Automated Diagnosis of Plus Disease in Retinopathy of Prematurity Using Deep Convolutional Neural Networks

Detection of active and inactive phases of thyroid-associated ophthalmopathy using deep convolutional neural network

Background

This study aimed to establish a deep learning system for detecting the active and inactive phases of thyroid-associated ophthalmopathy (TAO) using magnetic resonance imaging (MRI). This system could provide faster, more accurate, and more objective assessments across populations.

Methods

A total of 160 MRI images of patients with TAO, who visited the Ophthalmology Clinic of the Ninth People’s Hospital, were retrospectively obtained for this study. Of these, 80% were used for training and validation, and 20% were used for testing. The deep learning system, based on deep convolutional neural network, was established to distinguish patients with active phase from those with inactive phase. The accuracy, precision, sensitivity, specificity, F1 score and area under the receiver operating characteristic curve were analyzed. Besides, visualization method was applied to explain the operation of the networks.

Results

Network A inherited from Visual Geometry Group network. The accuracy, specificity and sensitivity were 0.863±0.055, 0.896±0.042 and 0.750±0.136 respectively. Due to the recurring phenomenon of vanishing gradient during the training process of network A, we added parts of Residual Neural Network to build network B. After modification, network B improved the sensitivity (0.821±0.021) while maintaining a good accuracy (0.855±0.018) and a good specificity (0.865±0.021).

Conclusions

The deep convolutional neural network could automatically detect the activity of TAO from MRI images with strong robustness, less subjective judgment, and less measurement error. This system could standardize the diagnostic process and speed up the treatment decision making for TAO.

Deep learning-enabled medical computer vision

A decade of unprecedented progress in artificial intelligence (AI) has demonstrated the potential for many fields-including medicine-to benefit from the insights that AI techniques can extract from data. Here we survey recent progress in the development of modern computer vision techniques-powered by deep learning-for medical applications, focusing on medical imaging, medical video, and clinical deployment. We start by briefly summarizing a decade of progress in convolutional neural networks, including the vision tasks they enable, in the context of healthcare. Next, we discuss several example medical imaging applications that stand to benefit-including cardiology, pathology, dermatology, ophthalmology-and propose new avenues for continued work. We then expand into general medical video, highlighting ways in which clinical workflows can integrate computer vision to enhance care. Finally, we discuss the challenges and hurdles required for real-world clinical deployment of these technologies.

Detection of Referable Horizontal Strabismus in Children's Primary Gaze Photographs Using Deep Learning

PURPOSE

This study implements and demonstrates a deep learning (DL) approach for screening referable horizontal strabismus based on primary gaze photographs using clinical assessments as a reference. The purpose of this study was to develop and evaluate deep learning algorithms that screen referable horizontal strabismus in children's primary gaze photographs.

METHODS

DL algorithms were developed and trained using primary gaze photographs from two tertiary hospitals of children with primary horizontal strabismus who underwent surgery as well as orthotropic children who underwent routine refractive tests. A total of 7026 images (3829 non-strabismus from 3021 orthoptics [healthy] subjects and 3197 strabismus images from 2772 subjects) were used to develop the DL algorithms. The DL model was evaluated by 5-fold cross-validation and tested on an independent validation data set of 277 images. The diagnostic performance of the DL algorithm was assessed by calculating the accuracy, sensitivity, specificity, and area under the receiver operating characteristic curve (AUC).

RESULTS

Using 5-fold cross-validation during training, the average AUCs of the DL models were approximately 0.99. In the external validation data set, the DL algorithm achieved an AUC of 0.99 with a sensitivity of 94.0% and a specificity of 99.3%. The DL algorithm's performance (with an accuracy of 0.95) in diagnosing referable horizontal strabismus was better than that of the resident ophthalmologists (with accuracy ranging from 0.81 to 0.85).

CONCLUSIONS

We developed and evaluated a DL model to automatically identify referable horizontal strabismus using primary gaze photographs. The diagnostic performance of the DL model is comparable to or better than that of ophthalmologists.

TRANSLATIONAL RELEVANCE

DL methods that automate the detection of referable horizontal strabismus can facilitate clinical assessment and screening for children at risk of strabismus.

Technology Can Augment, but Not Replace, Critical Human Skills Needed for Patient Care

The practice of medicine is changing rapidly as a consequence of electronic health record adoption, new technologies for patient care, disruptive innovations that breakdown professional hierarchies, and evolving societal norms. Collectively, these have resulted in the modification of the physician's role as the gatekeeper for health care, increased shift-based care, and amplified interprofessional team-based care. Technological innovations present opportunities as well as challenges. Artificial intelligence, which has great potential, has already transformed some tasks, particularly those involving image interpretation. Ubiquitous access to information via the Internet by physicians and patients alike presents benefits as well as drawbacks: patients and providers have ready access to virtually all of human knowledge, but some websites are contaminated with misinformation and many people have difficulty differentiating between solid, evidence-based data and untruths. The role of the future physician will shift as complexity in health care increases and as artificial intelligence and other technologies advance. These technological advances demand new skills of physicians; memory and knowledge accumulation will diminish in importance while information management skills will become more important. In parallel, medical educators must enhance their teaching and assessment of critical human skills (e.g., clear communication, empathy) in the delivery of patient care. The authors emphasize the enduring role of critical human skills in safe and effective patient care even as medical practice is increasingly guided by artificial intelligence and related technology, and they suggest new and longitudinal ways of assessing essential noncognitive skills to meet the demands of the future. The authors envision practical and achievable benefits accruing to patients and providers if practitioners leverage technological advancements to facilitate the development of their critical human skills.

Semantic Segmentation of Eye Fundus Images Using Convolutional Neural Networks

The article reviews the problems of eye bottom fundus analysis and semantic segmentation algorithms used to distinguish eye vessels, optical disk. Various diseases, such as glaucoma, hypertension, diabetic retinopathy, macular degeneration, etc., can be diagnosed by changes and anomalies of vesssels and optical disk. For semantic segmentation convolutional neural networks, especially U-Net architecture, are well suited. Recently a number of U-Net modifications have been developed that deliver excellent performance results.

