Development of a Machine-Learning-Based Tool for Overnight Orthokeratology Lens Fitting.

Purpose: Orthokeratology (ortho-K) is widely used to control myopia. Overnight ortho-K lens fitting with the selection of appropriate parameters is an important technique for achieving successful reductions in myopic refractive error. In this study, we developed a machine-learning model that could select ortho-K lens parameters at an expert level. Methods: Machine-learning models were established to predict the optimal ortho-K parameters, including toric lens option (toric or non-toric), overall diameter (OAD; 10.5 or 11.0 mm), base curve (BC), return zone depth (RZD), landing zone angle (LZA), and lens sagittal depth (LensSag). The analysis included 547 eyes of 297 Korean adolescents with myopia or astigmatism. The dataset was randomly divided into training (80%, n = 437 eyes) and validation (20%, n = 110 eyes) sets at the patient level. The model was trained based on clinical ortho-K lens fitting performed by highly experienced experts and ophthalmic measurements. Results: The final machine-learning models showed accuracies of 92.7% and 86.4% for predicting the toric lens option and OAD, respectively. The mean absolute errors for the BC, RZD, LZA, and LensSag predictions were 0.052 mm, 2.727 µm, 0.118°, and 5.215 µm, respectively. The machine-learning model outperformed the manufacturer's conventional initial lens selector in predicting BC and RZD. Conclusions: We developed an expert-level machine-learning-based model for determining comprehensive ortho-K lens parameters. We also created a web-based application. Translational Relevance: This model may provide more accurate fitting parameters for lenses than those of conventional calculations, thus reducing the need to rely on trial and error.

Association between metformin use and the risk of developing open-angle glaucoma among patients with diabetes: a retrospective cohort study and meta-analysis.

PURPOSE: Recent studies examining the neuroprotective effects of metformin on open-angle glaucoma (OAG) have failed to provide consistent results. In this study, we investigated the association between metformin use and OAG. METHODS: Data were obtained from a sample cohort of the Korean National Health Insurance database. Patients diagnosed with type-2 diabetes (T2DM) between 2004 and 2013 were included. We performed propensity score-matched analysis in a matched cohort (N = 20,646). The risk of the newly developed OAG was estimated using a Cox proportional hazards model. Including the present study, the meta-analysis included five studies to calculate the pooled risk for OAG based on metformin use. RESULTS: In the adjusted model, the analysis revealed no statistical association between metformin use and OAG incidence (hazard ratio [HR] 1.05; 95% confidence interval [CI] 0.79-1.40; P = 0.738). The highest tercile of metformin use demonstrated no statistical significance (HR 0.93 [95% CI 0.63-1.37]; P = 0.703). No significant dose-dependent association was observed between the cumulative dose and incidence of OAG (P-value for trend = 0.336). In a meta-analysis of four published articles and the present study, the common-effects and random-effects models indicated conflicting results in terms of significance. The random effects model demonstrated no significant association (pooled risk ratio 0.53; 95% CI 0.24-1.19; P = 0.123). CONCLUSION: We found no significant association between metformin use and OAG incidence in patients with T2DM in this population-based cohort study and meta-analysis. Further studies are needed to investigate the association between metformin use and the risk of OAG among patients with T2DM.

Association between cataract and fatty liver diseases from a nationwide cross-sectional study in South Korea.

This study examined the link between fatty liver disease (FLD) and cataracts, as previous research has suggested that FLD may contribute to metabolic syndrome, systemic inflammation, and potentially cataracts. We studied a nationwide cross-sectional cohort of the Fifth Korean National Health and Nutrition Examination Survey 2010-2011. FLD was defined as nonalcoholic FLD (NAFLD) and metabolic dysfunction-associated FLD (MAFLD). Multinomial logistic regression was utilized to investigate the relationship between cataracts and FLD after adjustment for potential confounders. Participants with cataracts had higher liver fibrosis scores, including the NAFLD fibrosis score (NFS; P < 0.001), fibrosis-4 index (FIB4; P < 0.001), and fatty liver index (FLI; P = 0.001). NAFLD was not associated with a higher odds ratio (OR) for cataracts in the fully adjusted model (OR = 1.23, P = 0.058). MAFLD was significantly associated with a higher OR (OR = 1.34, P = 0.006). After adjusting for all factors, the severity of FLD was linked to an increased risk of cataracts, with significant linear trends (P values for linear trends of NFS, FIB4, and FLI < 0.05). After adjusting for well-known cataract risk factors, MAFLD was significantly associated with cataracts. Our analysis suggests that FLD may serve as an independent risk factor for cataracts.

