J CATARACT REFRACT SURG - VOL 32, DECEMBER 2006

Burst-shot infrared digital photography

to determine scotopic pupil diameter

Hongseok Yang, MD, Marvin Lee, MD, Jun Bum Kim, MD, Jaehong Ahn, MD

PURPOSE: To evaluate the usefulness of infrared digital photography for determining scotopic pupilsize by comparing infrared digital photography with a Colvard pupillometer (Oasis Medical).

SETTING: Department of Ophthalmology, Ajou University School of Medicine, Suwon, Korea.

METHODS: Scotopic pupil size in 50 eyes of 25 healthy individuals was measured with a Colvard pu-pillometer and a digital camera (DSC-F828) using infrared burst shots after 5 minutes of dark adapta-tion. Measurements were performed by 2 independent examiners (E1 and E2). The digital photographimages were read using the ruler function of Adobe Photoshop 7.0 by 2 independent readers (R1 andR2). Agreement and repeatability were analyzed using the comparison method described by Blandand Altman.

RESULTS: The mean scotopic pupil diameter measured using the Colvard pupillometer was 6.69 mm G0.78 (SD) (E1) and 6.70 G 0.71 mm (E2). The mean scotopic pupil diameter measured from the dig-ital photograph images was 6.67 G 0.75 mm (E1) and 6.66 G 0.78 mm (E2). The mean difference be-tween E1 and E2 with both devices was small; however, the result with the infrared digital camera wasmarginally smaller than with the Colvard pupillometer. The limits of agreement were –0.01 G 0.70 mmwith the Colvard pupillometer and 0.01 G 0.20 mm with the digital photograph image. The digitalphotograph image showed better agreement. The coefficient of interrater repeatability was smallerfor the digital photograph image (0.39) than for the Colvard pupillometer (0.70).

CONCLUSIONS: Scotopic pupil measurement using an infrared digital camera with a burst shot hadgood agreement with the Colvard pupillometer and better repeatability. The infrared digital camera isless expensive, and pupil unrest can be overcome by taking serial images.

J Cataract Refract Surg 2006; 32:2113–2117 Q 2006 ASCRS and ESCRS

After refractive corneal or lens surgery, some patients report

decreased contrast sensitivity, increased glare, or nighttime

halos when their pupils exceed the optical zone. Thesedecreases in visual quality are the result of the optical glare

phenomenon at the optical transition zone of the cornea

or at the intraocular lens (IOL) optic edge. Therefore,

Accepted for publication June 5, 2006.

From the Department of Ophthalmology, Ajou University Schoolof Medicine, Suwon, Korea.

Presented as a poster at the 93rd Annual Meeting of the KoreanOphthalmological Society, Busan, Korea, April 2005.

No author has a financial or proprietary interest in any material ormethod mentioned.

Corresponding author: Jaehong Ahn, MD, Department ofOphthalmology, Ajou University School of Medicine, San 5,Wonchon-dong, Yeongtong-gu, Suwon, 443-729, Korea. E-mail:chrisahn@ajou.ac.kr.

Q 2006 ASCRS and ESCRS

Published by Elsevier Inc.

measuring pupil size is an important part of the preopera-

tive examination for refractive surgery, especially under

scotopic conditions. Refractive surgeons can determinethe diameter of the treatment zone using this measurement,

and an exact measurement of scotopic pupil size is an

essential parameter for determining patient suitability for

refractive surgery.1–4

To measure pupil diameter exactly, the illumination

and pupil unrest (the hippus cycle) must be considered.

Precise determination of the scotopic condition can be

made after dark adaptation for several minutes. Multiplemeasurements are needed for an exact determination of

pupil size because pupil unrest occurs continually and at

all levels of illumination.5

The device most commonly used for measuring scoto-

pic pupil size is a handheld infrared pupillometer (Colvard,

Oasis Medical). However, this device is expensive and er-

rors can result as it does not take pupil unrest into consid-

eration. Moreover, other errors can arise when different

0886-3350/06/$-see front matterdoi:10.1016/j.jcrs.2006.06.039

2113

BURST-SHOT INFRARED DIGITAL PHOTOGRAPHY FOR SCOTOPIC PUPIL SIZE

examiners read the scale. More accurate automatic devices

such as the Procyon instrument (Oasis Medical), Zywave

aberrometer (Bausch & Lomb), or WASCA aberrometer

(Asclepion-Meditec-Zeiss) are expensive.

This study shows the usefulness of a digital camera

with infrared function, which is less expensive and cantake continuous serial images of pupils (2.5 shots per sec-

ond) to overcome pupil unrest for determining scotopic

pupil size.