Convolutional Neural Networks with Transfer Learning for Recognition of COVID-19: A Comparative Study of Different Approaches

To judge the ability of convolutional neural networks (CNNs) to effectively and efficiently transfer image representations learned on the ImageNet dataset to the task of recognizing COVID-19 in this work, we propose and analyze four approaches. For this purpose, we use VGG16, ResNetV2, InceptionResNetV2, DenseNet121, and MobileNetV2 CNN models pre-trained on ImageNet dataset to extract features from X-ray images of COVID and Non-COVID patients. Simulations study performed by us reveal that these pre-trained models have a different level of ability to transfer image representation. We find that in the approaches that we have proposed, if we use either ResNetV2 or DenseNet121 to extract features, then the performance of these approaches to detect COVID-19 is better. One of the important findings of our study is that the use of principal component analysis for feature selection improves efficiency. The approach using the fusion of features outperforms all the other approaches, and with this approach, we could achieve an accuracy of 0.94 for a three-class classification problem. This work will not only be useful for COVID-19 detection but also for any domain with small datasets.

Retinal vascular tortuosity: Mechanisms and measurements

Retinal vessel tortuosity has been used in the diagnosis and management of different clinical situations. Notwithstanding, basic concepts, standards and tools of measurement, reliable normative data and clinical applications have many gaps or points of divergence. In this review we discuss triggering causes of retinal vessel tortuosity and resources used to assess and quantify it, as well as current limitations.

Automatic Identification of Referral-Warranted Diabetic Retinopathy Using Deep Learning on Mobile Phone Images

Purpose

To evaluate the performance of a deep learning algorithm in the detection of referral-warranted diabetic retinopathy (RDR) on low-resolution fundus images acquired with a smartphone and indirect ophthalmoscope lens adapter.

Methods

An automated deep learning algorithm trained on 92,364 traditional fundus camera images was tested on a dataset of smartphone fundus images from 103 eyes acquired from two previously published studies. Images were extracted from live video screenshots from fundus examinations using a commercially available lens adapter and exported as a screenshot from live video clips filmed at 1080p resolution. Each image was graded twice by a board-certified ophthalmologist and compared to the output of the algorithm, which classified each image as having RDR (moderate nonproliferative DR or worse) or no RDR.

Results

In spite of the presence of multiple artifacts (lens glare, lens particulates/smudging, user hands over the objective lens) and low-resolution images achieved by users of various levels of medical training, the algorithm achieved a 0.89 (95% confidence interval [CI] 0.83-0.95) area under the curve with an 89% sensitivity (95% CI 81%-100%) and 83% specificity (95% CI 77%-89%) for detecting RDR on mobile phone acquired fundus photos.

Conclusions

The fully data-driven artificial intelligence-based grading algorithm herein can be used to screen fundus photos taken from mobile devices and identify with high reliability which cases should be referred to an ophthalmologist for further evaluation and treatment.

Translational Relevance

The implementation of this algorithm on a global basis could drastically reduce the rate of vision loss attributed to DR.

Multi-Institutional Assessment and Crowdsourcing Evaluation of Deep Learning for Automated Classification of Breast Density

OBJECTIVE

We developed deep learning algorithms to automatically assess BI-RADS breast density.

METHODS

Using a large multi-institution patient cohort of 108,230 digital screening mammograms from the Digital Mammographic Imaging Screening Trial, we investigated the effect of data, model, and training parameters on overall model performance and provided crowdsourcing evaluation from the attendees of the ACR 2019 Annual Meeting.

RESULTS

Our best-performing algorithm achieved good agreement with radiologists who were qualified interpreters of mammograms, with a four-class κ of 0.667. When training was performed with randomly sampled images from the data set versus sampling equal number of images from each density category, the model predictions were biased away from the low-prevalence categories such as extremely dense breasts. The net result was an increase in sensitivity and a decrease in specificity for predicting dense breasts for equal class compared with random sampling. We also found that the performance of the model degrades when we evaluate on digital mammography data formats that differ from the one that we trained on, emphasizing the importance of multi-institutional training sets. Lastly, we showed that crowdsourced annotations, including those from attendees who routinely read mammograms, had higher agreement with our algorithm than with the original interpreting radiologists.

CONCLUSION

We demonstrated the possible parameters that can influence the performance of the model and how crowdsourcing can be used for evaluation. This study was performed in tandem with the development of the ACR AI-LAB, a platform for democratizing artificial intelligence.

Implementing the new normal in ophthalmology care beyond COVID-19

The COVID-19 pandemic has altered the clinical landscape immeasurably. The need to physical distance requires rethinking how we deliver ophthalmic care. Within healthcare, we will need to focus our resources on the five T's: Utilising technology, multidisciplinary clinical teams with wide professional talents need to work efficiently to reduce patient contact time. With regular testing, this will allow us to reduce the risk further. We also must acknowledge the explosion of different modalities to train our future ophthalmologists and the global challenges and advantages that these bring. Finally, we must not forget the psychological impact that this pandemic will have on ophthalmologists and ancillary staff, and need to have robust mechanisms for support.

Update on the Use of Anti-VEGF Drugs in the Treatment of Retinopathy of Prematurity

Retinopathy of prematurity (ROP) is one of the many significant consequences of premature birth and remains one of the leading causes of visual impairment in infants. Originally, cryotherapy was used to prevent the complications of vitreous hemorrhage and retinal detachment. Subsequently, laser photocoagulation, which is at least as effective and possibly safer than cryoretinopexy, was adopted as the primary treatment for type 1 ROP (stage 2 or 3 disease in zone II with plus disease or any stage disease in zone I with plus disease or stage 3 disease in zone I without plus disease). Laser therapy has been proven effective, and has a degree of permanence that is yet to be matched by alternative treatments, but can be associated with significant ocular side effects such as myopia. Treatment of type 1 ROP with anti-vascular endothelial growth factor (VEGF) agents seems to have fewer ocular side effects than laser ablation of the retina, particularly if used to treat type 1 ROP in zone I. However, ROP recurrence is a real threat after anti-VEGF therapy and long-term systemic side effects of this therapy remain under evaluation. This review focuses on the ophthalmic and systemic benefits and risks of anti-VEGF therapies for ROP as compared to retinal photocoagulation. Anti-VEGF therapies have dramatically altered the management of ROP and have also been shown to be beneficial with regard to the visual prognosis of patients with ROP, but patients so treated require frequent short- and long-term follow-up to detect and manage potential complications associated with this form of treatment. Such information also will allow clinicians to characterize the efficacy, side effect profile, and utility of intravitreal anti-VEGF agents for this condition. Prospective studies are needed to identify the optimum anti-VEGF drug and dose. [J Pediatr Ophthalmol Strabismus. 2020;57(6):351-362.].