Development of a generative deep learning model to improve epiretinal membrane detection in fundus photography.

BACKGROUND: The epiretinal membrane (ERM) is a common retinal disorder characterized by abnormal fibrocellular tissue at the vitreomacular interface. Most patients with ERM are asymptomatic at early stages. Therefore, screening for ERM will become increasingly important. Despite the high prevalence of ERM, few deep learning studies have investigated ERM detection in the color fundus photography (CFP) domain. In this study, we built a generative model to enhance ERM detection performance in the CFP. METHODS: This deep learning study retrospectively collected 302 ERM and 1,250 healthy CFP data points from a healthcare center. The generative model using StyleGAN2 was trained using single-center data. EfficientNetB0 with StyleGAN2-based augmentation was validated using independent internal single-center data and external datasets. We randomly assigned healthcare center data to the development (80%) and internal validation (20%) datasets. Data from two publicly accessible sources were used as external validation datasets. RESULTS: StyleGAN2 facilitated realistic CFP synthesis with the characteristic cellophane reflex features of the ERM. The proposed method with StyleGAN2-based augmentation outperformed the typical transfer learning without a generative adversarial network. The proposed model achieved an area under the receiver operating characteristic (AUC) curve of 0.926 for internal validation. AUCs of 0.951 and 0.914 were obtained for the two external validation datasets. Compared with the deep learning model without augmentation, StyleGAN2-based augmentation improved the detection performance and contributed to the focus on the location of the ERM. CONCLUSIONS: We proposed an ERM detection model by synthesizing realistic CFP images with the pathological features of ERM through generative deep learning. We believe that our deep learning framework will help achieve a more accurate detection of ERM in a limited data setting.

Deep learning prediction of steep and flat corneal curvature using fundus photography in post-COVID telemedicine era.

Recently, fundus photography (FP) is being increasingly used. Corneal curvature is an essential factor in refractive errors and is associated with several pathological corneal conditions. As FP-based examination systems have already been widely distributed, it would be helpful for telemedicine to extract information such as corneal curvature using FP. This study aims to develop a deep learning model based on FP for corneal curvature prediction by categorizing corneas into steep, regular, and flat groups. The EfficientNetB0 architecture with transfer learning was used to learn FP patterns to predict flat, regular, and steep corneas. In validation, the model achieved a multiclass accuracy of 0.727, a Matthews correlation coefficient of 0.519, and an unweighted Cohen's κ of 0.590. The areas under the receiver operating characteristic curves for binary prediction of flat and steep corneas were 0.863 and 0.848, respectively. The optic nerve and its peripheral areas were the main focus of the model. The developed algorithm shows that FP can potentially be used as an imaging modality to estimate corneal curvature in the post-COVID-19 era, whereby patients may benefit from the detection of abnormal corneal curvatures using FP in the telemedicine setting.

Deep Transfer Learning for Ethnically Distinct Populations: Prediction of Refractive Error Using Optical Coherence Tomography.

