Optimizing Hand-Held Spectral Domain Optical Coherence Tomography Imaging for Neonates, Infants, and Children

Retina Optimizing Hand-held Spectral Domain Optical Coherence Tomography Imaging for Neonates, Infants, and Children

Ramiro S. Maldonado,1 Joseph A. Izatt,1,2 Neeru Sarin,1 David K. Wallace,1,3 Sharon Freedman,1,3 C. Michael Cotten,3 and Cynthia A. Toth1,2

PURPOSE. To describe age-related considerations and methods diagnostic tool for management of vitreoretinal diseases, par- to improve hand-held spectral domain optical coherence to- ticularly involving the macula, since it provides information on mography (HH-SD OCT) imaging of eyes of neonates, infants, retinal architecture beyond that obtained by conventional oph- and children. thalmic methods.1,2 In older children, OCT has been shown to METHODS. Based on calculated optical parameters for neonatal be more sensitive than clinical examination in the detection of and infant eyes, individualized SD OCT scan parameters were retinal conditions such as edema, subretinal fluid, and retinal thinning.3 developed for improved imaging in pediatric eyes. Forty-two 4 subjects from 31 weeks postmenstrual age to 1.5 years were With 1.4 million blind children below age 15 worldwide and more than 50,000 legally blind children in the United imaged with a portable HH-SD OCT system. Images were ana- 5 lyzed for quality, field of scan, magnification, and potential States, there is a need to access this technology for use in clinical utility. younger children, particularly since the most common causes of ocular visual impairment are retinal diseases,6 such as reti- RESULTS. The axial length of the premature infant eye increases nopathy of prematurity (ROP), ocular albinism, and retinal rapidly in a linear pattern during the neonatal period and slows dystrophies.7,8 progressively with age. Refractive error shifts from mild myo- A few studies have reported time domain (TD) OCT and pia in neonates to mild hyperopia in infants. These factors spectral domain (SD) OCT imaging in the pediatric population affect magnification and field of view of optical diagnostic tools (Table 1). Although numerous publications show the advan- applied to the infant eye. When SD OCT parameters were tage of SD OCT ocular imaging in the adult,22–24 the higher corrected based on age-related optical parameters, SD OCT speed of scanning is even more important in young children image quality improved in young infants. The field of scan and with limited attention spans and difficulty with sustained fixa- ease of operation also improved, and the optic nerve, fovea, tion, as shown by Chong et al.12 in children with albinism. and posterior pole were successfully imaged in 74% and 87% of Scott et al.10 also demonstrated how SD OCT of infant eyes may individual eye imaging sessions in the intensive care nursery influence management in cases of shaken-infant syndrome. and clinic, respectively. No adverse events were reported. Chavala et al.9 described preretinal structures, schisis, and CONCLUSIONS. SD OCT in young children and neonates should retinal detachment found on SD OCT, but not on conventional be customized for the unique optical parameters of the infant clinical examination in infants with ROP. eye. This customization, not only improves image quality, but In those studies, limiting factors for imaging children were also allows control of the density of the optical sampling patient motion, poor fixation, patient position, and different directed onto the retina. (Invest Ophthalmol Vis Sci. 2010;51: optics compared with the adult eye. 2678–2685) DOI:10.1167/iovs.09-4403 In this study, we evaluated problems specific to imaging pediatric patients with SD OCT and identified and tested tech- ptical coherence tomography (OCT) is a diagnostic imag- nical corrections to solve these challenges. Oing technique that provides cross-sectional images of human retinal morphology in vivo. It has become a standard METHODS

We used a U.S. Food and Drug Administration–approved portable, 1 2 - From the Departments of Ophthalmology, Biomedical Engineer noncontact, hand-held SD OCT unit (Bioptigen Inc., Research Triangle ing, and 3Pediatrics, Duke University, Durham, North Carolina. Supported by grants from Angelica and Euan Baird; The Hartwell Park, NC) consisting of an imaging hand piece connected via a 1.3-m Foundation; Grant 1UL1 RR024128-01 from the National Center for flexible cable to an SD OCT engine mounted on a rolling cart. The SD Research Resources (NCRR), a component of the National Institutes of OCT system has a calibrated knob to adjust the reference arm position Health (NIH); and NIH Roadmap for Medical Research. The contents of with digital readout, as shown in Figure 1A; a manufacturer-supplied the article are solely the responsibility of the authors and do not calibration factor was used to convert the readout units to optical necessarily represent the official view of NCRR or NIH. distance in millimeters. The hand-held probe has a focus correction Submitted for publication July 30, 2009; revised October 7 and adjustment with a range of ϩ10 to Ϫ12 D (Fig. 1B). December 1, 2009; accepted December 14, 2009. On February 4, 2010, the manufacturer (Bioptigen Inc.) notified us Disclosure: , None; , Bioptigen Inc. R.S. Maldonado J.A. Izatt that the retinal scan length settings for this hand-held SD OCT unit (I, C, P); N. Sarin, None; D.K. Wallace, None; S. Freedman, None; were in error; and that each lateral scan setting label on this particular C.M. Cotten, None; C.A. Toth, Bioptigen Inc. (F), Alcon (F, C), Genentech (F, C) system created a scan that was 68% of the labeled length. Therefore, all Corresponding author: Cynthia A. Toth, Duke University Eye Cen- scan lengths (both x and y dimensions) recorded off the unit had to be ter, 2351 Erwin Rd, Box 3802, Durham, NC, 27710; multiplied by the 0.68 correction factor to represent the actual scan [email protected]. lengths. These corrections have been applied, and thus many scan