AI in the treatment of fertility: key considerations

Artificial intelligence (AI) has been proposed as a potential tool to help address many of the existing problems related with empirical or subjective assessments of clinical and embryological decision points during the treatment of infertility. AI technologies are reviewed and potential areas of implementation of algorithms are discussed, highlighting the importance of following a proper path for the development and validation of algorithms, including regulatory requirements, and the need for ecosystems containing enough quality data to generate it. As evidenced by the consensus of a group of experts in fertility if properly developed, it is believed that AI algorithms may help practitioners from around the globe to standardize, automate, and improve IVF outcomes for the benefit of patients. Collaboration is required between AI developers and healthcare professionals to make this happen.

Variability in Plus Disease Identified Using a Deep Learning-Based Retinopathy of Prematurity Severity Scale

PURPOSE

Retinopathy of prematurity is a leading cause of childhood blindness worldwide, but clinical diagnosis is subjective, which leads to treatment differences. Our goal was to determine objective differences in the diagnosis of plus disease between clinicians using an automated retinopathy of prematurity (ROP) vascular severity score.

DESIGN

This retrospective cohort study used data from the Imaging and Informatics in ROP Consortium, which comprises 8 tertiary care centers in North America. Fundus photographs of all infants undergoing ROP screening examinations between July 1, 2011, and December 31, 2016, were obtained.

PARTICIPANTS

Infants meeting ROP screening criteria who were diagnosed with plus disease and treatment initiated by an examining physician based on ophthalmoscopic examination results.

METHODS

An ROP severity score (1-9) was generated for each image using a deep learning (DL) algorithm.

MAIN OUTCOME MEASURES

The mean, median, and range of ROP vascular severity scores overall and for each examiner when the diagnosis of plus disease was made.

RESULTS

A total of 5255 clinical examinations in 871 babies were analyzed. Of these, 168 eyes were diagnosed with plus disease by 11 different examiners and were included in the study. The mean ± standard deviation vascular severity score for patients diagnosed with plus disease was 7.4 ± 1.9, median was 8.5 (interquartile range, 5.8-8.9), and range was 1.1 to 9.0. Within some examiners, variability in the level of vascular severity diagnosed as plus disease was present, and 1 examiner routinely diagnosed plus disease in patients with less severe disease than the other examiners (P < 0.01).

CONCLUSIONS

We observed variability both between and within examiners in the diagnosis of plus disease using DL. Prospective evaluation of clinical trial data using an objective measurement of vascular severity may help to define better the minimum necessary level of vascular severity for the diagnosis of plus disease or how other clinical features such as zone, stage, and extent of peripheral disease ought to be incorporated in treatment decisions.

Retinopathy of Prematurity: Evolving Treatment With Anti-Vascular Endothelial Growth Factor

PURPOSE

To discuss the evolution in retinopathy of prematurity since its first description as retrolental fibroplasia in the United States, including the changes in the understanding of pathophysiology; methods of diagnosis; destructive, anti-vascular endothelial growth factor (anti-VEGF), and supportive treatments; and differences in retinopathy of prematurity manifestations worldwide. The overall goal is to clarify retinopathy of prematurity currently and formulate questions to optimize future care.

STUDY DESIGN

Literature review and synthesis.

METHODS

Critical review and consideration of the literature with inclusion of historical articles and those regarding pathophysiologic risk factors, retinopathy of prematurity worldwide, basic and clinical science particularly regarding anti-VEGF mechanisms and agents tested in clinical trials.

RESULTS

Retinopathy of prematurity has evolved from affecting infants approximately 2 months premature to affecting extremely premature infants. Worldwide, retinopathy of prematurity differs and, in emerging countries, has features similar to that experienced in the United States when retinopathy of prematurity first manifested. Treatments have evolved from destruction of the peripheral avascular retina to inhibit angiogenic stimuli to anti-VEGF agents, which inhibit pathologic angiogenesis but also extend normal intraretinal angiogenesis by ordering the development of intraretinal vessels. Clinical trial evidence is accruing with the goal to develop less destructive treatments to optimize vision and that are protective to the retina and infant.

CONCLUSIONS

Goals for retinopathy of prematurity are to optimize prenatal and perinatal care, improve diagnostic acumen worldwide and refine treatment strategies, including with anti-VEGF agents, to inhibit intravitreal angiogenesis and facilitate vascularization of the previously avascular retina, which include supporting neural and vascular development of the premature infant and retina.

The state of artificial intelligence-based FDA-approved medical devices and algorithms: an online database

At the beginning of the artificial intelligence (AI)/machine learning (ML) era, the expectations are high, and experts foresee that AI/ML shows potential for diagnosing, managing and treating a wide variety of medical conditions. However, the obstacles for implementation of AI/ML in daily clinical practice are numerous, especially regarding the regulation of these technologies. Therefore, we provide an insight into the currently available AI/ML-based medical devices and algorithms that have been approved by the US Food & Drugs Administration (FDA). We aimed to raise awareness of the importance of regulatory bodies, clearly stating whether a medical device is AI/ML based or not. Cross-checking and validating all approvals, we identified 64 AI/ML based, FDA approved medical devices and algorithms. Out of those, only 29 (45%) mentioned any AI/ML-related expressions in the official FDA announcement. The majority (85.9%) was approved by the FDA with a 510(k) clearance, while 8 (12.5%) received de novo pathway clearance and one (1.6%) premarket approval (PMA) clearance. Most of these technologies, notably 30 (46.9%), 16 (25.0%), and 10 (15.6%) were developed for the fields of Radiology, Cardiology and Internal Medicine/General Practice respectively. We have launched the first comprehensive and open access database of strictly AI/ML-based medical technologies that have been approved by the FDA. The database will be constantly updated.

Detection of features associated with neovascular age-related macular degeneration in ethnically distinct data sets by an optical coherence tomography: trained deep learning algorithm

BACKGROUND

The ability of deep learning (DL) algorithms to identify eyes with neovascular age-related macular degeneration (nAMD) from optical coherence tomography (OCT) scans has been previously established. We herewith evaluate the ability of a DL model, showing excellent performance on a Korean data set, to generalse onto an American data set despite ethnic differences. In addition, expert graders were surveyed to verify if the DL model was appropriately identifying lesions indicative of nAMD on the OCT scans.