SUBJECTS AND METHODS

Scotopic pupil size was measured in 50 eyes of 25 healthy pa-tients. Ocular pathology, ocular surgery, and the use of topical orsystemic medication with potential effects on pupil size or behav-ior were exclusion criteria.

Measurements were taken with a Colvard handheld infraredpupillometer and a digital camera (Cyber Shot DSC-F828, SonyElectronics) with night-shot function after 5 minutes of dark ad-aptation. For illuminance measurements, 2 readings were takenwith an illuminance meter (Minolta) and the illuminance of thedark room was set to under 0.1 lux. Each examination wasperformed by 2 independent examiners (E1 and E2).

Colvard Pupillometer

The Colvard pupillometer uses light-amplification technol-ogy. The examiner can focus on the iris and pupil by moving thepupillometer forward and backward slightly. The patients wereasked to fixate on an infrared light-emitting diode that emits redlight at very low levels. In the device, a millimeter ruler was super-imposed by a reticule. In this study, the examiners took the hippuscycle into consideration and attempted to measure the greatesthorizontal scotopic pupil diameter. The examiners attempted toestimate the pupil size with a precision of 0.5 mm.

Infrared Digital Photography

The digital camera was mounted on a tripod in a darkroom,and the subject sat in front of the camera. To eliminate the effect ofaccommodation, the subject stared at a dark empty space behindthe camera. A millimeter ruler was placed in front of the superiororbital rim parallel to the long axis of the interpalpebral fissure.Seven sequential photographs were taken using burst shots toovercome the hippus cycle (2.5 photographs per second). Thedigital photograph images containing the patient’s pupil and themillimeter ruler were read using the ruler function of AdobePhotoshop 7.0 (Figure 1). The largest pupil size from the 7sequential photographs was selected.

To evaluate the interreader error that can occur with digitalphotograph image reading using Adobe Photoshop 7.0, readers1 and 2 (R1 and R2) read the digital images taken by E1.

Statistical Analysis

Statistical analysis was performed with a paired t test and themethod described by Bland and Altman6 using MedCalc software(version 8, Mariakerke). For each pupillometry technique, thecoefficient of interrater repeatability was defined as twice thestandard deviation of the differences between the 2 examiners’

J CATARACT REFRACT SURG -2114

measurements. Lower values indicated higher repeatability. A Pvalue of 0.05 or less was considered statistically significant.

RESULTS

The mean patient age was 28.0 years G 4.0 (SD) (range

20 to 36 years). The male-to-female ratio was 15:10.

Agreement of Pupil Size Measured by Each Device

The mean scotopic pupil diameter measured using the

Colvard pupillometer was 6.69 G 0.78 mm (range 5.0

to 8.0 mm) (E1) and 6.70 G 0.71 mm (range 5.0 to

8.0 mm) (E2). The mean scotopic pupil diameter measured

from the digital photograph images was 6.67 G 0.75 mm

(range 4.72 to 8.03 mm) (E1) and 6.66 G 0.78 mm (range

4.68 to 8.03 mm) (E2). The scotopic pupil size measuredby the same examiner using the 2 methods did not different

significantly for either examiner (P Z .622 and P Z .678,

paired sample t test). When the same examiner (E1)

measured pupil size using the 2 different devices, the mean

difference between devices was 0.02 G 0.35 mm and the

limit of agreement (LOA) was 0.02 G 0.69 mm (Table 1

and Figure 2, A).

Repeatability of Each Device Between Examiners

The mean difference between the examiners with the

Colvard pupillometer and the digital photograph image

was not statistically significant (P Z .844 and P Z .428,

paired sample t test). For the Colvard pupillometer, the dif-

ference was 0.01 G 0.36 mm and for the digital photograph

image, 0.01 G 0.10 mm. The mean difference for the digitalphotograph images was marginally smaller than for the

Colvard pupillometer. The LOA calculated from the Bland

and Altman plot was �0.01 G 0.70 mm with the Colvard

Figure 1. Infrared digital photograph image with the ruler function of

Adobe Photoshop 7.0.

VOL 32, DECEMBER 2006

BURST-SHOT INFRARED DIGITAL PHOTOGRAPHY FOR SCOTOPIC PUPIL SIZE

pupillometer and �0.01 G 0.20 mm with the digital pho-tograph images. The scotopic pupil diameter measured

from the digital photograph images showed better agree-

ment than the diameter measured using the Colvard pupil-

lometer (Table 1 and Figure 2, B and C). The coefficient of

the interrater repeatability was smaller for the digital pho-

tograph image method (0.39) than for the Colvard pupil-

lometer (0.70). The range of the mean differences

between examiners was smaller with the digital photographimages than with the Colvard pupillometer (Table 2).