Downloaded from jov.arvojournals.org on 09/25/2021 IOVS, May 2010, Vol. 51, No. 5 Technical Improvements to Optimize HH-SD OCT for Neonates 2679

TABLE 1. Publications Using OCT in the Pediatric Population, by Age Range

Patients Authors Topic (n) Age Range Type of OCT

Chavala S, et al.9 ROP 3 32, 37, 38 weeks SD OCT (Bioptigen) Scott AW, et al.10 Shaken-baby syndrome 2 6 and 14 months SD OCT (Bioptigen) Gerth C, et al.11 Normal children 30 7 months–9 years SD OCT (Bioptigen) Chong GT, et al.12 Ocular albinism 12 5–13 years SD OCT (Bioptigen) Querques G, et al.13 Isolated foveal hypoplasia 1 10 years SD OCT (Cirrus)* El-Dairi MA et al.14 Normal children 286 3–17 years TD OCT (Stratus)* El-Dairi MA, et al.15 Optic nerve head, glaucoma 222 3–17 years TD OCT (Stratus) Shields CL, et al.3 Intraocular tumors 44 4–17 years TD OCT (Stratus) El-Dairi MA, et al.16 Pseudotumor cerebri 48 5–14 years TD OCT (Stratus) Dickmann A, et al.17 Unilateral amblyopia 40 5–56 years TD OCT (Stratus) Ecsedy M, et al.18 Macula in pre-term and full-term infants 35 7–14 years TD OCT (Stratus) Harvey PS, et al.19 Foveal development in albinism 11 7–36 years TD OCT (Stratus) Skarmoutsos F, et al.20 Cystoid macular edema in pediatric uveitis 1 9 years TD OCT (Stratus) Meyer C, et al.21 Oculocutaneous albinism 1 10 years TD OCT (Stratus)

* Carl Zeiss Meditec, Inc., Dublin, CA.

sizes in this manuscript appear as decimal portions of millimeters (e.g., and in the event of deviation of either parameter by greater than 20%, a 6.8 X 6.8-mm scan). imaging was stopped unless the parameter returned to normal range We reviewed the literature for normative data regarding pediatric and the nurse approved resuming the imaging session. One to 2 hours optics from 1960 to the present25–29 and documented optical changes after imaging, the nurse was queried about adverse events, to identify that occur in the infant eye as a function of age. From this analysis, we alterations to the subject’s health that might be related to the imaging identified the problems that potentially affect pediatric imaging and session. In the clinic, after obtaining study consent, SD OCT was designed reference tables and imaging protocols to image pediatric performed at the time of clinical examination, in a protocol that did not patients. require monitoring of vital signs and allowed up to 30 minutes for the We obtained SD OCT imaging in 21 premature infants from the total imaging session. neonatal intensive unit care (NICU) of Duke University Medical Center In both groups, oral 0.5 mL sucrose 24% with pacifier was used to and 21 outpatient infants from Duke Eye Center. We imaged in parallel decrease infant stress.30–32 The eyelids were usually held open with with the clinical examination scheduled as part of standard of care, and two fingers of the examiner, although a lid speculum was sometimes the subjects were imaged more than once, for a total of 62 subject used when needed for a stable examination. Artificial tears (Systane; imaging sessions (113 individual eye sessions) in the NICU and 39 (59 Alcon, Inc., Fort Worth, TX) were used to lubricate the cornea during individual eye sessions) outpatient sessions. Imaging was not at- the examination. To avoid resting the flexible cable on the infant, tempted if there was a dense vitreous hemorrhage. imaging was always performed with the base of the hand-held scanner NICU subjects were selected by the pediatric ophthalmologists and hand-held cord extending over the forehead. (SFF and DKW) during ROP screening visits and evaluated by the Ease of operation of SD OCT imaging was scored in three areas: neonatologist to assure that they were sufficiently stable to undergo ability to (1) obtain an image within 15 seconds (2) orient with the imaging. Before imaging, consent was obtained from the parents and summed voxel projection (SVP): a two dimensional retinal image procedures were performed in accordance with the study approval created by axially summing the OCT volume (see Figs. 3A, 3C, 3E), and from the Duke University Medical Center Institutional Review Board. (3) image the region of interest (fovea). Time to scan was not measured The study conformed to the tenets of the Declaration of Helsinki. in less than 1-minute intervals as planned, because the start time was SD OCT imaging in the NICU was performed immediately after the recorded in whole minutes. Captured SD OCT images were viewed ROP eye examination and was restricted to 15 minutes. Heart rate, with the software on the SD OCT system (InVivoVue 1.2; Bioptigen, respiratory rate, and oxygen saturation were recorded every minute, Inc.). Image quality was considered good if the B-scan image was sharp enough to let the observer differentiate retinal layers and foveal con- tour (Fig. 2). A field of scan was considered adequate when the B-scan image occupied the full scan, and the SVP image captured part of the optic nerve and an arcade.