INTRODUCTION: The mismatch between training and testing data distribution causes significant degradation in the deep learning model performance in multi-ethnic scenarios. To reduce the performance differences between ethnic groups and image domains, we built a deep transfer learning model with adaptation training to predict uncorrected refractive errors using posterior segment optical coherence tomography (OCT) images of the macula and optic nerve. METHODS: Observational, cross-sectional, multicenter study design. We pre-trained a deep learning model on OCT images from the B&VIIT Eye Center (Seoul, South Korea) (N = 2602 eyes of 1301 patients). OCT images from Poona Eye Care (Pune, India) were chronologically sorted into adaptation training data (N = 60 eyes of 30 patients) for transfer learning and test data (N = 142 eyes of 71 patients) for validation. Deep learning models were trained to predict spherical equivalent (SE) and mean keratometry (K) values via transfer learning for domain adaptation. RESULTS: Both adaptation models for SE and K were significantly better than those without adaptation (P < 0.001). In myopia/hyperopia classification, the model trained on circular optic disc OCT images yielded the best performance (accuracy = 74.7%). It also performed best in estimating SE with the lowest mean absolute error (MAE) of 1.58 D. For classifying the degree of corneal curvature, the optic nerve vertical algorithm performed best (accuracy = 65.7%). The optic nerve horizontal model achieved the lowest MAE (1.85 D) when predicting the K value. Saliency maps frequently highlighted the retinal nerve fiber layers. CONCLUSIONS: Adaptation training via transfer learning is an effective technique for estimating refractive errors and K values using macular and optic nerve OCT images from ethnically heterogeneous populations. Further studies with larger sample sizes and various data sources are needed to confirm the feasibility of the proposed algorithm.

Comparison of implantation of posterior chamber phakic IOL implantation and laser vision correction in terms of corneal endothelial cells: 3-year observational paired-eye study.

PURPOSE: To compare the postoperative endothelial cell counts of EVO-implantable collamer lenses (ICLs) with a central hole (V4c and V5) and laser vision correction surgery (laser in situ keratomileusis or photorefractive keratectomy). SETTING: B&VIIT Eye Center, Seoul, South Korea. DESIGN: Retrospective observational and paired contralateral study. METHODS: 62 eyes of 31 patients who underwent EVO-ICLs with a central hole implantation in one eye (phakic intraocular lens [pIOL] group) and laser vision correction in the contralateral eye (LVC group) to correct refractive errors were retrospectively reviewed. Central endothelial cell density (ECD), percentage of hexagonal cells (HEX), coefficient of variation (CoV) in cell size, and adverse events were evaluated for at least 3 years. The endothelial cells were observed using a noncontact specular microscope. RESULTS: All surgeries were performed, without complications during the follow-up period. The mean ECD loss values compared with the preoperative measurements were 6.65% and 4.95% during the 3 years after pIOL and LVC, respectively. There was no significant difference in ECD loss compared with the preoperative values (paired t test, P = .188) between the 2 groups. No significant loss in ECD was observed at any timepoint. The pIOL group showed higher HEX ( P = .018) and lower CoV ( P = .006) values than the LVC group at the last visit. CONCLUSIONS: According to the authors' experience, the EVO-ICL with a central hole implantation was a safe and stable vision correction method. Moreover, it did not induce statistically significant changes in ECD at 3 years postoperatively compared with LVC. However, further long-term follow-up studies are required to confirm these results.

Development of an implantable collamer lens sizing model: a retrospective study using ANTERION swept-source optical coherence tomography and a literature review.

BACKGROUND: Optimal sizing for phakic intraocular lens (EVO-ICL with KS-AquaPort) implantation plays an important role in preventing postoperative complications. We aimed to formulate optimal lens sizing using ocular biometric parameters measured with a Heidelberg anterior segment optical coherence tomography (AS-OCT) device. METHODS: We retrospectively analyzed 892 eyes of 471 healthy subjects treated with an intraocular collamer lens (ICL) and assigned them to either the development (80%) or validation (20%) set. We built vault prediction models using the development set via classic linear regression methods as well as partial least squares and least absolute shrinkage and selection operator (LASSO) regression techniques. We evaluated prediction abilities based on the Bayesian information criterion (BIC) to select the best prediction model. The performance was measured using Pearson's correlation coefficient and the mean squared error (MAE) between the achieved and predicted results. RESULTS: Measurements of aqueous depth (AQD), anterior chamber volume, anterior chamber angle (ACA) distance, spur-to-spur distance, crystalline lens thickness (LT), and white-to-white distance from ANTERION were highly associated with the ICL vault. The LASSO model using the AQD, ACA distance, and LT showed the best BIC results for postoperative ICL vault prediction. In the validation dataset, the LASSO model showed the strongest correlation (r = 0.582, P < 0.001) and the lowest MAE (104.7 µm). CONCLUSION: This is the first study to develop a postoperative ICL vault prediction and lens-sizing model based on the ANTERION. As the measurements from ANTERION and other AS-OCT devices are not interchangeable, ANTERION may be used for optimal ICL sizing using our formula. Because our model was developed based on the East Asian population, further studies are needed to explore the role of this prediction model in different populations.