Agreement of Digital Photo Image Readingby 2 Different Readers

The mean difference between readers was small, 0.00 G0.20 mm, and the LOA was 0.00 G 0.39 mm (Table 3 and

Figure 2, D).

DISCUSSION

Recently, refractive surgery for myopia has become

popular worldwide. Refractive surgery for the cornea and

Table 1. Limits of agreement between the Colvard pupillometer and

digital photo Image (E1).

ComparisonMean Difference

G SD (mm)LOA

(mm)*Range(mm)

Colvard – digital photo 0.02 G 0.35 �0.67, 0.72 1.39

LOA Z limits of agreement

*Mean difference G 2 SD of the difference between Colvard and digital

photograph measurements

J CATARACT REFRACT SURG - V

the implantation of phakic IOLs (pIOLs) can increase ocu-

lar aberrations including glare, halos, and other visual qual-

ity problems. These phenomena seem to result from a small

decentered treatment zone, stromal opacity, and small de-

centered pIOL optics.1–4,7,8 Night-vision disturbance and

optical glare appear when the pupil dilates. Night visualfunction and quality of life can be affected.2,9 To prevent

these complications, precise scotopic pupil size measure-

ment is essential and the individual value must be consid-

ered when determining the treatment zone diameter.

External (illumination and drugs) and intrinsic factors

(psychological state, age, and refraction) affect pupil

size.5,10 In addition, pupil unrest makes a single measure-

ment of pupil size inaccurate; more than 1 measurementshould be made and the largest pupil size used for

correction.

Many comparative studies have sought the most reli-

able technique for measuring pupil size (Table 4). Schnit-

zler et al.11 compared an infrared pupillometer (Colvard)

with a video-driven infrared pupillometer (Video Vision

Analyzer [VIVA]) under dim light to evaluate scotopic

pupil diameter in a normal population. The mean scotopicpupil diameter was 6.08 G 1.16 mm (range 3.2 to 8.4 mm)

with the Colvard pupillometer and 6.24 G 1.28 mm (range

3.5 to 9.0 mm) with the VIVA pupillometer. The mean dif-

ference between examiners was 0.23 G 0.59 mm for VIVA

and 0.01 G 0.37 mm for the Colvard pupillometer. The

LOA was 1.4 (Colvard) versus 2.4 (VIVA). The coefficients

of repeatability were 0.7 (Colvard) and 1.1 (VIVA). Mea-

surements with the Colvard pupillometer were more reli-able and precise than those with the VIVA pupillometer.

Figure 2. A: Bland-Altman plot showing the inter-

instrument repeatability of the Colvard pupillometer

and infrared digital photograph image. B: Bland-

Altman plot showing the interrater repeatability of

the Colvard pupillometer. C: Bland-Altman plot show-

ing the interrater repeatability of the infrared digital

photograph image. D: Bland-Altman plot showing

the interreader repeatability of the infrared digital

photograph image by 2 different readers.

OL 32, DECEMBER 2006 2115

BURST-SHOT INFRARED DIGITAL PHOTOGRAPHY FOR SCOTOPIC PUPIL SIZE

Kohnen et al.12 compared the Procyon digital infrared

pupillometer and the handheld infrared Colvard pupillom-

eter for determining scotopic pupil size. The mean scotopic

pupil diameter was 5.90 G 0.97 mm with the Procyon dig-

ital infrared pupillometer and 5.78 G 0.98 mm with theColvard pupillometer. The coefficients of repeatability

were 0.64 (Procyon) and 1.16 (Colvard). The LOA was

–0.10 G 0.64 mm (Procyon) and 0.13 G 1.16 mm (Col-

vard). Although the differences between the readings

with the Procyon and Colvard devices appear to be small,

the digital pupillometer currently provides the most accu-

rate measurement of the scotopic pupil.

Schmitz et al.13 compared the WASCA workstationwavefront aberrometer, Procyon digital infrared pupillom-

eter, and Colvard pupillometer for determining scotopic

pupil size. Of the 3 instruments, the Colvard pupillometer

gave the largest pupil diameter.

Using infrared video recording as the standard, Twa

et al.14 evaluated the accuracy and repeatability of the dig-

ital photography technique and compared measurements

Table 2. Limits of agreement between E1 and E2.