RESULTS Infant Eye Optical Analysis The infant eye is unique in many aspects that affect optical imaging: axial length (AXL), refractive error (RE), corneal curvature (CC), and astigmatism. The AXL increases rapidly in the neonatal period growing 0.16 mm per week,29 according to a linear model27 (Fig. 3A). This growth slows with age from approximately 1 mm/year during the first 2 years to 0.4 mm/ year from 2 to 5 years, and to 0.1 mm/year from 5 to 15 years. FIGURE 1. Portable hand-held SD OCT workstation. (A) Arrow: refer- After age 15, no significant further growth occurs27,29 (Fig. 3B). ence arm position adjuster with digital readout. (B) Hand-held probe 29 with focus adjustment system and diopter scale bar. Infant is imaged in Gordon and Donzis reported a mean RE of Ϫ1.00 Ϯ 0.9 the incubator with the non-contact probe held over the left eye. The (range, Ϫ3.00 to ϩ1.00) D at 30 to 35 weeks postmenstrual age lids are held open by the examiner’s fingertips. (PMA), whereas from 36 weeks until the age of 6 years, a

Downloaded from jov.arvojournals.org on 09/25/2021 2680 Maldonado et al. IOVS, May 2010, Vol. 51, No. 5

Pediatric SD OCT Image Acquisition When imaging infant eyes, we found that the operator could readily hold the eyelids open in the awake or sleeping infant and that a speculum was usually not necessary, in part because the OCT scanner does not shine a bright light into the eye, and the subject sees only a faint red line against a black back- ground. Adding artiﬁcial tears before imaging provided stable tear ﬁlm and clearer images. During imaging in the NICU, there was adequate clearance for the SD OCT handpiece within the incubator; therefore, infants were not removed for imaging. Similarly, imaging was possible in one eye each of three infants around continuous positive airway pressure (CPAP) mask systems without removal. We did not need to use oral sucrose in 18% of NICU subjects and 10% of clinic subjects who appeared to be sleeping during imaging. We stopped imaging in two patients (8%) due to an elevation of heart rate above our 20% limit. Follow-up was performed at 1 to 2 hours after imaging for all NICU subjects, and no adverse events were reported in any of the subjects.

Improving Image Quality

FIGURE 2. (A) SD OCT scans of fovea in a premature infant without To correct the hand-held probe optics for the RE of the subject proper focus. Arrow: foveal area, blurred with low saturation in this being imaged, we used the reference table (Table 3) for set- image. (B) Good retinal layer differentiation in the SD OCT scan after tings based on age, since RE data were not available for most fine focusing of the HH-SD OCT system. subjects.27,29 We developed a calibrated RE readout scale applied to the hand-held probe to allow the operator to correct the RE in single-diopter increments (range, Ϫ10 to ϩ12 D) 27 hyperopic state prevails (mean ϩ0.5 Ϯ 0.2 D). Cook et al. (Fig. 1B). This calculation was made with manufacturer-sup- similarly reported an RE of Ϫ2.00 D at 32 weeks and Ϫ1.23 at plied data concerning the objective lens motion required per 36 weeks, with a shift to hyperopia (ϩ0.74 to ϩ2.12) by 40 to 52 weeks. The newborn cornea is generally steeper than the adult cornea, with a mean central corneal power of between 48 and 58.5 D, decreasing to adult values by 3 months.29,33–37 Al- though the power and axis of infant astigmatism varies in pediatric studies, the newborn eye has greater astigmatism than does the adult eye, but the condition also decreases by 50% in approximately 6 months.34 With retinoscopy, Dobson et al.38,39 reported greater than1Dofagainst-the-rule astigmatism in 100% of infant eyes under age 6 months, whereas by topography Isenberg et al.34 found a mean astigmatism of 6.0 D (range, 0.2–16.4) that was with the rule in 80% of newborn infant eyes. These optical properties result in a model infant eye unique from that of the adult. The schematic Gullstrand model eye is used to condense these optical properties into a summary simplified lens estimate that is useful in optimizing systems for viewing or imaging the retina.40 In 1976, Lotmar25 proposed a theoretical model for the eye of newborn infants. We used optical formulas of Gullstrand40 and of Gross and West41 to develop a theoretical eye model for prematurely born neonates using data available in biometric studies of the infant eye (Table 2).27,29 Refractive indices of the media were assumed to be equal to those of the adult eye. Note that the focal length of the premature infant schematic eye is 10.35 mm compared with 11.80 mm for Lotmar’s model newborn infant eye,25 and 17 mm for the adult eye, which is often used as a reference for ocular instrument analysis.42–44 Based on the foregoing analysis, we implemented age-specific considerations in our SD OCT imaging protocol for young children. This included changing the reference arm position, focus, and scan settings based on age (Table 3). For example, in the 32-week PMA infant, each millimeter of presumed scan length would actually be 0.629 mm at the retina in this eye (62.9% of the adult eye). Therefore, 27 performing a 10 mm retinal scan (set for an adult eye) would FIGURE 3. AXL change with age. (A) From 32 to 52 weeks PMA. (B) result in a 6.3 mm retinal scan in this infant eye. From birth to adulthood.29

