

fere with their coloration, homogeneity and binocu-

lar symmetry (Aguiar et al., 2007). The test should

be done early in a child’s development, given that

it is when the vision progresses, and if the diagno-

sis is reached later, the likelihood of full recovery is

smaller. This test is very important, because how it

HEALTHINF2015-InternationalConferenceonHealthInformatics

520

was seen, allows the detection of a huge part of dis-

eases with high incidence in childhood. Thus, this test

was chosen to be reproduced in our application.

3 PROTOCOL DESIGNED FOR

DATA COLLECTION

After reviewing the most common pathologies and the

clinical tests, determining which clinical signs may

be detected using technology available in a smart-

phone, a protocol for data collection needs to be de-

ﬁned within the application. The protocol is a guide-

line which encompass all the steps that are made by

the user, when handling with the mobile application

for data collection. For the development of this proto-

col we had to take into account the light environment,

which must be low in order not to allow the pupil to

close; the positioning of the person from the subject

to not obfuscate their vision; and all the necessary in-

formation shown in the screen, so the person fulﬁll

the minimum requirements for the photographs to be-

come perceptibles. The material necessary to do it is a

mobile phone with the application installed. The data

collection protocol was designed as follows:

1. Create a soft light environment: turning the lights

down and ensuring that any light sources - such as

table lamps or television - are behind the person

(so they dont reﬂect in the eyes).

2. Position yourself about 3 meters from the

child/person and use the zoom to capture the sub-

ject’s entire head (Figure 1).

Figure 1: Sketch ilustrating the positioning of the person.

3. Press the button when ready to initiate the collec-

tion.

(a) The head, eyes and pupil will be detected on the

screen.

4. When all is properly aligned the ﬁrst of 5 photos

will be taken automatically.

5. Follow the arrows to tilt your phone to where they

indicate: 1st upwards (Figure 2).

Figure 2: Drawing exemplifying the indications on the

phone.

(a) Repeat step 5 to down, left and right.

6. A menu will show up asking to send the data col-

lected now or later.

(a) Now: link your mobile phone to the internet (to

your computer by wiﬁ).

(b) Later: When you want, enter in ”Records” - se-

lect the folder with the collection you want to

send to analysis in a central server.

This protocol was included in the mobile applica-

tion developed by us, which will be described in more

detail in the section below.

4 DESIGN OF THE MOBILE

APLLICATION

The information previously referred was used to de-

sign an application for eye signs screening. It was

established that the user of the mobile application is

going to be a non medical person who wants to col-

lect data and receive a report on the eye health status,

preferably from a child. When opening the applica-

tion one disclaimer will appear, and the user may only

can continue after agreeing with the conditions: the

application does not replace a regularly ophthalmolo-

gist appointment. The reports act only as an alarm in

case of some abnormality detection, for further eval-

uation from a clinician, nonetheless it does not sub-

stitute a regular visit to the ophtalmologist. Then the

user will have three options: one for ”Collect”, other

for ”Alert”, and one for ”Records”. But, as in Fig-

ure 3, all will be disabled because the user has to reg-

ister a person in the application to start using it.

In a dropbox he can click on ”New”, and then will

be forwarded to a new window. In this window (Fig-

ure 4) the user can write the name and the age of the

child who he/she wants to screen and upload a photo

for ID.

After he/she saves it, the application returns to

the initial menu and the button ”Collect” will be now

available, to initiate the collection with the respective

person/child registered. After the ﬁrst data collection

DesignofaMobileApplicationforEyeSignsScreening

521

Figure 3: Mockup of the start menu.

Figure 4: Mockup of the user registration.

Figure 5: Mockup illustrating the pop-up guide to the user.

made, the user may have access to the remaining but-

tons. For the option ”Collect” some instructions to

this feature will pop-up to guide the user through data

collection: to turn the lights down, to step away 4

steps from the person who he wants to trace (which

will be in front of him/camera), and to use the zoom

to capture the entire head (Figure 5). This way the

user can apply the deﬁned protocol and obtain useful

data for processing/screening.

When the user ﬁnishes reading the instructions

he/she can click the button to start collection. The

camera will initiate, and the algorithm for face recog-

nition will pop-up. When the detection of the eyes

and pupil is reached, the phone shoots the ﬁrst of 5

photos with ﬂash. Afterwards, arrows will appear on

the screen, indicating where to tilt the camera, and the

process of detection of the face, eyes and pupil will

repeat for right, left, up and down. When everything

is in the right position, the photo is taken automati-

cally, without requiring the user to press the ”shutter”

button, like in 6.

Figure 6: Mockup exemplifying one procedure of the trac-

ing method.

At the end, the user is prompted to choosebetween

sending the collection now or later to the server. If

he chooses now, this data will be sent to the server.