METHODS

Model development data set-12 247 OCT scans from South Korea; external validation data set-91 509 OCT scans from Washington, USA. In both data sets, normal eyes or eyes with nAMD were included. After internal testing, the algorithm was sent to the University of Washington, USA, for external validation. Area under the receiver operating characteristic curve (AUC) and precision-recall curve (AUPRC) were calculated. For model explanation, saliency maps were generated using Guided GradCAM.

RESULTS

On external validation, AUC and AUPRC remained high at 0.952 (95% CI 0.942 to 0.962) and 0.891 (95% CI 0.875 to 0.908) at the individual level. Saliency maps showed that in normal OCT scans, the fovea was the main area of interest; in nAMD OCT scans, the appropriate pathological features were areas of model interest. Survey of 10 retina specialists confirmed this.

CONCLUSION

Our DL algorithm exhibited high performance for nAMD identification in a Korean population, and generalised well to an ethnically distinct, American population. The model correctly focused on the differences within the macular area to extract features associated with nAMD.

Deep Learning Models for Automated Diagnosis of Retinopathy of Prematurity in Preterm Infants

Retinopathy of prematurity (ROP) is a disease that can cause blindness in premature infants. It is characterized by immature vascular growth of the retinal blood vessels. However, early detection and treatment of ROP can significantly improve the visual acuity of high-risk patients. Thus, early diagnosis of ROP is crucial in preventing visual impairment. However, several patients refrain from treatment owing to the lack of medical expertise in diagnosing the disease; this is especially problematic considering that the number of ROP cases is on the rise. To this end, we applied transfer learning to five deep neural network architectures for identifying ROP in preterm infants. Our results showed that the VGG19 model outperformed the other models in determining whether a preterm infant has ROP, with 96% accuracy, 96.6% sensitivity, and 95.2% specificity. We also classified the severity of the disease; the VGG19 model showed 98.82% accuracy in predicting the severity of the disease with a sensitivity and specificity of 100% and 98.41%, respectively. We performed 5-fold cross-validation on the datasets to validate the reliability of the VGG19 model and found that the VGG19 model exhibited high accuracy in predicting ROP. These findings could help promote the development of computer-aided diagnosis.

Age-related Macular Degeneration: Nutrition, Genes and Deep Learning-The LXXVI Edward Jackson Memorial Lecture

PURPOSE

To evaluate the importance of nutritional supplements, dietary pattern, and genetic associations in age-related macular degeneration (AMD); and to discuss the technique of artificial intelligence/deep learning to potentially enhance research in detecting and classifying AMD.

DESIGN

Retrospective literature review.

METHODS

To review the studies of both prospective and retrospective (post hoc) analyses of nutrition, genetic variants, and deep learning in AMD in both the Age-Related Eye Disease Study (AREDS) and AREDS2.

RESULTS

In addition to demonstrating the beneficial effects of the AREDS and AREDS2 supplements of antioxidant vitamins and zinc (plus copper) for reducing the risk of progression to late AMD, these 2 studies also confirmed the importance of high adherence to Mediterranean diet in reducing progression of AMD in persons with varying severity of disease. In persons with the protective genetic alleles of complement factor H (CFH), the Mediterranean diet had further beneficial effect. However, despite the genetic association with AMD progression, prediction models found genetic information added little to the high predictive value of baseline severity of AMD for disease progression. The technique of deep learning, an arm of artificial intelligence, using color fundus photographs from AREDS/AREDS2 was superior in some cases and noninferior in others to clinical human grading (retinal specialists) and to the gold standard of the certified reading center graders.

CONCLUSIONS

Counseling individuals affected with AMD regarding the use of the AREDS2 supplements and the beneficial association of the Mediterranean diet is an important public health message. Although genetic testing is important in research, it is not recommended for prediction of disease or to guide therapies and/or dietary interventions in AMD. Techniques in deep learning hold great promise, but further prospective research is required to validate the use of this technique to provide improvement in accuracy and sensitivity/specificity in clinical research and medical management of patients with AMD.

Retinopathy of Prematurity: How to Prevent the Third Epidemics in Developing Countries

Retinopathy of prematurity (ROP) is vasoproliferative disease affecting preterm infants and is a leading cause of avoidable childhood blindness worldwide. The world is currently experiencing the third epidemic of ROP, where majority of the cases are from middle-income countries. Over 40% of the world's premature infants were born in India, China, Bangladesh, Pakistan, and Indonesia. Together with other neighboring nations, this region has unique challenges in ROP management. Key aspects of the challenges including heavier and more mature infants developing severe ROP. Current strategies include adoption of national screening guidelines, telemedicine, integrating vision rehabilitation and software innovations in the form of artificial intelligence. This review overviews some of these aspects.

The Role of Telemedicine to Alleviate an Increasingly Burdened Healthcare System: Retinopathy of Prematurity

Telemedicine-based remote digital fundus imaging (RDFI-TM) offers a promising platform for the screening of retinopathy of prematurity. RDFI-TM addresses some of the challenges faced by ophthalmologists in examining this vulnerable population in both low- and high-income countries. In this review, we studied the evidence on the use of RDFI-TM and analyzed the practical framework for RDFI-TM systems. We assessed the novel technological advances that can be deployed within RDFI-TM systems including noncontact imaging systems, smartphone-based imaging tools, and deep learning algorithms.

Comparison of Convolutional Neural Network Models for Determination of Vocal Fold Normality in Laryngoscopic Images

OBJECTIVES

Deep learning using convolutional neural networks (CNNs) is widely used in medical imaging research. This study was performed to investigate if vocal fold normality in laryngoscopic images can be determined by CNN-based deep learning and to compare accuracy of CNN models and explore the feasibility of application of deep learning on laryngoscopy.

METHODS

Laryngoscopy videos were screen-captured and each image was cropped to include abducted vocal fold regions. A total of 2216 image (899 normal, 1317 abnormal) were allocated to training, validation, and test sets. Augmentation of training sets was used to train a constructed CNN model with six layers (CNN6), VGG16, Inception V3, and Xception models. Trained models were applied to the test set; for each model, receiver operating characteristic curves and cutoff values were obtained. Sensitivity, specificity, positive predictive value, negative predictive value, and accuracy were calculated. The best model was employed in video-streams and localization of features was attempted using Grad-CAM.

RESULTS

All of the trained models showed high area under the receiver operating characteristic curve and the most discriminative cutoff levels of probability of normality were determined to be 35.6%, 61.8%, 13.5%, 39.7% for CNN6, VGG16, Inception V3, and Xception models, respectively. Accuracy of the CNN models selecting normal and abnormal vocal folds in the test set was 82.3%, 99.7%, 99.1%, and 83.8%, respectively.