Predicting Postoperative Anterior Chamber Angle for Phakic Intraocular Lens Implantation Using Preoperative Anterior Segment Metrics.

Purpose: The anterior chamber angle (ACA) is a critical factor in posterior chamber phakic intraocular lens (EVO Implantable Collamer Lens [ICL]) implantation. Herein, we predicted postoperative ACAs to select the optimal ICL size to reduce narrow ACA-related complications. Methods: Regression models were constructed using pre-operative anterior segment optical coherence tomography metrics to predict postoperative ACAs, including trabecular-iris angles (TIAs) and scleral-spur angles (SSAs) at 500 µm and 750 µm from the scleral spur (TIA500, TIA750, SSA500, and SSA750). Data from three expert surgeons were assigned to the development (N = 430 eyes) and internal validation (N = 108 eyes) datasets. Additionally, data from a novice surgeon (N = 42 eyes) were used for external validation. Results: Postoperative ACAs were highly predictable using the machine-learning (ML) technique (extreme gradient boosting regression [XGBoost]), with mean absolute errors (MAEs) of 4.42 degrees, 3.77 degrees, 5.25 degrees, and 4.30 degrees for TIA500, TIA750, SSA500, and SSA750, respectively, in internal validation. External validation also showed MAEs of 3.93 degrees, 3.86 degrees, 5.02 degrees, and 4.74 degrees for TIA500, TIA750, SSA500, and SSA750, respectively. Linear regression using the pre-operative anterior chamber depth, anterior chamber width, crystalline lens rise, TIA, and ICL size also exhibited good performance, with no significant difference compared with XGBoost in the validation sets. Conclusions: We developed linear regression and ML models to predict postoperative ACAs for ICL surgery anterior segment metrics. These will prevent surgeons from overlooking the risks associated with the narrowing of the ACA. Translational Relevance: Using the proposed algorithms, surgeons can consider the postoperative ACAs to increase surgical accuracy and safety.

Comparison of Corneal Higher-Order Aberrations Following Topography-Guided LASIK and SMILE for Myopic Correction: A Propensity Score Matching Analysis.

Ocular aberrations, particularly corneal higher-order aberrations (HOAs), which impair visual quality, should be minimized or corrected during any laser vision correction. We compared changes in visual outcomes, including HOAs, in patients who underwent Topography-Guided laser-assisted in situ keratomileusis (TG-LASIK) or small-incision lenticule extraction (SMILE) after propensity score matching (PSM) to reduce selection bias. Of 2749 patients who underwent SMILE or TG-LASIK for myopia, 152 eyes underwent complete ophthalmic examination preoperatively and over six months postoperatively. Visual outcomes were comparatively analyzed after PSM. As a result, 45 eyes were included in each group after PSM. There was a comparable improvement in visual acuity (VA) and refractive parameters postoperatively, with no difference between the two PSM-groups. However, 6.6% in the SMILE PSM-group lost two or more lines of Snellen VA at the six-month follow-up, while none in the TG-LASIK PSM-group did. Specifically, the SMILE PSM-group showed a significant increase in corneal HOAs, including spherical aberration, coma, and total HOAs (0.0736 ± 0.162 µm; 0.181 ± 0.233 µm; and 0.151 ± 0.178 µm, respectively), whereas TG-LASIK PSM-group did not. Furthermore, SMILE PSM-group had greater postoperative corneal HOAs than those in TG-LASIK PSM-group. Collectively, TG-LASIK induces fewer corneal HOAs even after facilitating between-group comparability using PSM analysis. TG-LASIK provides better visual quality than SMILE for myopia.

Oculomics for sarcopenia prediction: a machine learning approach toward predictive, preventive, and personalized medicine.

