Anterior segment photography in pediatric eyes usingthe Lytro light field handheld noncontact cameraInna Marcus, MD,a Irene T. Tung, MD,a Eniolami O. Dosunmu, MD,a Warakorn Thiamthat, MD,a,b

and Sharon F. Freedman, MDa

PURPOSE To compare anterior segment findings identified in young children using digital photo-

Author affiliations: aDuke Eye COphthalmology Lerdsin HospitalThailandSubmitted May 15, 2013.Revision accepted August 3, 2Correspondence: Sharon F. Fr

Durham, North Carolina 2771Copyright � 2013 by the Am

Strabismus.1091-8531/$36.00http://dx.doi.org/10.1016/j.ja

572

graphic images from the Lytro light field camera to those observed clinically.

METHODS This was a prospective study of children\9 years of age with an anterior segment abnor-

mality. Clinically observed anterior segment examination findings for each child wererecorded and several digital images of the anterior segment of each eye captured withthe Lytro camera. The images were later reviewed by a masked examiner. Sensitivityof abnormal examination findings on Lytro imaging was calculated and compared to theclinical examination as the gold standard.

RESULTS A total of 157 eyes of 80 children (mean age, 4.4 years; range, 0.1-8.9) were included.

Clinical examination revealed 206 anterior segment abnormalities altogether: lids/lashes(n 5 21 eyes), conjunctiva/sclera (n 5 28 eyes), cornea (n 5 71 eyes), anterior chamber(n 5 14 eyes), iris (n 5 43 eyes), and lens (n 5 29 eyes). Review of Lytro photographs ofeyes with clinically diagnosed anterior segment abnormality correctly identified 133of 206 (65%) of all abnormalities. Additionally, 185 abnormalities in 50 children weredocumented at examination under anesthesia.

CONCLUSIONS The Lytro camera was able to document most abnormal anterior segment findings in

un-sedated young children. Its unique ability to allow focus change after image captureis a significant improvement on prior technology. ( J AAPOS 2013;17:572-577)

Obtaining anterior segment images in childrenusing slit-lamp-mounted cameras is challengingbecause patients often have difficulty remaining

images taken with the Lytro light field camera to thosefindings observed from the clinical anterior segmenteye examination.

still. Several studies have compared imaging using portablepersonal cameras to slit-lamp examination in children.1-5

Adjusting the focus when photographing a child’s eyewith a traditional handheld camera can prove challenging.

Light field photography, developed by Gabriel Lipp-mann in 1911, uses a microarray of lenses to capturemany images of a scene at slightly different angles.6 Theseimages can be reconstructed to produce a single three-dimensional master image. Building on prior work in thisfield, Ren Ng created the Lytro in 2012 (Lytro Inc, Moun-tain View, CA), the first commercially available light fielddigital camera (Figure 1A).7-8 The Lytro camera’s lightfield technology allows the focus to be adjusted afterimage capture.

The goal of this study was to compare anteriorsegment findings identifiable on digital photographic

enter, Durham, North Carolina; bDepartment of, Medical Bureau Ministry of Public Health, Bangkok,

013.eedman, MD, Duke Eye Center, 2351 Erwin Road,0 (email: sharon.freedman@duke.edu).erican Association for Pediatric Ophthalmology and

apos.2013.08.011

Subjects and Methods

This is a prospective study of consecutive patients under 9 years

of age presenting for a scheduled eye examination at the Duke

Eye Center by one pediatric ophthalmologist (SFF) from

September 2012 through January 2013. Inclusion criteria were

the presence of one or more abnormal anterior segment findings

in one or both eyes and written consent of a parent or guardian.

Children unable to cooperate with at least a penlight examination

of the anterior segment were excluded. The study was approved

by the Duke Hospital Institutional Review Board and complied

with guidelines of the US Health Insurance Portability and

Accountability Act of 1996.

Each subject received an anterior segment examination as

part of routine ophthalmological care. The anterior segment

findings for each eye were recorded, including lids and lashes,

conjunctiva and sclera, cornea, anterior chamber, iris, and lens.

The subject’s age and mode of anterior segment examination

(penlight, portable slit-lamp biomicroscope, or stationary

tabletop slit-lamp biomicroscope) were recorded.

Lytro Camera

The Lytro camera is 41 mm high by 41 mmwide by 112 mm long

and weighs 214 g (Figure 1). The 8 gigabyte model used in this

Journal of AAPOS

FIG 1. A, Lytro handheld camera. B, The camera is held approximately10 cm from the eye being imaged.