CONCLUSION

All four models showed acceptable diagnostic accuracy. Performance of VGG16 and Inception V3 was better than the simple CNN6 model and the recently published Xception model. Real-time classification with a combination of the VGG16 model, OpenCV, and Grad-CAM on a video stream showed the potential clinical applications of the deep learning model in laryngoscopy.

Automated detection of early-stage ROP using a deep convolutional neural network

BACKGROUND/AIM

To automatically detect and classify the early stages of retinopathy of prematurity (ROP) using a deep convolutional neural network (CNN).

METHODS

This retrospective cross-sectional study was conducted in a referral medical centre in Taiwan. Only premature infants with no ROP, stage 1 ROP or stage 2 ROP were enrolled. Overall, 11 372 retinal fundus images were compiled and split into 10 235 images (90%) for training, 1137 (10%) for validation and 244 for testing. A deep CNN was implemented to classify images according to the ROP stage. Data were collected from December 17, 2013 to May 24, 2019 and analysed from December 2018 to January 2020. The metrics of sensitivity, specificity and area under the receiver operating characteristic curve were adopted to evaluate the performance of the algorithm relative to the reference standard diagnosis.

RESULTS

The model was trained using fivefold cross-validation, yielding an average accuracy of 99.93%±0.03 during training and 92.23%±1.39 during testing. The sensitivity and specificity scores of the model were 96.14%±0.87 and 95.95%±0.48, 91.82%±2.03 and 94.50%±0.71, and 89.81%±1.82 and 98.99%±0.40 when predicting no ROP versus ROP, stage 1 ROP versus no ROP and stage 2 ROP, and stage 2 ROP versus no ROP and stage 1 ROP, respectively.

CONCLUSIONS

The proposed system can accurately differentiate among ROP early stages and has the potential to help ophthalmologists classify ROP at an early stage.

Interpretation of artificial intelligence studies for the ophthalmologist

PURPOSE OF REVIEW

The use of artificial intelligence (AI) in ophthalmology has increased dramatically. However, interpretation of these studies can be a daunting prospect for the ophthalmologist without a background in computer or data science. This review aims to share some practical considerations for interpretation of AI studies in ophthalmology.

RECENT FINDINGS

It can be easy to get lost in the technical details of studies involving AI. Nevertheless, it is important for clinicians to remember that the fundamental questions in interpreting these studies remain unchanged - What does this study show, and how does this affect my patients? Being guided by familiar principles like study purpose, impact, validity, and generalizability, these studies become more accessible to the ophthalmologist. Although it may not be necessary for nondomain experts to understand the exact AI technical details, we explain some broad concepts in relation to AI technical architecture and dataset management.

SUMMARY

The expansion of AI into healthcare and ophthalmology is here to stay. AI systems have made the transition from bench to bedside, and are already being applied to patient care. In this context, 'AI education' is crucial for ophthalmologists to be confident in interpretation and translation of new developments in this field to their own clinical practice.

Artificial intelligence and deep learning in ophthalmology - present and future (Review)

Since its introduction in 1959, artificial intelligence technology has evolved rapidly and helped benefit research, industries and medicine. Deep learning, as a process of artificial intelligence (AI) is used in ophthalmology for data analysis, segmentation, automated diagnosis and possible outcome predictions. The association of deep learning and optical coherence tomography (OCT) technologies has proven reliable for the detection of retinal diseases and improving the diagnostic performance of the eye's posterior segment diseases. This review explored the possibility of implementing and using AI in establishing the diagnosis of retinal disorders. The benefits and limitations of AI in the field of retinal disease medical management were investigated by analyzing the most recent literature data. Furthermore, the future trends of AI involvement in ophthalmology were analyzed, as AI will be part of the decision-making regarding the scientific investigation, diagnosis and therapeutic management.

Deep Learning for Automated Sorting of Retinal Photographs

PURPOSE

Though the domain of big data and artificial intelligence in health care continues to evolve, there is a lack of systemic methods to improve data quality and streamline the preparation process. To address this, we aimed to develop an automated sorting system (RetiSort) that accurately labels the type and laterality of retinal photographs.

DESIGN

Cross-sectional study.

PARTICIPANTS

RetiSort was developed with retinal photographs from the Singapore Epidemiology of Eye Diseases (SEED) study.

METHODS

The development of RetiSort was composed of 3 steps: 2 deep-learning (DL) algorithms and 1 rule-based classifier. For step 1, a DL algorithm was developed to locate the optic disc, the "landmark feature." For step 2, based on the location of the optic disc derived from step 1, a rule-based classifier was developed to sort retinal photographs into 3 types: macular-centered, optic disc-centered, or related to other fields. Step 2 concurrently distinguished laterality (i.e., the left or right eye) of macular-centered photographs. For step 3, an additional DL algorithm was developed to differentiate the laterality of disc-centered photographs. Via the 3 steps, RetiSort sorted and labeled retinal images into (1) right macular-centered, (2) left macular-centered, (3) right optic disc-centered, (4) left optic disc-centered, and (5) images relating to other fields. Subsequently, the accuracy of RetiSort was evaluated on 5000 randomly selected retinal images from SEED as well as on 3 publicly available image databases (DIARETDB0, HEI-MED, and Drishti-GS). The main outcome measure was the accuracy for sorting of retinal photographs.

RESULTS

RetiSort mislabeled 48 out of 5000 retinal images from SEED, representing an overall accuracy of 99.0% (95% confidence interval [CI], 98.7-99.3). In external tests, RetiSort mislabeled 1, 0, and 2 images, respectively, from DIARETDB0, HEI-MED, and Drishti-GS, representing an accuracy of 99.2% (95% CI, 95.8-99.9), 100%, and 98.0% (95% CI, 93.1-99.8), respectively. Saliency maps consistently showed that the DL algorithm in step 3 required pixels in the central left lateral border and optic disc of optic disc-centered retinal photographs to differentiate the laterality.

CONCLUSIONS

RetiSort is a highly accurate automated sorting system. It can aid in data preparation and has practical applications in DL research that uses retinal photographs.

Developing a pediatric ophthalmology telemedicine program in the COVID-19 crisis

PURPOSE

To describe our methodology for implementing synchronous telemedicine during the 2019 novel coronavirus (COVID-19) pandemic.