Aims: Sarcopenia is characterized by a gradual loss of skeletal muscle mass and strength with increased adverse outcomes. Recently, large-scale epidemiological studies have demonstrated a relationship between several chronic disorders and ocular pathological conditions using an oculomics approach. We hypothesized that sarcopenia can be predicted through eye examinations, without invasive tests or radiologic evaluations in the context of predictive, preventive, and personalized medicine (PPPM/3PM). Methods: We analyzed data from the Korean National Health and Nutrition Examination Survey (KNHANES). The training set (80%, randomly selected from 2008 to 2010) data were used to construct the machine learning models. Internal (20%, randomly selected from 2008 to 2010) and external (from the KNHANES 2011) validation sets were used to assess the ability to predict sarcopenia. We included 8092 participants in the final dataset. Machine learning models (XGBoost) were trained on ophthalmological examinations and demographic factors to detect sarcopenia. Results: In the exploratory analysis, decreased levator function (odds ratio [OR], 1.41; P value <0.001), cataracts (OR, 1.31; P value = 0.013), and age-related macular degeneration (OR, 1.38; P value = 0.026) were associated with an increased risk of sarcopenia in men. In women, an increased risk of sarcopenia was associated with blepharoptosis (OR, 1.23; P value = 0.038) and cataracts (OR, 1.29; P value = 0.010). The XGBoost technique showed areas under the receiver operating characteristic curves (AUCs) of 0.746 and 0.762 in men and women, respectively. The external validation achieved AUCs of 0.751 and 0.785 for men and women, respectively. For practical and fast hands-on experience with the predictive model for practitioners who may be willing to test the whole idea of sarcopenia prediction based on oculomics data, we developed a simple web-based calculator application (https://knhanesoculomics.github.io/sarcopenia) to predict the risk of sarcopenia and facilitate screening, based on the model established in this study. Conclusion: Sarcopenia is treatable before the vicious cycle of sarcopenia-related deterioration begins. Therefore, early identification of individuals at a high risk of sarcopenia is essential in the context of PPPM. Our oculomics-based approach provides an effective strategy for sarcopenia prediction. The proposed method shows promise in significantly increasing the number of patients diagnosed with sarcopenia, potentially facilitating earlier intervention. Through patient oculometric monitoring, various pathological factors related to sarcopenia can be simultaneously analyzed, and doctors can provide personalized medical services according to each cause. Further studies are needed to confirm whether such a prediction algorithm can be used in real-world clinical settings to improve the diagnosis of sarcopenia. Supplementary Information: The online version contains supplementary material available at 10.1007/s13167-022-00292-3.

Clinical outcomes of the immediate reapplication of small-incision lenticule extraction without adjusting the surgical parameters after suction loss.

This study was to analyze the clinical outcomes of immediate reapplication of small-incision lenticule extraction (SMILE) without adjusting the surgical parameters after suction loss and to compare the outcomes with contralateral eyes that underwent uneventful SMILE. A total of 74 patients who underwent uneventful SMILE in one eye (Uneventful group) and immediate reapplication of SMILE without adjusting the surgical parameters after suction loss in the contralateral eye (Suction loss group) were included. Suction loss occurred during the posterior lenticule surface cut in 39 eyes (53%) and the cap cut in 35 eyes (47%). Surgical outcomes, including visual acuity, manifest refraction, keratometry, and corneal wavefront aberrations, were evaluated at 6 months postoperatively. The mean uncorrected distance visual acuity (UDVA), corrected distance visual acuity (CDVA), and spherical equivalent were - 0.02 ± 0.07, - 0.04 ± 0.04, and - 0.10 ± 0.46 diopters (D), respectively, in the Suction loss group and - 0.02 ± 0.07, - 0.04 ± 0.05, and - 0.19 ± 0.53 D, respectively (P = 0.965, 0.519, and 0.265, respectively), in the Uneventful group. Changes between the preoperative and 6-month postoperative total corneal aberrations, spherical aberrations, and horizontal and vertical coma did not significantly differ between the Suction loss and Uneventful groups. Immediate reapplication of SMILE without adjusting the surgical parameters after suction loss resulted in good surgical outcomes that were comparable with those of uneventful SMILE.