Choosing to send it later, he returns to the start menu

and may send it later in the ”Records” feature. In

option ”Alert” of the front menu, the user will have

indications to set an alarm of thirty days between col-

lections. After this, the calendar of his phone will

appear, where he can see other events that have been

scheduled. Clicking on the desired day, and setting

the time will set the alarm. Then, on the selected date,

he will receive a notiﬁcation to remind him of the col-

lection. Nonetheless the application is available for

data collection whenever the user decides to use it,

regardless of the alarm set. Finally, when the user

clicks on the button ”Records” (History), the applica-

tion will redirect him to the gallery where he can see

the different folders of previous data collection ses-

sions. In each screening, one folder will be created to

store all the information collected and corresponding

reports. When the user opens the selected folder, he

will have access to the photos recorded, getting 3 dif-

ferent options: ”Properties”, ”Delete” or ”Send”. In

”Properties” the non medical user can see the name,

date and hour of the photo, and also the age of the per-

son/child, in months, at the time of the collection. In

”Delete” he can erase it, and, in the ”Send” option he

can send the chosen session to the server, if it has not

been done before or if he wants to repeat it, receiving

a new report.

HEALTHINF2015-InternationalConferenceonHealthInformatics

522

4.1 Test Beds

The camera of the mobile phone needs to be con-

trolled in such a way that face, eyes, and pupils are au-

tomatically detected, and then the necessary controls

and instructions for the user are presented. Figure 7

is referent to the ﬁnal aspect of the camera preview.

Implementation of different sensors, like accelerom-

eter, to detect motion when the user needs to follow

the instructions, and light sensor to measure the il-

lumination of the room, would be very useful in the

application.

Figure 7: Camera’s preview of the mobile application.

The developed mobile application (”EyePic”) was

then tested following the required protocol and with-

out the use of any external hardware. Figure 8 shows

the result of the application with the proposed proto-

col and the white reﬂex necessary to evaluate the de-

gree of strabismus (”Hirschberg test”). The dual ﬂash

of the mobile phone leads pupil closure, not allow-

ing the presence of a red reﬂex, being this an issue to

evaluate in further tests.

Figure 8: Binocular photograph taken with ”EyePic” mo-

bile application.

This application was shown to some profession-

als of the area who express their opinion about the

designed protocol, respective mobile application and

early results. They agree that this kind of mobile ap-

plication is very useful for variousreasons, mainly be-

cause it is low-cost, but also because most parents are

not informed or do not even have knowledge on the

subject, the majority thinks they can not take a baby

to an appointment, that he will not cooperate. So, be-

cause nowadays everybody adhere to mobile applica-

tions quite well, this new one, easy to access, allow-

ing to perform a data collection at home and trans-

mitting important information, becomes very usefull

for an early ophthalmologist appointment. This is an

important tool to battle with a delayed diagnosis, hav-

ing, consequently, huge impact in the medical ﬁeld.

In fact, Br¨uckner test plays an important role in de-

tection of these eye pathologies, that must be detected

as soon as possible.The main concern is the quality of

the camera phone, which may not have the required

conditions for a complete assessment of lunar reﬂec-

tions like in the Br¨uckner test.

Relatively to the deﬁned protocol, the lighting in

the room should be low (as we noted when starting

the data collection) in order to allow scotopic vision

(vision produced by the eye in low light conditions)

and frontal photos will be the ideal, since taken from

another perspective may mislead.

A ”beta” version of this application will be tested

at Hospital de Braga in a screening project named

”Pimpolho”, aiming to make an evaluation of the eye

health of children aged 3 to 4 years, with special focus

on amblyopia or some amblyogenic risk factors.

5 DISCUSSION AND

CONCLUSIONS

In this paper, we propose the framework for eye

pathology screening using a mobile application, with

special emphasis in children and fast degrading eye

pathologies. The most common diseases that ben-

eﬁt from early detection, and respective signs were

described. This allowed for the design of a novel

data collection protocol for a range of eye pathology

screening, specially designed for non-medical users,

and with only the technology available on a smart-

phone: camera and ﬂash. In this application there is

no need for external devices. Besides the features pre-

sented before, in the future we intend to make avail-

able a video recording what could add more value to

the application, covering more eye pathologies, like

nystagmus or other evaluation for strabismus (Train-

ing, 2014). But there are some important issues to be

considered like the child’s attention during the record-

ing, and the accuracy required to implement it.

Given the importance of the eye diseases covered

in this work, early detection is essential to a timely

and effective treatment, making this mobile applica-

tion quite advantageous, due to its practical and easy

to use aspects for a regular monitoring, something

extremely important for this diseases with high inci-

dence during childhood.

In conclusion we can say that one of the strengths

of this application is its accessibility: it can reage ev-

eryone, since low income countries where medical in-

struments are scarce (this can become an alternative);

and mainly because allows a ﬁrst screening, collect-

ing usefull images to the detection of serious diseases.

Its great advantage is the ability to store all collec-

tions made by date (records / history), and thus can be

viewed by a medical expert. As parents with young

DesignofaMobileApplicationforEyeSignsScreening

523

children usually attend a pediatrician, and he only per-

forms any non routine eye exam in case of alarm, it

will be a great asset to have a visual history of the

development of eye disease (if that manifest) for an

accurate diagnosis. Also, a image repository can be

achievedwith this collection, not only for doctor eval-

uation, but for science research as well.

