Background: Accurately predicting refractive error in children is crucial for detecting amblyopia, which can lead to permanent visual impairment, but is potentially curable if detected early. Various tools have been adopted to more easily screen a large number of patients for amblyopia risk. Objective: For efficient screening, easy access to screening tools and an accurate prediction algorithm are the most important factors. In this study, we developed an automated deep learning–based system to predict the range of refractive error in children (mean age 4.32 years, SD 1.87 years) using 305 eccentric photorefraction images captured with a smartphone. Methods: Photorefraction images were divided into seven classes according to their spherical values as measured by cycloplegic refraction. Results: The trained deep learning model had an overall accuracy of 81.6%, with the following accuracies for each refractive error class: 80.0% for ≤−5.0 diopters (D), 77.8% for >−5.0 D and ≤−3.0 D, 82.0% for >−3.0 D and ≤−0.5 D, 83.3% for >−0.5 D and <+0.5 D, 82.8% for ≥+0.5 D and <+3.0 D, 79.3% for ≥+3.0 D and <+5.0 D, and 75.0% for ≥+5.0 D. These results indicate that our deep learning–based system performed sufficiently accurately. Conclusions: This study demonstrated the potential of precise smartphone-based prediction systems for refractive error using deep learning and further yielded a robust collection of pediatric photorefraction images.

We investigated the characteristics of children with peripapillary hyperreflective ovoid mass-like structures (PHOMS) and evaluated the associated risk factors. This cross-sectional study included 132 eyes of 66 children with PHOMS and 92 eyes of 46 children without PHOMS (controls) who were assessed by disc enhanced-depth image spectral-domain (SD) optical coherence tomography (OCT). Univariable and multivariable logistic analyses were performed to evaluate risk factors associated with presence of PHOMS. Among the 66 children with PHOMS, 53 (80.3%) had bilateral and 13 (19.7%) had unilateral PHOMS. The mean age of the PHOMS group was 11.7 ± 2.6 years, and that of the control group was 11.4 ± 3.1 years. The mean spherical equivalent (SE) as determined by cycloplegic refraction was −3.13 ± 1.87 diopters (D) in the PHOMS group and −0.95 ± 2.65 D in the control group. Additionally, mean astigmatism was 0.67 ± 0.89 D and 0.88 ± 1.02 D in the PHOMS group and the control group, respectively. Mean disc size was 1,735 ± 153 µm in the PHOMS group and 1,741 ± 190 µm in the control group, while mean optic nerve head (ONH) tilt angle was 9.84 ± 5.38 degrees in the PHOMS group and 3.71 ± 4.41 degrees in the control group. SE and ONH tilt angle were significantly associated with PHOMS according to both univariable [odds ratio (OR): 1.59; p < 0.001 and OR: 1.35; p < 0.001, respectively] and multivariable (OR: 1.71; p = 0.001 and OR: 1.29; p = 0.001, respectively) logistic regression analyses. There was a significant correlation between SE and ONH tilt (r = −0.46; p < 0.001). In conclusion, PHOMS is associated with myopic shift in children, and optic disc tilt may be a mediator between myopia and PHOMS.

Background: To evaluate the potential of vessel density alterations for predicting postoperative visual field (VF) improvement in chiasmal compression using optical coherence tomography angiography (OCT-A). Methods: The study investigated 57 eyes of 57 patients diagnosed with pituitary tumors and 42 eyes of 42 age and refractive error matched controls. All eyes with chiasmal compression for which preoperative optical coherence tomography (OCT) and OCT-A, and pre- and postoperative VF data were available. Preoperative vessel densities of superficial retinal capillary plexus (SRCP), deep retinal capillary plexus (DRCP), and radial peripapillary capillary (RPC) segment were utilized by OCT-A. Results: Preoperative peripapillary retinal nerve fiber layer and ganglion cell layer complex thickness and vessel densities of SRCP and RPC segments in eyes with chiasmal compression were significantly reduced compared with healthy controls (p < 0.001, p < 0.001, p = 0.007, and p = 0.020, respectively). In multivariate regression analysis, preoperative perimetric mean deviation (MD) (p = 0.002) and vessel density of SRCP (p = 0.025) were correlated significantly with postoperative perimetric MD. Spearman’s correlation analysis revealed significant correlations between preoperative MD on perimetry (r = 0.443, p = 0.001), vessel densities of SRCP (r = 0.288, p = 0.035) and RPC segment (r = 0.347, p = 0.009), and postoperative perimetric MD. Conclusions: Structural degeneration referred to as microvascular alterations measured by OCT-A and preoperative VF defects were associated with worse postoperative VF prognosis. Parafoveal and peripapillary vessel densities may serve as a sensitive, structural prognostic factors in the preoperative judgement of chiasmal compression.

Background: It is necessary to consider myopic optic disc tilt as it seriously impacts normal ocular parameters. However, ophthalmologic measurements are within inter-observer variability and time-consuming to get. This study aimed to develop and evaluate deep learning models that automatically recognize a myopic tilted optic disc in fundus photography. Methods: This study used 937 fundus photographs of patients with normal or myopic tilted disc, collected from Samsung Medical Center between April 2016 and December 2018. We developed an automated computer-aided recognition system for optic disc tilt on color fundus photographs via a deep learning algorithm. We preprocessed all images with two image resizing techniques. GoogleNet Inception-v3 architecture was implemented. The performances of the models were compared with the human examiner’s results. Activation map visualization was qualitatively analyzed using the generalized visualization technique based on gradient-weighted class activation mapping (Grad-CAM++). Results: Nine hundred thirty-seven fundus images were collected and annotated from 509 subjects. In total, 397 images from eyes with tilted optic discs and 540 images from eyes with non-tilted optic discs were analyzed. We included both eye data of most included patients and analyzed them separately in this study. For comparison, we conducted training using two aspect ratios: the simple resized dataset and the original aspect ratio (AR) preserving dataset, and the impacts of the augmentations for both datasets were evaluated. The constructed deep learning models for myopic optic disc tilt achieved the best results when simple image-resizing and augmentation were used. The results were associated with an area under the receiver operating characteristic curve (AUC) of 0.978 ± 0.008, an accuracy of 0.960 ± 0.010, sensitivity of 0.937 ± 0.023, and specificity of 0.963 ± 0.015. The heatmaps revealed that the model could effectively identify the locations of the optic discs, the superior retinal vascular arcades, and the retinal maculae. Conclusions: We developed an automated deep learning-based system to detect optic disc tilt. The model demonstrated excellent agreement with the previous clinical criteria, and the results are promising for developing future programs to adjust and identify the effect of optic disc tilt on ophthalmic measurements.

Kim H, et al. Korean J Ophthalmol. 2018 Dec;33:e10. https://doi.org/10.3341/kjo.2018.0050