Machine learning predicting myopic regression after corneal refractive surgery using preoperative data and fundus photography.

PURPOSE: Myopic regression after surgery is the most common long-term complication of refractive surgery, but it is difficult to identify myopic regression without long-term observation. This study aimed to develop machine learning models to identify high-risk patients for refractive regression based on preoperative data and fundus photography. METHODS: This retrospective study assigned subjects to the training (n = 1606 eyes) and validation (n = 403 eyes) datasets with chronological data splitting. Machine learning models with ResNet50 (for image analysis) and XGBoost (for integration of all variables and fundus photography) were developed based on subjects who underwent corneal refractive surgery. The primary outcome was the predictive performance for the presence of myopic regression at 4 years of follow-up examination postoperatively. RESULTS: By integrating all factors and fundus photography, the final combined machine learning model showed good performance to predict myopic regression of more than 0.5 D (area under the receiver operating characteristic curve [ROC-AUC], 0.753; 95% confidence interval [CI], 0.710-0.793). The performance of the final model was better than the single ResNet50 model only using fundus photography (ROC-AUC, 0.673; 95% CI, 0.627-0.716). The top-five most important input features were fundus photography, preoperative anterior chamber depth, planned ablation thickness, age, and preoperative central corneal thickness. CONCLUSION: Our machine learning algorithm provides an efficient strategy to identify high-risk patients with myopic regression without additional labor, cost, and time. Surgeons might benefit from preoperative risk assessment of myopic regression, patient counseling before surgery, and surgical option decisions.

A deep learning approach for detection of shallow anterior chamber depth based on the hidden features of fundus photographs.

BACKGROUND AND OBJECTIVES: Patients with angle-closure glaucoma (ACG) are asymptomatic until they experience a painful attack. Shallow anterior chamber depth (ACD) is considered a significant risk factor for ACG. We propose a deep learning approach to detect shallow ACD using fundus photographs and to identify the hidden features of shallow ACD. METHODS: This retrospective study assigned healthy subjects to the training (n = 1188 eyes) and test (n = 594) datasets (prospective validation design). We used a deep learning approach to estimate ACD and build a classification model to identify eyes with a shallow ACD. The proposed method, including subtraction of the input and output images of CycleGAN and a thresholding algorithm, was adopted to visualize the characteristic features of fundus photographs with a shallow ACD. RESULTS: The deep learning model integrating fundus photographs and clinical variables achieved areas under the receiver operating characteristic curve of 0.978 (95% confidence interval [CI], 0.963-0.988) for an ACD ≤ 2.60 mm and 0.895 (95% CI, 0.868-0.919) for an ACD ≤ 2.80 mm, and outperformed the regression model using only clinical variables. However, the difference between shallow and deep ACD classes on fundus photographs was difficult to be detected with the naked eye. We were unable to identify the features of shallow ACD using the Grad-CAM. The CycleGAN-based feature images showed that area around the macula and optic disk significantly contributed to the classification of fundus photographs with a shallow ACD. CONCLUSIONS: We demonstrated the feasibility of a novel deep learning model to detect a shallow ACD as a screening tool for ACG using fundus photographs. The CycleGAN-based feature map showed the hidden characteristic features of shallow ACD that were previously undetectable by conventional techniques and ophthalmologists. This framework will facilitate the early detection of shallow ACD to prevent overlooking the risks associated with ACG.

Application of generative adversarial networks (GAN) for ophthalmology image domains: a survey.