Volume 17 Number 6 / December 2013 Marcus et al 573

study can store up to 350 photographs. This camera has no

flash. It does have a zoom slider (up to 8� with a constant

focal length, f/2 lens), and a universal USB port. The camera

has a 38.6 mm square touch screen display for managing settings

and reviewing photographs. There are two basic modes, “every-

day” and “creative,” the latter allowing more control in setting

the range of focus available after the photograph is downloaded.

In “creative mode” the photographer can set the middle of the

focus range. In November 2012 a software update enabled the

photographer to manually set the shutter speed (8 to 1/250 sec-

onds) and/or the ISO (80 to 3200). The touch screen also allows

the user to adjust the shutter speed and ISO. While reviewing

the photographs on the camera, the photographer can refocus

the image on areas of interest.

The Lytro camera includes a proprietary application for image

storage and manipulation, and the software reflects the settings

used when the photographs were taken. The images can be

exported from the software as 1024 � 1024 JPEG files.

One software update that became available during the study

included some artistic filters and a function called “perspective

shift,” which allows the angle from which the photographs are

taken to be changed by several degrees in any direction: in this

mode, the image in the photograph appears to be three dimen-

sional due to the change in perspective. No caliper function is

currently available in the Lytro software.

Image Capture

Several digital photographs were taken of each eye with the Lytro

camera (by IM, ITT, or EOM). To ensure masked review of the

results, care was taken to avoid photographing identifying facial

features. The images were captured with room lighting most

Journal of AAPOS

appropriate to documenting the pathology. Lighting was adjusted

to suit the child’s level of cooperation as well as the pathology to

be documented; hence a penlight, portable slit lamp, and/or

tabletop slit-lamp biomicroscope provided additional lighting

for select cases. Total imaging time was less than 5 minutes.

There was no contact of the camera with the eye. Fluorescein

was instilled in the eyes to document pathology when appropriate

for clinical examination (Figure 2B).

The camera was always used in “creative mode” to allow for

control over the range of focus. The lens cap, which approxi-

mates the distance between the medial and lateral canthus, was

used as a target to set the focus range, with the lens cap filling

the field of view at approximately 50% of maximum zoom.

This permitted the photographer to hold the camera approxi-

mately 10 cm from the subject (Figure 1B). Most photographs

in this study were taken at a shutter speed ranging from 1/25

to 1/100 with the ISO set to automatic (range, 160-3200).

Shutter speeds ranging from 1/30 to 1/80 second minimized

motion artifacts and prevented overexposure with bright lighting

(such as with a slit-beam in a darkened room). In cases where the

size of an abnormality was of interest, such as large corneal

diameter, a ruler was held next to the subject (Figure 2E). All

Lytro images presented here have not been manipulated (with

the exception of Figure 3, where cropping allowed comparison

against similar RetCam image).

Image Review and Data Analysis

The photographic images were reviewed by a pediatric ophthal-

mologist (SFF) who was masked to patient identity and in

random order at least 1 month after the subject’s clinical visit.

Each could be refocused by the reviewer using the Lytro soft-

ware, without knowledge of the pathology noted on physical

examination. The reviewer was cautioned to document only

those abnormalities that were clearly visible in the photographs.

The physical examination findings for each eye were recorded

for each component of the anterior segment examination: lids

and lashes, conjunctiva and sclera, cornea, anterior chamber,

iris, and lens. Overall image quality and iris color (light vs

dark) were also judged by the reader. The primary outcome of

the study was success in detecting abnormal findings on image

review compared to examination. Study data was analyzed using

relational database software. All results are presented as mean �standard deviation unless otherwise noted. Because many

pathologic findings were unilateral or asymmetric, the unit of

evaluation was the eye rather than the child. Each examination

part was analyzed and recorded separately. Throughout the

study period, Lytro imaging was also used as a clinical tool to

capture selected findings during examination under anesthesia;

during some examinations RetCam photographs were also taken

when clinically indicated.

Results

A total of 157 eyes of 80 children were included. Meanage was 4.36 � 2.54 years (range, 0.1-8.9). Three subjectshad only one eye photographed due to poor cooperation.

FIG 2. Examples of Lytro camera images of the anterior segment. A, Right eye of 2-month-old infant referred with corneal opacities, with focuson examiner’s fingers. B, Post-capture focus on cornea showing opacities (arrow). C, Right eye of a 5-year-old girl born with Peters anomalywho has thinning of the border between the corneal transplant and host tissue showing image captured under diffuse handheld light. D, Photographof the same eye captured with blue light illumination from a portable slit lamp showing positive Seidel test after placement of fluorescein on the eye.E, Use of a ruler to measure a lid hemangioma in a 2-month-old (17 mm horizontal diameter). F, The right eye of a 6-year-old child with a pars planaglaucoma drainage tube behind a Sommering ring with posterior synechiae from the iris.