Downloaded from jov.arvojournals.org on 09/25/2021 IOVS, May 2010, Vol. 51, No. 5 Technical Improvements to Optimize HH-SD OCT for Neonates 2681

TABLE 2. Comparison of Schematic Eye for Adult, Newborn, and Premature Infants

Schematic Eye Gullstrand Benner-Rannets Lotmar Current Study Properties Adult40 Adult-Modern41 Newborn25 Premature

Total eye power, D 58.64 60 84.8 96.54 Eye axial length, mm 24 24.09 17.49 15.1 Anterior focal length, mm Ϫ17.055 Ϫ16.67 Ϫ11.8 Ϫ10.35 Posterior focal length, mm 22.78 22.27 15.74 13.81 Corneal power, D 43.05 D 43.08 48.9 D 55.25 Corneal curvature radius 7.7 7.8 7.26 6.1 Lens power, D 19.11 20.83 43.4 43.5 Refractive error, D 1 0 ϩ2.8 Ϫ1 Refractive index: Air 1 1 1 1 Cornea 1.376 1.377 1.377 1.377 Aqueous 1.336 1.336 1.336 1.336 Lens 1.386 1.422 1.43 1.43 Vitreous 1.336 1.336 1.334 1.334

diopter of RE correction and the pitch of the objective lens smaller infant eye, the OCT scanning pivot location is displaced screw mount. anteriorly relative to the pupil, and thus the peripheral portion We started imaging premature neonates with the focus set of the image is clipped due to image vignetting by the iris. A at Ϫ1.00 D and term infants at ϩ2.00 D and then adjusted the pronounced clipping effect loses peripheral information as focus to optimize the image clarity. Changes smaller than 2 D seen in Figures 4A and 4B. This may be corrected by shortening do not produce a discernable improvement in the quality of SD the OCT reference arm delay such that the pivot point is OCT B-scan image; thus, we generally adjusted the focus in 2-D positioned in the iris plane. The reference arm position is jumps. A properly focused image allows the observer to differ- corrected in the SD OCT system as follows: entiate retinal layers and small pathologic structures (Fig. 2B). In less than 10% of cases the RE obtained from Table 3 did ⌬ in Reference Arm Position in mm not correlate with the best focus used for imaging. As a result ϭ ⌬ in AXL of the Eye in mm ϫ n of these focus adjustments, 91% of the 113 sessions had adequate image quality. where, n is the index of refraction of the vitreous (1.334). The Adequate Field of Scan: Correcting OCT AXL of the pediatric subject may be calculated based on age (reported for our system in Table 3). Commercial OCT systems Image Vignetting have a reference arm position pre-established by the manufac- Imagers achieve two-dimensional scanning of the retina by turer for a standard adult eye, and this calculated change is pivoting the OCT beam in the plane of the patient’s iris. In the with respect to that preset value.

TABLE 3. Standard Reference Table for Axial Length, Refractive Error, Reference Arm Position, and A-scans per B-scan by Age, with an Example of a 10-mm Adult Scan

For a 10-mm (Ϸ35 degree) Adult Scan Setting*

Increase in Scan Scan Relative Reference Length Length Number of Increment Scan Axial Arm† on on A-scans A-scans per Length to Refractive SD Length SD (Readout Retina Retina per Each1mmof Adult Scan Group Age Error D (D) (mm) (mm) Units) (mm) (deg) B-scans‡ Scan Length Length (%)

30–35 wk Ϫ1 0.9 15.1 0.9 97 6.3 22 925 92 63 35–39 wk 0.3 1.6 16.1 0.6 86 6.7 23 986 99 67 39–41 wk 0.4 1.5 16.8 0.6 79 7.0 25 1029 103 70 0–1 mo 0.9 0.9 17.4 0.5 72 7.3 25 1066 107 73 1–2 mo 0.3 0.6 18.6 0.5 59 7.8 27 1139 114 78 2–6 mo 0.5 0.6 18.9 0.4 56 7.9 28 1158 116 79 6–12 mo 0.6 0.2 19.2 0.5 52 8.0 28 1176 118 80 12–18 mo 0.7 0.6 20.1 0.7 43 8.4 29 1231 123 84 18 mo–2 y 0.9 1.5 21.3 0.3 30 8.9 31 1305 130 89 2–3 y 1 1.1 21.8 0.1 24 9.1 32 1335 134 91 3–4 y 0.6 1.8 22.2 0.4 20 9.3 32 1360 136 93 4–5 y Ϫ0.8 0.9 22.3 0.2 19 9.3 33 1366 137 93 5–9 y Ϫ0.6 1 22.7 0.4 14 9.5 33 1390 139 95 10 y-adult Ϫ0.5 1.5 24 0.7 0 10.0 35 1470 147 100 Axial myopia 26 Ϫ22 10.8 38 1593