DeviceMean Difference

G SD (mm)LOA

(mm)*Range(mm)

Colvard �0.01 G 0.36 �0.71, 0.69 1.40Digital photo 0.01 G 0.10 �0.21, 0.19 0.40

LOA Z limits of agreement

*Mean difference G 2 SD of the difference between 2 examiners

J CATARACT REFRACT SURG -2116

with this technique to those taken with the Colvard pupill-

ometer, a series of semicircular templates, and a millimeter

ruler. The Colvard pupillometer was better than other tech-

niques with respect to simplicity and test time. In contrast,

digital photography was more repeatable and accurate than

the common clinical techniques. Although not as quick as

other methods, digital photography is relatively inexpen-

sive and provides lasting documentation and independentgrading suitable for clinical research purposes.

We used the infrared function of an inexpensive com-

mercial digital camera without a flash. The mean scotopic

pupil diameters measured using the Colvard pupillometer

were 6.69 G 0.78 mm (E1) and 6.70 G 0.71 mm (E2)

and those obtained using the digital photograph images,

6.67 G 0.75 mm (E1) and 6.66 G 0.78 mm (E2). There

were no statistically significant differences. Our datashow larger pupil diameters than in previous studies.

This may have resulted from taking the measurement in

a perfectly dark room without a flash and the younger

ages of our subjects. The mean age of our subjects was

Table 3. Limits of agreement between 2 different readers of digital

photograph Image.

ComparisonMean Difference

G SD (mm)LOA

(mm)*Range(mm)

Reader 1 – reader 2 0.00 G 0.20 �0.38, 0.39 0.77

LOA Z limits of agreement

*Mean difference G 2 SD of the difference between 2 readers

Table 4. Comparison of scotopic pupil size measured with different devices.

Age (Y) Scotopic Pupil Size (mm)

Study* Mean G SD Range Device Mean G SD Range Coefficient of Interrater Repeatability

Schnitzler11 36 G 9 19–55 Colvard 6.08 G 1.18 (E1) d 0.746.08 G 1.13 (E2) d

VIVA 6.35 G 1.34 (E1) d 1.186.12 G 1.21 (E2) d

Rosen5 46 G 11.7 d Procyon 6.61 d dKohnen12 38.8 G 10.7 20–59 Procyon 5.85 G 1.01 (E1) d 0.64

5.95 G 0.93 (E2) dColvard 5.84 G 1.03 (E1) d 1.16

5.71 G 0.92 (E2) dSchmitz13 23 G 3 d Colvard 6.61 4.78–8.11 d

Procyon 6.56 4.78–8.11 dWASCA 6.30 4.84–7.12 d

Current 28.0 G 4.0 20–36 Colvard 6.69 G 0.78 (E1) d 0.706.70 G 0.71 (E2) d

Infrared camera 6.67 G 0.75 (E1) d 0.396.66 G 0.78 (E2) d

E1 Z examiner 1; E2 Z examiner 2

*First author

VOL 32, DECEMBER 2006

BURST-SHOT INFRARED DIGITAL PHOTOGRAPHY FOR SCOTOPIC PUPIL SIZE

28 years versus about 36 years in previous studies. Ryou

et al.15 compared scotopic pupil size in a dark room. The

mean scotopic pupil size was 7.5 G 0.8 mm in younger pa-

tients (range 20 to 29 years) and 6.3 G 0.5 mm in older

patients (range 40 to 59 years).

In our study, both instruments gave the same pupil sizemeasurement when pupil size was measured by the same

examiner (E1) using 2 different devices. The mean differ-

ence between devices was 0.02 G 0.35 mm, and the LOA

was 0.02 G 0.69 mm. In the study by Schnitzler et al.,11

which compared the Colvard and VIVA pupillometers,

the mean difference between devices was 0.05 G 0.7 mm

and the LOA was 0.05 G 1.45 mm, so our results showed

better agreement between devices.The digital photograph image had a better LOA calcu-

lated from the Bland-Altman plot: –0.01 G 0.70 mm with

the Colvard pupillometer and 0.01 G 0.20 mm with the

digital photograph images. In other words, the digital pho-

tograph image method had better repeatability. In fact, the

coefficient of interrater repeatability was smaller using the

digital photograph image method (0.39) than with the Col-

vard pupillometer (0.70). Schnitzler et al.11 report acoefficient of interrater repeatability of the VIVA infrared

pupillometer and Colvard pupillometer of 1.18 and 0.74,

respectively. Kohnen et al.12 report a coefficient of interrat-

er repeatability of the Procyon digital infrared pupillometer

of 0.64 and of the Colvard pupillometer of 1.16. We

conclude the repeatability of digital photograph images

for measuring scotopic pupil diameter is excellent.