METHODS

A retrospective review of outpatient records at a single children's hospital from March 21 to April 10, 2020, was carried out to determine the outcome of already-scheduled face-to-face outpatient appointments. The week leading up to the March 21, all appointments in the study period were categorized as follows: (1) requiring an in-person visit, (2) face-to-face visit that could be postponed, and (3) consultation required but could be virtual. Teams of administrators, schedulers, and ophthalmic technicians used defined scripts and standardized emails to communicate results of categorization to patients. Flowcharts were devised to schedule and implement telemedicine visits. Informational videos were made accessible on social media to prepare patients for the telemedicine experience. Simultaneously our children's hospital launched a pediatric on-demand e-consult service, the data analytics of which could be used to determine how many visits were eye related.

RESULTS

A total of 237 virtual ophthalmology consult visits were offered during the study period: 212 were scheduled, and 206 were completed, of which 43 were with new patients and 163 with returning patients. Following the initial virtual visit, another was required on average in 4 weeks by 21 patients; in-person follow-up was required for 170 patients on average 4.6 months after the initial virtual visit. None needed review within 72 hours. The pediatric on-demand service completed 290 visits, of which 25 had eye complaints.

CONCLUSIONS

With proper materials, technology, and staffing, a telemedicine strategy based on three patient categories can be rapidly implemented to provide continued patient care during pandemic conditions. In our study cohort, the scheduled clinic e-visits had a low no-show rate (3%), and 8% of the on-demand virtual access for pediatric care was eye related.

Workforce Shortage for Retinopathy of Prematurity Care and Emerging Role of Telehealth and Artificial Intelligence

Retinopathy of prematurity (ROP) is the leading cause of childhood blindness in very-low-birthweight and very preterm infants in the United States. With improved survival of smaller babies, more infants are at risk for ROP, yet there is an increasing shortage of providers to screen and treat ROP. Through a literature review of new and emerging technologies, screening criteria, and analysis of a national survey of pediatric ophthalmologists and retinal specialists, the authors found the shortage of ophthalmology workforce for ROP a serious and growing concern. When used appropriately, emerging technologies have the potential to mitigate gaps in the ROP workforce.

Automated identification of retinopathy of prematurity by image-based deep learning

BACKGROUND

Retinopathy of prematurity (ROP) is a leading cause of childhood blindness worldwide but can be a treatable retinal disease with appropriate and timely diagnosis. This study was performed to develop a robust intelligent system based on deep learning to automatically classify the severity of ROP from fundus images and detect the stage of ROP and presence of plus disease to enable automated diagnosis and further treatment.

METHODS

A total of 36,231 fundus images were labeled by 13 licensed retinal experts. A 101-layer convolutional neural network (ResNet) and a faster region-based convolutional neural network (Faster-RCNN) were trained for image classification and identification. We applied a 10-fold cross-validation method to train and optimize our algorithms. The accuracy, sensitivity, and specificity were assessed in a four-degree classification task to evaluate the performance of the intelligent system. The performance of the system was compared with results obtained by two retinal experts. Moreover, the system was designed to detect the stage of ROP and presence of plus disease as well as to highlight lesion regions based on an object detection network using Faster-RCNN.

RESULTS

The system achieved an accuracy of 0.903 for the ROP severity classification. Specifically, the accuracies in discriminating normal, mild, semi-urgent, and urgent were 0.883, 0.900, 0.957, and 0.870, respectively; the corresponding accuracies of the two experts were 0.902 and 0.898. Furthermore, our model achieved an accuracy of 0.957 for detecting the stage of ROP and 0.896 for detecting plus disease; the accuracies in discriminating stage I to stage V were 0.876, 0.942, 0.968, 0.998 and 0.999, respectively.

CONCLUSIONS

Our system was able to detect ROP and differentiate four-level classification fundus images with high accuracy and specificity. The performance of the system was comparable to or better than that of human experts, demonstrating that this system could be used to support clinical decisions.

Artificial intelligence for retinopathy of prematurity

PURPOSE OF REVIEW

In this article, we review the current state of artificial intelligence applications in retinopathy of prematurity (ROP) and provide insight on challenges as well as strategies for bringing these algorithms to the bedside.

RECENT FINDINGS

In the past few years, there has been a dramatic shift from machine learning approaches based on feature extraction to 'deep' convolutional neural networks for artificial intelligence applications. Several artificial intelligence for ROP approaches have demonstrated adequate proof-of-concept performance in research studies. The next steps are to determine whether these algorithms are robust to variable clinical and technical parameters in practice. Integration of artificial intelligence into ROP screening and treatment is limited by generalizability of the algorithms to maintain performance on unseen data and integration of artificial intelligence technology into new or existing clinical workflows.

SUMMARY

Real-world implementation of artificial intelligence for ROP diagnosis will require massive efforts targeted at developing standards for data acquisition, true external validation, and demonstration of feasibility. We must now focus on ethical, technical, clinical, regulatory, and financial considerations to bring this technology to the infant bedside to realize the promise offered by this technology to reduce preventable blindness from ROP.

Application of Artificial Intelligence in Targeting Retinal Diseases

Retinal diseases affect an increasing number of patients worldwide because of the aging population. Request for diagnostic imaging in ophthalmology is ramping up, while the number of specialists keeps shrinking. Cutting-edge technology embedding artificial intelligence (AI) algorithms are thus advocated to help ophthalmologists perform their clinical tasks as well as to provide a source for the advancement of novel biomarkers. In particular, optical coherence tomography (OCT) evaluation of the retina can be augmented by algorithms based on machine learning and deep learning to early detect, qualitatively localize and quantitatively measure epi/intra/subretinal abnormalities or pathological features of macular or neural diseases. In this paper, we discuss the use of AI to facilitate efficacy and accuracy of retinal imaging in those diseases increasingly treated by intravitreal vascular endothelial growth factor (VEGF) inhibitors (i.e. anti-VEGF drugs), also including integration and interpretation features in the process. We review recent advances by AI in diabetic retinopathy, age-related macular degeneration, and retinopathy of prematurity that envision a potentially key role of highly automated systems in screening, early diagnosis, grading and individualized therapy. We discuss benefits and critical aspects of automating the evaluation of disease activity, recurrences, the timing of retreatment and therapeutically potential novel targets in ophthalmology. The impact of massive employment of AI to optimize clinical assistance and encourage tailored therapies for distinct patterns of retinal diseases is also discussed.