* Note that this would apply to an adult scan setting selected from any OCT system. † A higher number equates with a shorter reference arm length on the system we used. To convert, we used manufacturer data (Bioptigen, Inc., Research Triangle Park, NC) of Ϫ10.86 readout units/mm of change in reference arm length. (Note that on the Bioptigen unit used in this study, the 10-mm adult scan appears as a 16-mm scan setting due to a manufacturer labeling error, as explained in Methods.) ‡ Presuming one wants 6.8 ␮m of separation between A-scans. This would be varied for a different A-scan density.

Downloaded from jov.arvojournals.org on 09/25/2021 2682 Maldonado et al. IOVS, May 2010, Vol. 51, No. 5

lateral movement of the hand-held probe to identify whether the reference arm should be adjusted longer or shorter. When the reference arm is too short, the clipping shadow moves in the same direction as lateral probe movement; when it is too long, the shadow moves in a direction opposite to the probe movement. A proper reference arm position produces a wider ﬁeld of scanning that allows for better orientation on the retina (Figs. 4C, 4F). Clipping may still occur, even for a perfectly set reference arm position, in large lateral scans.

Correction for Different Lateral Magniﬁcation in the Pediatric Eye Variations in the optical system of the eye affect the magniﬁ- cation of retinal images.45 Sanchez-Cano et al.45 studied this effect for time-domain OCT images, reporting an inversely proportional relationship between AXL and retinal image size. In infants, because of the very short eye length, this relationship is even more exaggerated. Most OCT system software does not adjust for this. To calculate the correct scan length on the retina (SLOR) for the infant eye, we adapted the following formula:

SLOR ϭ ͑Infant Eye AXL/Standard Adult AXL͒ ϫ System Scan Length FIGURE 4. SD OCT images with (A, B) and without (C–F) vignetting from incorrect reference arm length. (A, B) A 6.8 ϫ 6.8-mm volumetric scan of a 38-week PMA infant obtained without reference arm correc- For example in Figures 5A and 5B, a 6.8 ϫ 6.8-mm volumetric tion, focus, or scan density correction. After reference arm correction, scan is projected onto the retina of an 8-month-old patient the 10.9 ϫ 10.9-mm scan of a 35-week (C, D) and of a 47-week (E, F) (AXL ϭ 19.2 mm). To correct the lateral dimension: PMA infant show a decrease in vignetting. The SVP demonstrates the loss of peripheral image in (A) compared with when the reference arm is correct (C, E). SLOR ϭ ͑19.2 mm/24 mm͒ ϫ 6.8 SLOR ϭ 5.4 mm Per our protocol, we set the calculated reference arm before imaging and check for any OCT clipping. If clipping is We need to increase the length of the scan to correct for the observed, we then test the scan response to slight manual smaller SLOR in the young infant’s eye. For example, to en-

FIGURE 5. Retinal field of view and magnification with SD OCT in an infant. (A–C) The right eye at 8 months; (C) the projection of the SD OCT imaged area on the retinal cam- era photograph. (D, G) Additional SD OCT imaging of the same eye at 12 months. Left: the SVP retinal image with a selected B-scan in the middle column (corresponding to the green line in the SVP): (A, B) At high magnification, 6.8 ϫ 6.8-mm scan projecting to 5.4 ϫ 5.4-mm scan length on retina (80% of the adult); (D, E) with moderate magnification and wider field of view, a 10.9 ϫ 10.9-mm scan projecting to 8.7 ϫ 8.7-mm scan length on retina; and (F, G) with an even wider field of view, 13.6 ϫ 13.6-mm volumetric scan projecting to 10.9 ϫ 10.9-mm scan length (38°) on the retina. With SD OCT imaging of a large area, the scan area is limited by the pupil and vignetting occurs as in (F).