574 Marcus et al Volume 17 Number 6 / December 2013

Eleven subjects had anterior segment examination inclinic by penlight, 28 by portable slit lamp, and 41 bytabletop slit lamp. Of the 80 children, 46 (58%) haddark iris color. Seven subjects (9%) had photographsthat were of insufficient quality for one or both eyes tofully assess all examination parts. Subject movement wasthe cause of poor-quality photographs and this did notcorrelate with age. Eye color did not have a great effecton the ability to identify pathology in the photographs(data not shown).

Review of Lytro photographs revealed abnormalities ofthe lids and lashes in 21 eyes, of the conjunctiva and sclerain 28 eyes, of the cornea in 71 eyes, of the anterior chamberin 14 eyes, of the iris in 43 eyes, and of the lens in 29 eyes(Table 1). Overall, masked review of Lytro photographs ofeyes with clinically diagnosed anterior segment abnormal-ities correctly identified 133 of 206 (65%) abnormalities,with the highest rate of identification (79%) for lens abnor-malities in the setting of a dilated iris and the lowest rate ofidentification (50%) for anterior chamber abnormalities(Table 1).

Many different anterior segment pathologies werenoted on clinical examination (Table 2). The most com-mon pathologies of the various anterior segment compo-nents were as follows: lids and lashes, hemangioma;conjunctiva and sclera, glaucoma drainage device; cornea,Haab striae; anterior chamber, drainage device tube;iris, peaked or irregular pupil; and lens, intraocular lens(IOL) implant. The Sommering ring from lensectomywas identified more often on review of the imagesthan an IOL implant itself. The lens component wasexcluded in the analysis of eyes that were not dilated atthe time the image was captured. Some eyes had morethan one abnormality present per anterior segmentcomponent. Figure 2 and e-Supplements 1-1 to 1-6(available at jaapos.org) highlight some of the pathologiesmentioned above.

The Lytro camera was used as a clinical tool duringexamination under anesthesia to document findings in 76eyes of 50 patients (e-Supplements 1-6 to 1-24, availableat jaapos.org). Table 3 lists the various pathologies identi-fied for each examination component. A total of 132

Journal of AAPOS

FIG 3. A 19-month-old with acute glaucoma in aphakic left eye. Thevitreous is against the cornea and filling the anterior chambercausing iris bombe. Comparison of Lytro photographs (A, B) withRetCam photographs (C, D) using diffuse illumination (A, C) andslit beam (B, D).

Volume 17 Number 6 / December 2013 Marcus et al 575

distinct abnormalities were identified in the anteriorsegment components: lids/lashes (2), conjunctiva/sclera(31), cornea (31), anterior chamber (21), iris (29), andlens (45). Some eyes had more than one abnormality iden-tified. In selected cases, the Lytro camera demonstratedanterior segment features that were sometimes difficult tocapture by RetCam imaging (Figure 3).

Discussion

With the advent of portable personal cameras, severalstudies have compared ophthalmic findings seen on

Table 1. Comparison of anterior segment abnormalities identified by clinicaphotographs (n 5 80 children)

Anterior segmentcomponent

Total eyes withcomponent visibleon photographsa

Eyes with abnormalityon clinical

examination (%)

Fron

Lids/lashes 157 21 (13)Conjunctiva/ sclera 157 28 (18)Cornea 157 71 (45)Anterior chamberb 152 14 (9)Irisb 152 43 (28)Lensb,c 61 29 (48)

aOf 80 total subjects, 3 had only one eye photographed. Study eyes had anclinical examination.bThree subjects had unilateral corneal opacity and one had bilateral corneal op5 eyes.cThirty-eight subjects had one or both eyes dilated allowing examination of tand 3 right eyes were not photographed.

