Study subjects and baseline examinations
This is a single-center, prospective, interventional study conducted at Yeungnam University Medical Center between March 2018 and April 2018. The study protocol was approved by the Institutional Review Board of Yeungnam University Medical Center (IRB No. 2017-04-026). All participants provided signed informed consent, and the study adhered to the tenets of the Declaration of Helsinki.
The participants included healthy young subjects aged 18 years or older. Normal subjects were defined as those with no history of systemic or ocular disease or no presence of ocular surgery. The exclusion criteria were as follows: myopia < - 6 D, astigmatism > 1.5 D, intraocular pressure (IOP) > 21 mmHg, media opacity that obscured image acquisition, pathologic myopic changes of the retina such as posterior staphyloma, Lacquer crack, tessellated fundus, or myopic foveoschisis, or changes of the optic disc such as tilted disc configuration, myopic parapillary atrophy or glaucomatous changes. Media opacity was defined to be the presence of corneal scarring, corneal edema, cataract, or vitreous haze. Each participant underwent a routine ophthalmic examination, including evaluation of past medical history, best-corrected visual acuity, refractive error without pupil dilation (Auto-refracto-keratometer, HRK-7000A; Huvitz Co., Ltd., Korea), IOP (TX-20 Full Auto Tonometers; Canon, Tokyo, Japan), and axial length (IOL Master500; Carl Zeiss Meditec, Gena, Germany).
Considering the fact that axial length and degree of baseline refractive error are capable of influencing the measurement of area in UWF images, we classified the subjects into three groups based on their corresponding axial lengths: below-average axial length group (Group A, between 22 and 24 mm), above-average axial length group (Group B, between 24 and 26 mm), and long axial length group (Group C, 26 mm or above). Further, we classified the subjects into two groups based on refractive error: mild myopia (Plano to - 3 D) and moderate myopia (- 3 D to - 6 D).
Image acquisition, projection, and quantification
UWF imaging was performed using an Optos UWF system (Optos California; Dunfermline, Scotland, UK). UWF images were acquired first without the use of soft CL, and then the process was repeated using CLs of six different diopters (+ 9 D, + 6 D, + 3 D, - 3 D, - 6 D, and - 9 D). Acquired images were transformed to a stereographic projection image using proprietary software from the manufacturer [1]. With stereographically projected UWF images, two masked, trained ophthalmologists manually outlined the optic disc area by 15–17 points where the disc margin meets the blood vessel using Image J V.1.49b (US National Institutes of Health, Bethesda, Maryland).
The area of the optic disc corresponding to each image was measured in square millimeters (mm2) by summing the anatomically-correct sizes of all pixels that comprise the disc margin (Figure 1). Two independent, masked graders performed annotations of the optic disc twice, and the average value was used for subsequent statistical analyses. These values obtained using CLs were compared with the measurements obtained from the baseline image without CL (100.0%), and the respective percentage differences were determined. The maximal difference (%) in each case was defined as absolute value differences (% area difference) compared to those for the baseline image without CL. The magnification group comprised subjects exhibiting maximal difference > 10%.
Statistical analysis
Statistical analyses were conducted using SPSS software (version 19.0; IBM Corp., Armonk, NY) and MedCalc (version 15.8; MedCalc, Inc., Ostend, Belgium). The Kolmogorov-Smirnov test was used to assess sample distribution. The differences in numerical data were analyzed using repeated measure analysis of variance (ANOVA), Kruskal Wallis test, independent t-test, and Mann-Whitney test. Categorical variables were evaluated using Chi-square tests. Area under the curve, sensitivity, and specificity were calculated using the receiver operating characteristics (ROC) curve. A multiple comparison with Bonferroni correction was performed in cases that exhibited significant difference. Intra-grader and inter-grader agreement values were evaluated by intra-class correlation coefficient (ICC) values. Statistical significance was defined as P < 0.05.