Downloaded from jov.arvojournals.org on 09/25/2021 IOVS, May 2010, Vol. 51, No. 5 Technical Improvements to Optimize HH-SD OCT for Neonates 2683

compass both macula and optic nerve on SD OCT of this infant’s eye (Figs. 5D–G), the lateral scan length was increased based on the reference table (Table 3). Without correction, measurement of lesions based on presumed mm sweep of the OCT system would be incorrect. The SVP images in Figures 5A, 5D, and 5F show a difference in lateral magniﬁcation of the retinal structures on SD OCT depending on the scan length. This difference may not be appreciated in observing the B- scans alone. Although the American National Standards Institute (ANSI) standards for ophthalmic imaging42,44 allow scanning with denser A-scans per B-scan, we elected to perform pediatric imaging using the same scan density as in the adult. If one does not consider the optics of the infant eye, then scan density will be greater in the infant eye. From the conversion table (Table 3) we could easily preset pediatric scan settings as follows:

A-scans/B-scan ϭ ͑Pediatric AXL mm/24 mm͒ ϫ ͑Scan Length mm͒ ϫ Selected number of A-scans per mm

The 30° SD OCT scan spans 8.57 mm of retina in the adult (3.5° per mm) but only 5.39 mm of retina in the 32-week-old PMA infant (Table 3). The density of A-scans per mm of retina would therefore be higher in the infant retina if not decreased to 63% of the adult number.

Impact of Applying Pediatric-Specific Settings Using these steps to image pediatric eyes with SD OCT (Fig. 6), we scored the ease of operation in three areas: time to scan, orienting from the retinal SVP image, and ability to image the fovea for the 113 sessions. In the NICU group, we captured an SD OCT image in Ͻ1 minute in 31% of sessions, between 1 and 2 minutes in 26%, and Ͼ2 minutes in 43%. We captured an average of 6 Ϯ 3 scan sets per eye, including rectangular volumetric scans in different areas (100 scans per volume) and linear scans. The total NICU session was 15 minutes per subject; thus, the average time per eye was 8 Ϯ 4 minutes. Four subjects in the NICU group had nasal CPAP apparatus; in these cases, we imaged only one eye because of the difficulty in FIGURE 6. Suggested sequence of steps for pediatric imaging with alignment. In the clinic group, an average of 6 Ϯ 3 scan sets HH-SD OCT. were captured per eye and average imaging time for each eye was 7 Ϯ 5 minutes. Imaging sessions in some children ex- tended over 30 minutes. Nystagmus, poor fixation, and move- The fovea was captured 74% of the time in the 113 NICU ment accounted for prolonged imaging sessions, more com- sessions. Pronounced Bell’s response during sleep or with eye monly in the older children. closure and eye or patient movement were the main reasons These steps have cut the infant imaging time in half by for failure. The success rate was 87%, in the 59 clinic sessions, improving ease of operation of SD OCT imaging. Previously, with poor fixation, nystagmus, and patient motion as the major Ͼ3 minutes per scan was necessary, with poorer image quali- obstacles for obtaining a scan set centered on the fovea. ty,9 compared with under 1.5 minutes with the current protocol. It is important to note that the time reported herein is not scanning time over the retina, but total time spent on an DISCUSSION imaging session including time for saving scans, repositioning the subject, instilling artificial tears, and giving oral sucrose. Portable SD OCT imaging can now be used to examine infants The operator was generally disoriented when viewing the and young children, such as those with ROP and those requir- SVP retinal image, because this was either inverted or rotated ing examinations under anesthesia. We modified optical pa- and inverted (when scanning at 0° or 90°, respectively). We rameters to address age-specific optics of the infant eye and to partially overcame this by creating a reference card and placing improve SD OCT imaging of very young infants. Retinal SD labels over the screen to mark the SVP image. To improve OCT imaging was tolerated by NICU infants as performed in orientation, we also used the Bioptigen system Free Run mode this study without adverse events. that creats the SVP image in a continuous real-time scanning The SD OCT system used in this study, according to the mode. In squirming infants, one could save a useful scan after manufacturer, meets the ANSI42–44standards for ocular expo- it appeared on the screen. Note that, peripheral imaging with sure to light (limited to Ͻ700 ␮W of continuous-wave power our current techniques is limited (Fig. 5). The field of imaging in the 800–900-nm spectral region within a 7-mm limiting may extend beyond the posterior pole, depending on eye aperture) allowable for a patient exposed continuously for up position; however, unlike fundus photographs, it does not to 8 hours. Imaging sessions, in our experience, represented a provide a peripheral view. total beam illumination time in the eye of less than 5 minutes.

Downloaded from jov.arvojournals.org on 09/25/2021 2684 Maldonado et al. IOVS, May 2010, Vol. 51, No. 5