As Rosen et al.5 point out, pupil unrest can lead to anunexpected result in determining scotopic pupil size.

Therefore, physicians measure pupil size more than once.

We took 7 sequential photographs (2.5 photographs per

second) using a burst shot with the camera and used the

maximum value as the pupil size for each subject to over-

come pupil unrest.

Cheng et al.16 state that the largest recorded pupil size

might represent the scotopic pupil better than an average.We agreed and used the largest value as the scotopic pupil

size. It takes more time to read 7 digital photograph images

and choose the largest value than to use a CCD camera

image with software such as a Zywave aberrometer and

Procyon pupillometer, but the latter devices are very

expensive.

In conclusion, infrared digital photography for mea-

suring scotopic pupil size showed good agreement withthe Colvard pupillometer. The pupil size measured using

each device did not differ significantly. The infrared digital

J CATARACT REFRACT SURG

photography produced better interrater repeatability than

the Colvard pupillometer and had good agreement between

readers. Although not as quick and easy as automatic de-

vices, infrared digital photography is economical and had

good interrater repeatability. Moreover, pupil unrest can

be overcome by taking serial pupil images and the physi-cian can obtain more accurate data. Therefore, burst-shot

infrared digital photography can be used as another

method for measuring scotopic pupil size in clinical

practice.

REFERENCES

1. Roberts CW, Koester CJ. Optical zone diameters for photorefractive

corneal surgery. Invest Ophthalmol Vis Sci 1993; 34:2275–2281

2. Nixon WS. Pupil size in refractive surgery [letter]. J Cataract Refract

Surg 1997; 23:1435–1436

3. Davidorf JM, Salz J, Campion M, Hoffman RS. Response to letter by

WS Nixon. J Cataract Refract Surg 1998; 24:291–294

4. Colvard M. Preoperative measurement of scotopic pupil dilation

using an office pupillometer. J Cataract Refract Surg 1998; 24:

1594–1597

5. Rosen ES, Gore CL, Taylor D, et al. Use of a digital infrared pupillometer

to assess patient suitability for refractive surgery. J Cataract Refract

Surg 2002; 28:1433–1438

6. Bland JM, Altman DG. Statistical methods for assessing agreement be-

tween two methods of clinical measurement. Lancet 1986; 1:307–310

7. Maguire LJ, Bechara S. Epithelial distortions at the ablation zone

margin after excimer laser photorefractive keratectomy for myopia

[letter]. Am J Ophthalmol 1994; 117:809–810

8. Alio JL, de la Hoz F, Perez-Santonja JJ, et al. Phakic anterior chamber

lenses for the correction of myopia; a 7-year cumulative analysis of

complications in 263 cases. Ophthalmology 1999; 106:458–466

9. Applegate RA, Howland HC. Refractive surgery, optical aberrations,

and visual performance. J Refract Surg 1997; 13:295–299

10. Winn B, Whitaker D, Elliott DB, Phillips NJ. Factors affecting light-adap-

ted pupil size in normal human subjects. Invest Ophthalmol Vis Sci

1994; 35:1132–1137

11. Schnitzler E-M, Baumeister M, Kohnen T. Scotopic measurement of

normal pupils: Colvard versus Video Vision Analyzer infrared pupill-

ometer. J Cataract Refract Surg 2000; 26:859–866

12. Kohnen T, Terzi E, Buhren J, Kohnen E-M. Comparison of a digital and

a handheld infrared pupillometer for determining scotopic pupil

diameter. J Cataract Refract Surg 2003; 29:112–117

13. Schmitz S, Krummenauer F, Henn S, Dick HB. Comparison of three dif-

ferent technologies for pupil diameter measurement. Graefes Arch

Clin Exp Ophthalmol 2003; 241:472–477

14. Twa MD, Bailey MD, Hayes J, Bullimore M. Estimation of pupil size by

digital photography. J Cataract Refract Surg 2004; 30:381–389

15. Ryou J-H, Kim H-W, Park M-G, Paik H-J. Pupil size in the dark in normal

adults. J Korean Ophthalmol Soc 1996; 37:1722–1729

16. Cheng ACK, Rao SK, Lam DSC. Reply to letter by AM Khanani,

SM Brown: Determining scotopic pupil size. J Cataract Refract Surg

2005; 31:1267–1268

- VOL 32, DECEMBER 2006 2117