Convolutional Neural Network Based on Fluorescein Angiography Images for Retinopathy of Prematurity Management

Purpose

The purpose of this study was to explore the use of fluorescein angiography (FA) images in a convolutional neural network (CNN) in the management of retinopathy of prematurity (ROP).

Methods

The dataset involved a total of 835 FA images of 149 eyes (90 patients), where each eye was associated with a binary outcome (57 “untreated” eyes and 92 “treated”; 308 “untreated” images, 527 “treated”). The resolution of the images was 1600 and 1200 px in 20% of cases, whereas the remaining 80% had a resolution of 640 and 480 px. All the images were resized to 640 and 480 px before training and no other preprocessing was applied. A CNN with four convolutional layers was trained on 90% of the images (n = 752) randomly chosen. The accuracy of the prediction was assessed on the remaining 10% of images (n = 83). Keras version 2.2.0 for R with Tensorflow backend version 1.11.0 was used for the analysis.

Results

The validation accuracy after 100 epochs was 0.88, whereas training accuracy was 0.97. The receiver operating characteristic (ROC) presented an area under the curve (AUC) of 0.91.

Conclusions

Our study showed, we believe for the first time, the applicability of artificial intelligence (CNN) technology in the ROP management driven by FA. Further studies are needed to exploit different fields of applications of this technology.

Translational Relevance

This algorithm is the basis for a system that could be applied to both ROP as well as experimental oxygen induced retinopathy.

Classification of pachychoroid disease on ultrawide-field indocyanine green angiography using auto-machine learning platform

AIMS

Automatic identification of pachychoroid maybe used as an adjunctive method to confirm the condition and be of help in treatment for macular diseases. This study investigated the feasibility of classifying pachychoroid disease on ultra-widefield indocyanine green angiography (UWF ICGA) images using an automated machine-learning platform.

METHODS

Two models were trained with a set including 783 UWF ICGA images of patients with pachychoroid (n=376) and non-pachychoroid (n=349) diseases using the AutoML Vision (Google). Pachychoroid was confirmed using quantitative and qualitative choroidal morphology on multimodal imaging by two retina specialists. Model 1 used the original and Model 2 used images of the left eye horizontally flipped to the orientation of the right eye to increase accuracy by equalising the mirror image of the right eye and left eye. The performances were compared with those of human experts.

RESULTS

In total, 284, 279 and 220 images of central serous chorioretinopathy, polypoidal choroidal vasculopathy and neovascular age-related maculopathy were included. The precision and recall were 87.84% and 87.84% for Model 1 and 89.19% and 89.19% for Model 2, which were comparable to the results of the retinal specialists (90.91% and 95.24%) and superior to those of ophthalmic residents (68.18% and 92.50%).

CONCLUSIONS

Auto machine-learning platform can be used in the classification of pachychoroid on UWF ICGA images after careful consideration for pachychoroid definition and limitation of the platform including unstable performance on the medical image.

Deep learning algorithms for detection of diabetic retinopathy in retinal fundus photographs: A systematic review and meta-analysis

BACKGROUND

Diabetic retinopathy (DR) is one of the leading causes of blindness globally. Earlier detection and timely treatment of DR are desirable to reduce the incidence and progression of vision loss. Currently, deep learning (DL) approaches have offered better performance in detecting DR from retinal fundus images. We, therefore, performed a systematic review with a meta-analysis of relevant studies to quantify the performance of DL algorithms for detecting DR.

METHODS

A systematic literature search on EMBASE, PubMed, Google Scholar, Scopus was performed between January 1, 2000, and March 31, 2019. The search strategy was based on the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) reporting guidelines, and DL-based study design was mandatory for articles inclusion. Two independent authors screened abstracts and titles against inclusion and exclusion criteria. Data were extracted by two authors independently using a standard form and the Quality Assessment of Diagnostic Accuracy Studies (QUADAS-2) tool was used for the risk of bias and applicability assessment.

RESULTS

Twenty-three studies were included in the systematic review; 20 studies met inclusion criteria for the meta-analysis. The pooled area under the receiving operating curve (AUROC) of DR was 0.97 (95%CI: 0.95-0.98), sensitivity was 0.83 (95%CI: 0.83-0.83), and specificity was 0.92 (95%CI: 0.92-0.92). The positive- and negative-likelihood ratio were 14.11 (95%CI: 9.91-20.07), and 0.10 (95%CI: 0.07-0.16), respectively. Moreover, the diagnostic odds ratio for DL models was 136.83 (95%CI: 79.03-236.93). All the studies provided a DR-grading scale, a human grader (e.g. trained caregivers, ophthalmologists) as a reference standard.

CONCLUSION

The findings of our study showed that DL algorithms had high sensitivity and specificity for detecting referable DR from retinal fundus photographs. Applying a DL-based automated tool of assessing DR from color fundus images could provide an alternative solution to reduce misdiagnosis and improve workflow. A DL-based automated tool offers substantial benefits to reduce screening costs, accessibility to healthcare and ameliorate earlier treatments.

The American Society of Retina Specialists Artificial Intelligence Task Force Report

Artificial intelligence (AI) is a growing area that relies on the heavy use of diagnostic imaging within the field of retina to offer exciting advancements in diagnostic capability to better understand and manage retinal conditions such as diabetic retinopathy, diabetic macular edema, age-related macular degeneration, and retinopathy of prematurity. However, there are discrepancies between the findings of these AI programs and their referral recommendations compared with evidence-based referral patterns, such as Preferred Practice Patterns by the American Academy of Ophthalmology. The overall focus of this task force report is to first describe the work in AI being completed in the management of retinal conditions. This report also discusses the guidelines of the Preferred Practice Pattern and how they can be used in the emerging field of AI.

Smartphone-Based Fundus Imaging-Where Are We Now?

With the advent of smartphone-based fundus imaging (SBFI), a low-cost alternative to conventional digital fundus photography has become available. SBFI allows for a mobile fundus examination, is applicable both with and without pupil dilation, comes with built-in connectivity and post-processing capabilities, and is relatively easy to master. Furthermore, it is delegable to paramedical staff/technicians and, hence, suitable for telemedicine. Against this background a variety of SBFI applications have become available including screening for diabetic retinopathy, glaucoma, and retinopathy of prematurity and its applications in emergency medicine and pediatrics. In addition, SBFI is convenient for teaching purposes and might serve as a surrogate for direct ophthalmoscopy. First wide-field montage techniques are available and the combination of SBFI with machine learning algorithms for image analyses is promising. In conclusion, SBFI has the potential to make fundus examinations and screenings for patients particularly in low- and middle-income settings more accessible and, therefore, aid tackling the burden of diabetic retinopathy, glaucoma, and retinopathy of prematurity screening. However, image quality for SBFI varies substantially and a reference standard for grading appears prudent. In addition, there is a strong need for comparison of different SBFI approaches in terms of applicability to disease screening and cost-effectiveness.