Calculations for ophthalmic beam exposure are based on 13. Querques G, Bux AV, Iaculli C, Noci ND. Isolated foveal hypopla- 17-mm secondary focal length of the 24-mm average AXL adult sia. Retina. 2008;28:1552–1553. human eye and do not consider the pediatric eye AXL, even 14. El-Dairi MA, Asrani SG, Enyedi LB, Freedman SF. Optical coherence though light exposure safety studies on which ANSI standards tomography in the eyes of normal children. Arch Ophthalmol. are based, were performed in Macaca eyes (AXL, 13.1–19.5 2009;127:50–58. mm), which is comparable to very young children.27,29 Pub- 15. El-Dairi MA, Holgado S, Asrani S, Enyedi L, Freedman S. Correlation lished ocular instrument recommendations43 of a task group of between Optical Coherence Tomography and glaucomatous optic nerve head damage in children. Br J Ophthalmol. 2009;93(10): the International Commission on Non-ionizing Radiation Pro- 1325–1330. tection44 and an ANSI summary with emphasis on ophthalmic 13 16. El-Dairi MA, Holgado S, O’Donnell T, Buckley EG, Asrani S, Freed- devices all reference the adult focal length for calculations of man SF. Optical coherence tomography as a tool for monitoring exposure. The anterior focal length of the newborn infant eye pediatric pseudotumor cerebri. J AAPOS. 2007;11:564–570. 25 was calculated to be 11.8 mm, and we calculated it to be 17. Dickmann A, Petroni S, Salerni A, Dell’Omo R, Balestrazzi E. Uni- 10.35 mm in the premature eye. The optics of the pediatric eye lateral amblyopia: an optical coherence tomography study. J AA- will increase the density of SD OCT A-scans beyond that ex- POS. 2009;13:148–150. pected from the standard adult settings. To maintain the same 18. Ecsedy M, Szamosi A, Karko C, et al. A comparison of macular density of scanning for the pediatric eyes as in adult eyes, we structure imaged by optical coherence tomography in preterm and recommend decreasing the number of A-scans (focal sites im- full-term children. Invest Ophthalmol Vis Sci. 2007;48:5207–5211. aged) per B-scan, based on AXL. 19. Harvey PS, King RA, Summers CG. Spectrum of foveal develop- In pediatric patients, access to the eye may be limited by ment in albinism detected with optical coherence tomography. J infant inattention and movement. Thus, it is critical to optimize AAPOS. 2006;10:237–242. scan quality in a short capture time. We have demonstrated the 20. Skarmoutsos F, Sandhu SS, Voros GM, Shafig A. The use of optical coherence tomography in the management of cystoid macular importance of customizing SD OCT imaging settings for the edema in pediatric uveitis. J AAPOS. 2006;10:173–174. pediatric population. Imagers should consider age and optics 21. Meyer CH, Lapolice DJ, Freedman SF. Foveal hypoplasia in oculo- of the eye when setting imaging parameters for infants and cutaneous albinism demonstrated by optical coherence tomogra- measuring ocular structures. By detecting critical early retinal phy. Am J Ophthalmol. 2002;133:409–410. changes, a customized approach to SD OCT imaging in chil- 22. Schuman SG, Koreishi AF, Farsiu S, Jung SH, Izatt JA, Toth CA. dren may be a valuable adjunct to clinical examination in Photoreceptor layer thinning over drusen in eyes with age- predicting infant eyes at high risk of vision loss from ROP and related macular degeneration imaged in vivo with spectral- other pediatric retinal disorders. This imaging in neonates, domain optical coherence tomography. Ophthalmology. 2009; infants, and young children could lead to a better understand- 116(3):488–496. ing of infant retinal development and pediatric retinal diseases 23. Forooghian F, Cukras C, Meyerle CB, Chew EY, Wong WT. Evalu- and to improved therapeutic decisions in the future. ation of time domain and spectral domain optical coherence tomography in the measurement of diabetic macular edema. Invest References Ophthalmol Vis Sci. 2008;49(10):4290–4296. 24. Han IC, Jaffe GJ. Comparison of spectral- and time-domain optical 1. Mirza RG, Johnson MW, Jampol LM. Optical coherence tomogra- coherence tomography for retinal thickness measurements in phy use in evaluation of the vitreoretinal interface: a review. Surv healthy and diseased eyes. Am J Ophthalmol. 2009;147(5):847– Ophthalmol. 2007;52:397–421. 858. 2. Stopa M, Bower BA, Davies E, Izatt JA, Toth CA. Correlation of 25. Lotmar W. A theoretical model for the eye of new-born infants. pathologic features in Spectral Domain Optical Coherence Tomog- Graefes Arch Clin Exp Ophthalmol. 1976;198:179–185. raphy with conventional retinal studies. Retina. 2008;28:298–308. 26. Mactier H, Maroo S, Bradnam M, Hamilton R. Ocular biometry in 3. Shields CL, Mashayekhi A, Luo CK, Materin MA, Shields JA. Optical preterm infants: implications for estimation of retinal illuminance. coherence tomography in children: analysis of 44 eyes with in- Invest Ophthalmol Vis Sci. 2008;49(1):453–457. traocular tumors and simulating conditions. J Pediatr Ophthalmol 27. Cook A, White S, Batterbury M, Clark D. Ocular growth and Strabismus. 2004;41:338–344. refractive error development in premature infants without retinop- 4. World Health Organization. Magnitude and causes of visual impair- athy of prematurity. Invest Ophthalmol Vis Sci. 2008;49(12): ment. Geneva: WHO; 2004:282. 5199–5207. 5. Printing House for the Blind. Facts and Figures on Americans with 28. Quinn GE, Francis EL, Nipper KS, et al. Highly precise eye length Vision Loss. New York: American Foundation for the Blind; 2007. measurements in children aged 3 through 12 years. Arch Ophthal- 6. Foster A. Childhood blindness. Eye. 1988;2(suppl):13–18. mol. 2003;121(7):985–990. 7. Lu Q, Zheng Y, Sun B, et al. A population-based study of visual 29. Gordon RA, Donzis PB. Refractive development of the human eye. impairment among pre-school children in Beijing: the Beijing study Arch Ophthalmol. 1985;103(6):785–789. of visual impairment in children. Am J Ophthalmol. 2009;147: 30. Stevens B, Yamada J, Ohlsson A. Sucrose for analgesia in newborn 1075–1081. infants undergoing painful procedures. Cochrane Database Syst 8. Foster A, Gilbert C. Epidemiology of visual impairment in child- Rev. 2004;3. hood. In: Taylor D, ed. Paediatric Ophthalmology. 2nd ed. Ox- 31. Elserafy FA, Alsaedi SA, Louwrens J, Bin Sadiq B, Mersal AY. Oral ford, UK: Blackwell; 1996. sucrose and a pacifier for pain relief during simple procedures in 9. Chavala SH, Farsiu S, Maldonado R, Wallace DK, Freedman SF, preterm infants: a randomized controlled trial. Ann Saudi Med. Toth CA. Insights into advanced retinopathy of prematurity utiliz- 2009;29(3):184–188. ing handheld spectral domain optical coherence tomography im- 32. Curtis SJ, Jou H, Ali S, Vandermeer B, Klassen T. A randomized aging. Ophthalmology. 2009;116(12):2448–2456. controlled trial of sucrose and/or pacifier as analgesia for infants 10. Scott AW, Farsiu S, Enyedi LB, Wallace DK, Toth CA. Imaging the receiving venipuncture in a pediatric emergency department. infant retina with a hand-held spectral-domain optical coherence BMC Pediatr. 2007:7–27. tomography device. Am J Ophthalmol. 2009;147(2):364–373. 33. Friling R, Weinberger D, Kremer I, Avisar R, Sirota L, Snir M. 11. Gerth C, Zawadzki RJ, Heon E, Werner JS. High-resolution Keratometry measurements in preterm and full term newborn retinal imaging in young children using a handheld scanner and infants. Br J Ophthalmol. 2004;88:8–10. Fourier-domain optical coherence tomography. J AAPOS. 2007; 34. Isenberg S, Del Signore M, Chen A, Wei J, Christenson P. Corneal 13:72–74. topography of neonates and infants. Arch Ophthalmol. 2004;122: 12. Chong GT, Farsiu S, Freedman SF, et al. Abnormal foveal morphol- 1767–1771. ogy in ocular albinism imaged with spectral-domain optical coher- 35. Inagaki Y. The rapid change of corneal curvature in the neonatal ence tomography. Arch Ophthalmol. 2009;127(1):37–44. period and infancy. Arch Ophthalmol. 1986;104:1026–1027.