Journal of AAPOS

imaging to anterior segment examination findings in chil-dren.1-5 A recent study using traditional digital photo-graphy to assess trachoma in adults and children showeda high level of agreement between imaging and physicalexamination: 79.7% for follicles, 81.6% for trachomatousinflammation, and 89.3% for scarring.3

In our experience with the Lytro camera young chil-dren readily accepted the camera close to their faces,probably because its small blocklike shape is more inter-esting than threatening. The Lytro has few controlsand can be easily held and operated with one hand,freeing the other hand for a toy, penlight, or portableslit lamp. It is equally useful in older children who areable to position before a tabletop slit lamp. The camerais small enough to comfortably hold next to the slitlamp, with the slit beam illuminating the anteriorsegment. This versatility allows for both external andslit-lamp photographs and decreases the need to send pa-tients elsewhere for formal imaging. Furthermore, theLytro camera’s unique technology makes it ideal forophthalmic photography, in which the right focus iscrucial for capturing pathology.

The cost of the Lytro cameria is closer to that ofconsumer-grade digital cameras than to, for example,slit-lamp cameras and the RetCam. The Lytro can takemany photos in quick succession, and it requires nosetup. The Lytro has a few simple controls, and we foundthat settings rarely had to be changed to capture good-quality photographs. In creative mode, with a shutterspeed of 1/30 second and ISO set to automatic, we ob-tained clear, sharp photographs with good exposure inthe clinic setting, even of very young and very activechildren with variable levels of illumination. Because thecamera lacks a flash, patients cannot be surprised with asudden bright light, and there is little chance of imageoverexposure.

The main feature of the Lytro software is the abilityto refocus images on areas of interest after capture. Theperspective shift feature was not found to be useful in

l examination versus masked review of Lytro anterior segment

action of eyes with abnormalityphotograph review vs clinical

examination (%)Eyes with poor photographic

documentation (%)

15/21 (71) 019/28 (68) 037/71 (52) 4 (3)7/14 (50) 9 (6)32/43 (74) 4 (3)23/29 (79) 6 (10)

abnormality identified in one or more anterior segment component on

acities preventing examination of the anterior chamber, iris and lens in

he lens; 12 had only the left eye dilated, 3 had only the right eye dilated

Table 2. Detailed anterior segment abnormalities identified byclinical examination versus masked review of Lytro anteriorsegment photographs (n 5 80 children)

Anterior segmentcomponent

Abnormal finding byclinical examination Total eyes

Total identifiedon photographreview (%)

Lids/lashes Hemangioma 15 10 (67)Chalazion 3 3 (100)Dermoid 1 1 (100)Ptosis 1 0Crusting 1 1 (100)

Conjunctiva/sclera Glaucoma drainagedevicea

16 12 (75)

Conjunctival vesselabnormalityb

9 6 (67)

Scleral patch graft 6 4 (67)Other 8 6 (75)

Cornea Haab striae 19 13 (68)Large cornea 18 12 (67)Band keratopathy 8 8 (100)Corneal opacity 8 8 (100)Pannus 8 5 (63)Small cornea 7 5 (71)Scar 4 2 (50)Other 10 7 (70)

Anterior chamber Tube in AC 15 11 (73)Shallow AC 1 1 (100)Silicone oil in AC 1 0

Iris Peaked or irregularpupil

13 9 (69)

Aniridia 11 10 (90)Peripheraliridectomy

5 1 (20)

Posterior synechiae 3 3 (100)Optical iridectomy 2 2 (100)Sector iridectomy 2 2 (100)Pupillary membrane 2 2 (100)Iris strands 2 0Other 4 1 (25)

Lens Intraocular lensimplantc

12 10 (83)

Cataract, anteriorpolar

5 4 (80)

Aphakiac 5 5 (100)Open posteriorcapsule

4 0

Cataract, posteriorsubcapsular

3 0

Anterior lens plaque 2 1 (50)Other 3 1 (33)

AC, anterior chamber.aGlaucoma drainage device includes visualization of the tube shadow,device plate, or bleb over the implant.bConjunctival vessel abnormalities included prominent vessels in thesetting of hemangioma and dilated vessels in the setting of hyperemia.cIn 7 eyes with intraocular lens implant, Sommering ring was theabnormality identified. In 2 cases with aphakia, Sommering ring wasidentified. In these cases the Sommering ring was counted as a correctidentification for the absence of natural lens.