Deployment of Artificial Intelligence in Real-World Practice: Opportunity and Challenge

Artificial intelligence has rapidly evolved from the experimental phase to the implementation phase in many image-driven clinical disciplines, including ophthalmology. A combination of the increasing availability of large datasets and computing power with revolutionary progress in deep learning has created unprecedented opportunities for major breakthrough improvements in the performance and accuracy of automated diagnoses that primarily focus on image recognition and feature detection. Such an automated disease classification would significantly improve the accessibility, efficiency, and cost-effectiveness of eye care systems where it is less dependent on human input, potentially enabling diagnosis to be cheaper, quicker, and more consistent. Although this technology will have a profound impact on clinical flow and practice patterns sooner or later, translating such a technology into clinical practice is challenging and requires similar levels of accountability and effectiveness as any new medication or medical device due to the potential problems of bias, and ethical, medical, and legal issues that might arise. The objective of this review is to summarize the opportunities and challenges of this transition and to facilitate the integration of artificial intelligence (AI) into routine clinical practice based on our best understanding and experience in this area.

Explainable end-to-end deep learning for diabetic retinopathy detection across multiple datasets

Purpose: Diabetic retinopathy (DR) is characterized by retinal lesions affecting people having diabetes for several years. It is one of the leading causes of visual impairment worldwide. To diagnose this disease, ophthalmologists need to manually analyze retinal fundus images. Computer-aided diagnosis systems can help alleviate this burden by automatically detecting DR on retinal images, thus saving physicians' precious time and reducing costs. The objective of this study is to develop a deep learning algorithm capable of detecting DR on retinal fundus images. Nine public datasets and more than 90,000 images are used to assess the efficiency of the proposed technique. In addition, an explainability algorithm is developed to visually show the DR signs detected by the deep model. Approach: The proposed deep learning algorithm fine-tunes a pretrained deep convolutional neural network for DR detection. The model is trained on a subset of EyePACS dataset using a cosine annealing strategy for decaying the learning rate with warm up, thus improving the training accuracy. Tests are conducted on the nine datasets. An explainability algorithm based on gradient-weighted class activation mapping is developed to visually show the signs selected by the model to classify the retina images as DR. Result: The proposed network leads to higher classification rates with an area under curve (AUC) of 0.986, sensitivity = 0.958, and specificity = 0.971 for EyePACS. For MESSIDOR, MESSIDOR-2, DIARETDB0, DIARETDB1, STARE, IDRID, E-ophtha, and UoA-DR, the AUC is 0.963, 0.979, 0.986, 0.988, 0.964, 0.957, 0.984, and 0.990, respectively. Conclusions: The obtained results achieve state-of-the-art performance and outperform past published works relying on training using only publicly available datasets. The proposed approach can robustly classify fundus images and detect DR. An explainability model was developed and showed that our model was able to efficiently identify different signs of DR and detect this health issue.

Diving Deep into Deep Learning: An Update on Artificial Intelligence in Retina

Purpose of Review

In the present article, we will provide an understanding and review of artificial intelligence in the subspecialty of retina and its potential applications within the specialty.

Recent Findings

Given the significant use of diagnostic imaging within retina, this subspecialty is a fitting area for the incorporation of artificial intelligence. Researchers have aimed at creating models to assist in the diagnosis and management of retinal disease as well as in the prediction of disease course and treatment response. Most of this work thus far has focused on diabetic retinopathy, age-related macular degeneration, and retinopathy of prematurity, although other retinal diseases have started to be explored as well.

Summary

Artificial intelligence is well-suited to transform the practice of ophthalmology. A basic understanding of the technology is important for its effective implementation and growth.

A Retrospective Comparison of Deep Learning to Manual Annotations for Optic Disc and Optic Cup Segmentation in Fundus Photographs

Purpose

Optic disc (OD) and optic cup (OC) segmentation are fundamental for fundus image analysis. Manual annotation is time consuming, expensive, and highly subjective, whereas an automated system is invaluable to the medical community. The aim of this study is to develop a deep learning system to segment OD and OC in fundus photographs, and evaluate how the algorithm compares against manual annotations.

Methods

A total of 1200 fundus photographs with 120 glaucoma cases were collected. The OD and OC annotations were labeled by seven licensed ophthalmologists, and glaucoma diagnoses were based on comprehensive evaluations of the subject medical records. A deep learning system for OD and OC segmentation was developed. The performances of segmentation and glaucoma discriminating based on the cup-to-disc ratio (CDR) of automated model were compared against the manual annotations.

Results

The algorithm achieved an OD dice of 0.938 (95% confidence interval [CI] = 0.934–0.941), OC dice of 0.801 (95% CI = 0.793–0.809), and CDR mean absolute error (MAE) of 0.077 (95% CI = 0.073 mean absolute error (MAE)0.082). For glaucoma discriminating based on CDR calculations, the algorithm obtained an area under receiver operator characteristic curve (AUC) of 0.948 (95% CI = 0.920 mean absolute error (MAE)0.973), with a sensitivity of 0.850 (95% CI = 0.794–0.923) and specificity of 0.853 (95% CI = 0.798–0.918).

Conclusions

We demonstrated the potential of the deep learning system to assist ophthalmologists in analyzing OD and OC segmentation and discriminating glaucoma from nonglaucoma subjects based on CDR calculations.

Translational Relevance

We investigate the segmentation of OD and OC by deep learning system compared against the manual annotations.

Evaluation of artificial intelligence-based telemedicine screening for retinopathy of prematurity

Retrospective evaluation of a deep learning-derived retinopathy of prematurity (ROP) vascular severity score in an operational ROP screening program demonstrated high diagnostic performance for detection of type 2 or worse ROP. To our knowledge, this is the first report in the literature that evaluated the use of artificial intelligence for ROP screening and represents a proof of concept. With further prospective validation, this technology might improve the accuracy, efficiency, and objectivity of diagnosis and facilitate earlier detection of disease progression in patients with potentially blinding ROP.