Downloaded from jov.arvojournals.org on 09/25/2021 IOVS, May 2010, Vol. 51, No. 5 Technical Improvements to Optimize HH-SD OCT for Neonates 2685

36. Mohindra I, Held R, Gwiazda J, Brill S. Astigmatism in infants. 42. Delori FC, Webb RH, Sliney DH. Maximum permissible expo- Science. 1978;202:329. sures for ocular safety (ANSI 2000), with emphasis on ophthal- 37. Howland HC, Atkinson J, Braddock O, French J. Infant astigmatism mic devices. J Opt Soc Am A Opt Image Sci Vis. 2007;24:1250– measured by photorefraction. Science. 1978;202:331–333. 1265. 38. Dobson V, Fulton A, Sebris SL. Cycloplegic refractions of infants 43. Sliney D, Aron-Rosa D, DeLori F, et al. Adjustment of guidelines for and young children: the axis of astigmatism. Invest Ophthalmol exposure of the eye to optical radiation from ocular instruments: Vis Sci. 1984;25:83–87. statement from a task group of the International Commission on 39. Dobson V, Fulton AB, Manning K. Cycloplegic refractions of pre- Non-Ionizing Radiation Protection (ICNIRP). Appl Opt. 2005;44: mature infants. Am J Ophthalmol. 1981;91:490–495. 2162–2176 . 40. Gullstrand A. Appendices II and IV. In: Southall JPC, ed. Helm- 44. American National Standards Institute. American National Stan- holtz’s Treatise on Physiological Optics. Menasha, WI: Optical dard for Safe of Use of Laser. New York: ANSI; 2007. Society of America; 1909;1:301–358, 382–415. 45. Sanchez-Cano A, Baraibar B, Pablo LE, Honrubia FM. Magniﬁcation 41. Gross DA, West RW. Introduction to the Optics of the Eye. New characteristics of the optical coherence tomograph STRATUS OCT York: Elsevier Science; 2001. 3000. Ophthalmic Physiol Opt. 2008;28:21–28.

Downloaded from jov.arvojournals.org on 09/25/2021