Besides all of that this is a very innovative mo-

bile application, there are no currently similar appli-

cations, to our knowledge, including the same fea-

tures an without the use of additional hardware to ob-

tain the imagens for the analysis (Gallagher, 2013).

ACKNOWLEDGEMENTS

The authors would like to acknowledge B´arbara Mota

(Faculty of Fine Arts of the University of Porto) and

the Ophthalmology Service of the Hospital de Braga.

This work was partially funded by Project I-CITY

- ICT for Future Health/Faculdade de Engenharia

da Universidade do Porto, NORTE-07-0124-FEDER-

000068, funded by the Fundo Europeu de Desen-

volvimento Regional (FEDER) through the Programa

Operacional do Norte (ON2) and by national funds

through FCT/MEC (PIDDAC).

REFERENCES

AAOP et al. (2008). Section on ophthalmology; american

association for pediatric ophthalmology and strabis-

mus, et al. red reﬂex examination in neonates, infants,

and children. Pediatrics, 122:1401–4.

Aguiar, A. S. C., Cardoso, M. V. L. M. L., and L´ucio, I.

M. L. (2007). Teste do reﬂexo vermelho: forma de

prevenc¸˜ao `a cegueira na infˆancia. Revista Brasileira

de Enfermagem, 60(5):541–545.

Alhammouri, Q., Gr¨af, M., and Lorenz, B. (2011).

Br¨uckner test sensitivity to detect foveal dimming.

JRMS, 18(3):10–15.

Amitava, A. K., Kewlani, D., Khan, Z., and Raz-

zak, A. (2010). Assessment of a modiﬁcation

of br¨uckner’s test as a screening modality for ani-

sometropia and strabismus. Oman journal of ophthal-

mology, 3(3):131.

Angart, L. (2014). Strabismus test. Vision Testing.

Anne (2013). How a photo can save a childs life.

Bethke, W. (2010). Review of ophthalmology. vision test-

ing? there’s an app for that.

E.C. Campos, G. V. N. (2002). Binocular Vision and Ocu-

lar Motility: Theory and Management of Strabismus,

volume 2, chapter 12 - Examination of the Patient - II:

Motor Signs in Heterophoria and Heterotropia, pages

174–176. Mosby, 6 edition.

Fund, D. E. C. (2010). Life and sight for every child. Daisys

Eye Cancer Fund - PhotoRED Technique.

Gallagher, J. (2013). Optician’s clinic that ﬁts a pocket -

peek. BBC News Health.

Gr¨af, M. (2010). The br¨uckner test revisited. In Pedi-

atric Ophthalmology, Neuro-Ophthalmology, Genet-

ics, pages 113–124. Springer.

J. Chhablani, S.Kaja, V. A. S. (2012). Smartphones

in ophthalmology. Smartphones in ophthalmology,

60(2):127–131.

J. Liebmann-Smith, E. J. (2010). Baby photos can reﬂect

serious eye disorders. Huffpost Healthy Living.

JornalNot´ıcias (2014). Menina salva de ﬁcar cega grac¸as a

uma foto no facebook.

Mail, D. (2007). The ’strange white light’ that nearly killed

baby grace.

Medical, N. (2011). Strabismus diagnosis.

Miller, J. M., Hall, H. L., Greivenkamp, J. E., and Guyton,

D. L. (1995). Quantiﬁcation of the br¨uckner test for

strabismus. Investigative ophthalmology & visual sci-

ence, 36(5):897–905.

P.J. Magalhes, I. M. V. (2009). Divulgac¸˜ao e treinamento

do teste do reﬂexo vermelho em rec´em-nascidos como

estrat´egia pol´ıtica em defesa da sa´ude ocular infan-

til no cear´a. Minist´erio da Sa´ude - Departamento de

Ciencia e Tecnologia.

Ramasubramanian, A. and Shields, C. L. (2012).

Retinoblastoma. JP Medical Ltd.

Rutstein, R. P., Cogen, M. S., Cotter, S. A., Daum, O.

K. M., Amos, J. F., Barry Barresi, O., Beebe, K. L.,

and Cavallerano, O. J. (2011). Optometric clinical

practice guideline care of the patient with strabismus:

Esotropia and exotropia. Lindbergh Blvd. St. Louis:

American Optometric Association.

Stampler, L. (2014). How posting a facebook picture saved

a 3 year olds sight.

TheRoyalCollegeofOphthalmologists (2012). Guidelines

for the management of strabismus in childhood. Tech-

nical report, The Royal College of Ophthalmologists -

Scientiﬁc Department.

Today, M. N. (2011). Remote diagnosis via camera phones.

Medical Devices and Diagnostics.

Training, V. (2014). Test for strabismus. Cover and Cross

Over tests.

Tvi24 (2014). Facebook salva visao de menina de trˆes anos.

Vela, C., Samson, E., Zunzunegui, M. V., Haddad, S.,

Aubin, M.-J., and Freeman, E. E. (2012). Eye care

utilization by older adults in low, middle, and high in-

come countries. BMC ophthalmology, 12(1):5.

HEALTHINF2015-InternationalConferenceonHealthInformatics

524