BACKGROUND: Recent advances in deep learning techniques have led to improved diagnostic abilities in ophthalmology. A generative adversarial network (GAN), which consists of two competing types of deep neural networks, including a generator and a discriminator, has demonstrated remarkable performance in image synthesis and image-to-image translation. The adoption of GAN for medical imaging is increasing for image generation and translation, but it is not familiar to researchers in the field of ophthalmology. In this work, we present a literature review on the application of GAN in ophthalmology image domains to discuss important contributions and to identify potential future research directions. METHODS: We performed a survey on studies using GAN published before June 2021 only, and we introduced various applications of GAN in ophthalmology image domains. The search identified 48 peer-reviewed papers in the final review. The type of GAN used in the analysis, task, imaging domain, and the outcome were collected to verify the usefulness of the GAN. RESULTS: In ophthalmology image domains, GAN can perform segmentation, data augmentation, denoising, domain transfer, super-resolution, post-intervention prediction, and feature extraction. GAN techniques have established an extension of datasets and modalities in ophthalmology. GAN has several limitations, such as mode collapse, spatial deformities, unintended changes, and the generation of high-frequency noises and artifacts of checkerboard patterns. CONCLUSIONS: The use of GAN has benefited the various tasks in ophthalmology image domains. Based on our observations, the adoption of GAN in ophthalmology is still in a very early stage of clinical validation compared with deep learning classification techniques because several problems need to be overcome for practical use. However, the proper selection of the GAN technique and statistical modeling of ocular imaging will greatly improve the performance of each image analysis. Finally, this survey would enable researchers to access the appropriate GAN technique to maximize the potential of ophthalmology datasets for deep learning research.

Simple Code Implementation for Deep Learning-Based Segmentation to Evaluate Central Serous Chorioretinopathy in Fundus Photography.

PURPOSE: Central serous chorioretinopathy (CSC) is a retinal disease that frequently shows resolution and recurrence with serous detachment of the neurosensory retina. Here, we present a deep learning analysis of subretinal fluid (SRF) lesion segmentation in fundus photographs to evaluate CSC. METHODS: We collected 194 fundus photographs of SRF lesions from the patients with CSC. Three graders manually annotated of the entire SRF area in the retinal images. The dataset was randomly separated into training (90%) and validation (10%) datasets. We used the U-Net segmentation model based on conditional generative adversarial networks (pix2pix) to detect the SRF lesions. The algorithms were trained and validated using Google Colaboratory. Researchers did not need prior knowledge of coding skills or computing resources to implement this code. RESULTS: The validation results showed that the Jaccard index and Dice coefficient scores were 0.619 and 0.763, respectively. In most cases, the segmentation results overlapped with most of the reference areas in the annotated images. However, cases with exceptional SRFs were not accurate in terms of prediction. Using Colaboratory, the proposed segmentation task ran easily in a web-based environment without setup or personal computing resources. CONCLUSIONS: The results suggest that the deep learning model based on U-Net from the pix2pix algorithm is suitable for the automatic segmentation of SRF lesions to evaluate CSC. TRANSLATIONAL RELEVANCE: Our code implementation has the potential to facilitate ophthalmology research; in particular, deep learning-based segmentation can assist in the development of pathological lesion detection solutions.

Deep learning for predicting uncorrected refractive error using posterior segment optical coherence tomography images.

BACKGROUND/OBJECTIVES: This study aimed to evaluate a deep learning model for estimating uncorrected refractive error using posterior segment optical coherence tomography (OCT) images. METHODS: In this retrospective study, we assigned healthy subjects to development (N = 688 eyes of 344 subjects) and test (N = 248 eyes of 124 subjects) datasets (prospective validation design). We developed and validated OCT-based deep learning models to estimate refractive error. A regression model based on a pretrained ResNet50 architecture was trained using horizontal OCT images to predict the spherical equivalent (SE). The performance of the deep learning model for detecting high myopia was also evaluated. A saliency map was generated using the Grad-CAM technique to visualize the characteristic features. RESULTS: The developed model showed a low mean absolute error for SE prediction (2.66 D) and a significant Pearson correlation coefficient of 0.588 (P < 0.001) in the test dataset validation. To detect high myopia, the model yielded an area under the receiver operating characteristic curve of 0.813 (95% confidence interval [CI], 0.744-0.881) and an accuracy of 71.4% (95% CI, 65.3-76.9%). The inner retinal layers and relatively steepened curvatures were highlighted using a saliency map to detect high myopia. CONCLUSION: A deep learning algorithm showed that OCT could potentially be used as an imaging modality to estimate refractive error. This method will facilitate the evaluation of refractive error to prevent clinicians from overlooking the risks associated with refractive error during OCT assessment.

