Automated Explainable Multidimensional Deep Learning Platform of Retinal Images for Retinopathy of Prematurity Screening.

Importance: A retinopathy of prematurity (ROP) diagnosis currently relies on indirect ophthalmoscopy assessed by experienced ophthalmologists. A deep learning algorithm based on retinal images may facilitate early detection and timely treatment of ROP to improve visual outcomes. Objective: To develop a retinal image-based, multidimensional, automated, deep learning platform for ROP screening and validate its performance accuracy. Design, Setting, and Participants: A total of 14 108 eyes of 8652 preterm infants who received ROP screening from 4 centers from November 4, 2010, to November 14, 2019, were included, and a total of 52 249 retinal images were randomly split into training, validation, and test sets. Four main dimensional independent classifiers were developed, including image quality, any stage of ROP, intraocular hemorrhage, and preplus/plus disease. Referral-warranted ROP was automatically generated by integrating the results of 4 classifiers at the image, eye, and patient levels. DeepSHAP, a method based on DeepLIFT and Shapley values (solution concepts in cooperative game theory), was adopted as the heat map technology to explain the predictions. The performance of the platform was further validated as compared with that of the experienced ROP experts. Data were analyzed from February 12, 2020, to June 24, 2020. Exposure: A deep learning algorithm. Main Outcomes and Measures: The performance of each classifier included true negative, false positive, false negative, true positive, F1 score, sensitivity, specificity, receiver operating characteristic, area under curve (AUC), and Cohen unweighted κ. Results: A total of 14 108 eyes of 8652 preterm infants (mean [SD] gestational age, 32.9 [3.1] weeks; 4818 boys [60.4%] of 7973 with known sex) received ROP screening. The performance of all classifiers achieved an F1 score of 0.718 to 0.981, a sensitivity of 0.918 to 0.982, a specificity of 0.949 to 0.992, and an AUC of 0.983 to 0.998, whereas that of the referral system achieved an F1 score of 0.898 to 0.956, a sensitivity of 0.981 to 0.986, a specificity of 0.939 to 0.974, and an AUC of 0.9901 to 0.9956. Fine-grained and class-discriminative heat maps were generated by DeepSHAP in real time. The platform achieved a Cohen unweighted κ of 0.86 to 0.98 compared with a Cohen κ of 0.93 to 0.98 by the ROP experts. Conclusions and Relevance: In this diagnostic study, an automated ROP screening platform was able to identify and classify multidimensional pathologic lesions in the retinal images. This platform may be able to assist routine ROP screening in general and children hospitals.

Inter-eye Differences in Ocular Biometric Parameters of Concomitant Exotropia.

Purpose: To evaluate the ocular biometric parameters in patients with constant and intermittent exotropia by the measurement of swept-source optical coherence tomography (SS-OCT) optical biometer OA-2000 and comparing it with the normal control subjects. Design: Cross-sectional case-control study. Participants: Fifty-five constant and 24 intermittent patients with exotropia with central fixation and 77 orthotropic normal control participants aged 4-18 years old. Methods: Non-contact and high-resolution optical biometric OA-2000 measurements were conducted under uniform ambient light conditions. The statistical analysis included intraclass correlation coefficient (ICC), Bland-Altman plot, and independent t-tests. Main Outcome Measures: Spherical equivalent (SE), ocular biological parameters such as pupil diameter (PD), anterior chamber depth (ACD), lens thickness (LT), and axial lengths (AL). The absolute values of inter-eye differences for SE, PD, ACD, LT, and AL were recorded as AnisoSE, AnisoPD, AnisoACD, AnisoLT, and AnisoAL, respectively. Results: AnisoSE (0.878 vs. 0.577, P = 0.019), AnisoAL (0.395 vs. 0.208, P = 0.001), AnisoACD (0.060 vs. 0.032, P < 0.001), AnisoLT (0.060 vs. 0.031, P = 0.002), and AnisoPD (0.557 vs. 0.340, P = 0.002) were significantly larger in concomitant patients with exotropia. The SE, AL, ACD, LT, and PD showed excellent binocular correlation with ICC values that ranged from 0.943 to 0.987 in control participants and from 0.767 to 0.955 in concomitant exotropia patients. Bland-Altman plots showed the wider range of agreement in patients with concomitant exotropia than the control participants (SE: 5.0288 vs. 3.3258; AL: 2.2146 vs. 1.3172; ACD:0.3243 vs. 0.1682; PD: 2.4691 vs. 1.9241; and LT:0.3723 vs 0.1858). Conclusion: Patients with concomitant exotropia showed larger inter-eye differences in SE, ACD, LT, PD, and AL. Advice should be given to suspicious children to avoid or delay the development of concomitant exotropia.

Corneal biomechanics, refractive error, and axial length in Chinese primary school children.

PURPOSE: Low corneal hysteresis is associated with longer axial length in Chinese secondary school children. The authors sought to explore this association in primary school children. METHODS: LogMAR presenting visual acuity, cycloplegic refractive error, ocular biometry, central corneal thickness (CCT), and corneal hysteresis (CH) was assessed for children in grades 1 to 3 at an academically competitive urban school in Shantou, China. RESULTS: Among 872 eligible children (mean age, 8.6 ± 2.1 years), 651 (74.7%) completed the examination. Among 1299 examined eyes, 111 (8.5%) had uncorrected vision ≤6/12. Mean spherical equivalent refractive error for all eyes was +0.26 ± 1.41 D, and axial length (AL) was 22.7 ± 0.90 mm. CH for the lowest (mean AL, 21.7 ± 0.39 mm), two middle (mean AL, 22.4 ± 0.15 and 22.9 ± 0.15 mm), and highest quartiles (mean AL, 23.7 ± 0.74 mm) of AL were 10.6 ± 2.1 mm Hg, 10.4 ± 2.1 mm Hg, 10.3 ± 2.3 mm Hg, and 10.2 ± 2.3 mm Hg respectively (age- and gender-adjusted Pearson's correlation coefficient r = -0.052; P = 0.001). In generalized estimating equation models adjusting for age, gender, and CCT, lower CH was significantly associated with longer AL (P < 0.001) and more myopic refractive error (P = 0.001). CONCLUSIONS: CH measurement is practical in young children because this is when myopia undergoes its most rapid progression. Prospective follow-up of this cohort at high risk for myopia is under way to determine whether low CH is predictive, or a consequence, of long AL.