Table 3. Anterior segment abnormalities identified duringexamination under anesthesia which were documented by Lytrophotography (n 5 50 children)

Anterior segmentcomponent

Abnormal finding by clinicalexamination Total eyes

Lids/lashes Lower lid scarring with lash loss 2Crusting 1

Conjunctiva/sclera Glaucoma drainage devicea 10Conjunctival vessel abnormalityb 7Scleral patch graft 5Abnormal limbus 3Shallow fornix 2Enlarged lacrimal gland 2Scleral thinning 2Other 3

Cornea Corneal opacity 14Haab striae 12Corneal laceration 2Band keratopathy 3Other 5

Anterior chamber Anterior chamber tube 16Shallow anterior chamber 3Hyphema 2Other 4

Iris Irregular or peaked pupil 11Iris atrophy 5Posterior synechiae 4Peripheral iridectomy 3Brushfield spots 2Sector iridectomy 2Ectropion uvea 2Other 9

Lens Sommering ring 17Aphakia 9Intraocular lens implant 8Cataract, nuclear 6Cataract , cortical 3Cataract, posterior subcapsular 2Other 4

aGlaucoma drainage device includes visualization of the tube shadow,device plate, or bleb over the implant.bConjunctival vessel abnormalities included prominent vessels in thesetting of hemangioma and dilated vessels in the setting of hyperemia.

576 Marcus et al Volume 17 Number 6 / December 2013

the present study, but it could theoretically be used toconstruct stereographic images.

This study has several limitations. It was not designedto determine the utility of the camera as a screening

tool because the sample contained 100% pathology.Thus we cannot demonstrate the specificity of detectingpathology from the photographic images. The Lytrocamera can be used only for external and anterior segmentabnormalities; it cannot image the posterior segmentor capture video. While the present study demonstratesthat a wide range of abnormalities can be photographed,interpretation can be challenging without clinicaldocumentation for comparison. Despite steps taken tomaximize masking, we did not achieve complete maskingin this study because the clinical examination andthe photographic review were performed by the same per-son, who was uniquely qualified to identify the broadrange of pathology in this sample population. The corneaand anterior chamber posed the greatest challenges toidentifying abnormalities, probably because these compo-nents are transparent. Moreover, the tear film of thecornea caused reflection artifacts in some cases. Finally,

Journal of AAPOS

Volume 17 Number 6 / December 2013 Marcus et al 577

this study does not compare the Lytro camera to otherdigital cameras. Furthermore, due to the short attentionspan of our young subjects it was impractical to use twocameras.This study showed moderately successful results with

masked examiner detection of anterior segment pathologyin Lytro images; nevertheless, the camera’s ability tocapture a wide range of anterior segment pathologies,its unique ability to refocus after image capture, itslight weight and small size, and its affordability make itan excellent tool for supplementing other forms of clinicaldocumentation.

Acknowledgments

The authors thank Laura B. Enyedi MD, for providing the Lytrophotographs for e-Supplements 1-9 and 1-20.

Seeing Is Believing

When Falchuk told me that “the picture didn’taphor. Donald Redelmeier, a physician at Sunonto, has a particular interest in physican corefers to a phenomenon called the “eyeball test,tifies “something intangible yet unsettling in thof course, be wrong. But it should not be ignorecognize that the information before him has

—Jerome Groopman, How Doctors Think (Bost

Contributed by Alex V. Levin, MD, MHSc, Philad

Journal of AAPOS

References

1. Robinson J, Gilmore KJ, Fielder AR. Validation of a photographicmethodofmeasuring corneal diameter. Br JOphthalmol 1989;73:570-73.

2. Lagreze WA, Zobor G. A method for noncontact measurement ofcorneal diameter in children. Am J Ophthalmol 2007;144:141-2.

3. Roper KG, Taylor HR. Comparison of clinical and photographicassessment of trachoma. Br J Ophthalmol 2009;93:811-14.

4. Puvanachandra N, Lyons CJ. Rapid measurement of corneal diameterin children: validation of a clinic-based digital photographic technique.J AAPOS 2009;13:287-8.

5. Cranston ME, Mhanni AA, Marles SL, Chudley AE. Concordance ofthree methods for palpebral fissure length measurement in the assess-ment of fetal alcohol spectrum disorder. Can J Clin Pharmacol 2009;16:e234-41.

6. Integral photography—a new discovery by Professor Lippmann.Scientific American 1911;164.

7. Adelson EH, Wang JYA. Single lens stereo with a plenoptic camera.IEEE Trans Pattern Anal Mach Intell 1992;14:99-106.

8. Ng R, et al. Light field photography with a hand-held plenopticcamera. Computer Science Technical Report CSTR 2005;2:11.

fit,” his words were more than mere met-nybrook Health Sciences Centre in Tor-gnition and its relation to diagnosis. He” the pivotal moment when a doctor iden-e patient’s presentation” The instinct may,red, because it can cause the physician tobeen improperly “framed.”

on: Houghton Mifflin, 2007), 21-22.

elphia, Pennsylvania