Development of a Web-Based Ensemble Machine Learning Application to Select the Optimal Size of Posterior Chamber Phakic Intraocular Lens.

Purpose: Selecting the optimal lens size by predicting the postoperative vault can reduce complications after implantation of an implantable collamer lens with a central-hole (ICL with KS-aquaport). We built a web-based machine learning application that incorporated clinical measurements to predict the postoperative ICL vault and select the optimal ICL size. Methods: We applied the stacking ensemble technique based on eXtreme Gradient Boosting (XGBoost) and a light gradient boosting machine to pre-operative ocular data from two eye centers to predict the postoperative vault. We assigned the Korean patient data to a training (N = 2756 eyes) and internal validation (N = 693 eyes) datasets (prospective validation). Japanese patient data (N = 290 eyes) were used as an independent external dataset from different centers to validate the model. Results: We developed an ensemble model that showed statistically better performance with a lower mean absolute error for ICL vault prediction (106.88 µm and 143.69 µm in the internal and external validation, respectively) than the other machine learning techniques and the classic ICL sizing methods did when applied to both validation datasets. Considering the lens size selection accuracy, our proposed method showed the best performance for both reference datasets (75.9% and 67.4% in the internal and external validation, respectively). Conclusions: Applying the ensemble approach to a large dataset of patients who underwent ICL implantation resulted in a more accurate prediction of vault size and selection of the optimal ICL size. Translational Relevance: We developed a web-based application for ICL sizing to facilitate the use of machine learning calculators for clinicians.

Deep-learning-based cardiovascular risk stratification using coronary artery calcium scores predicted from retinal photographs.

BACKGROUND: Coronary artery calcium (CAC) score is a clinically validated marker of cardiovascular disease risk. We developed and validated a novel cardiovascular risk stratification system based on deep-learning-predicted CAC from retinal photographs. METHODS: We used 216 152 retinal photographs from five datasets from South Korea, Singapore, and the UK to train and validate the algorithms. First, using one dataset from a South Korean health-screening centre, we trained a deep-learning algorithm to predict the probability of the presence of CAC (ie, deep-learning retinal CAC score, RetiCAC). We stratified RetiCAC scores into tertiles and used Cox proportional hazards models to evaluate the ability of RetiCAC to predict cardiovascular events based on external test sets from South Korea, Singapore, and the UK Biobank. We evaluated the incremental values of RetiCAC when added to the Pooled Cohort Equation (PCE) for participants in the UK Biobank. FINDINGS: RetiCAC outperformed all single clinical parameter models in predicting the presence of CAC (area under the receiver operating characteristic curve of 0·742, 95% CI 0·732-0·753). Among the 527 participants in the South Korean clinical cohort, 33 (6·3%) had cardiovascular events during the 5-year follow-up. When compared with the current CAC risk stratification (0, >0-100, and >100), the three-strata RetiCAC showed comparable prognostic performance with a concordance index of 0·71. In the Singapore population-based cohort (n=8551), 310 (3·6%) participants had fatal cardiovascular events over 10 years, and the three-strata RetiCAC was significantly associated with increased risk of fatal cardiovascular events (hazard ratio [HR] trend 1·33, 95% CI 1·04-1·71). In the UK Biobank (n=47 679), 337 (0·7%) participants had fatal cardiovascular events over 10 years. When added to the PCE, the three-strata RetiCAC improved cardiovascular risk stratification in the intermediate-risk group (HR trend 1·28, 95% CI 1·07-1·54) and borderline-risk group (1·62, 1·04-2·54), and the continuous net reclassification index was 0·261 (95% CI 0·124-0·364). INTERPRETATION: A deep learning and retinal photograph-derived CAC score is comparable to CT scan-measured CAC in predicting cardiovascular events, and improves on current risk stratification approaches for cardiovascular disease events. These data suggest retinal photograph-based deep learning has the potential to be used as an alternative measure of CAC, especially in low-resource settings. FUNDING: Yonsei University College of Medicine; Ministry of Health and Welfare, Korea Institute for Advancement of Technology, South Korea; Agency for Science, Technology, and Research; and National Medical Research Council, Singapore.