Application of Comprehensive Artificial intelligence Retinal Expert (CARE) system: a national real-world evidence study

Trends and hotspots in the field of diabetic retinopathy imaging research from 2000-2023

Background

Diabetic retinopathy (DR) poses a major threat to diabetic patients' vision and is a critical public health issue. Imaging applications for DR have grown since the 21st century, aiding diagnosis, grading, and screening. This study uses bibliometric analysis to assess the field's advancements and key areas of interest.

Methods

This study performed a bibliometric analysis of DR imaging articles collected from the Web of Science Core Collection database between January 1st, 2000, and December 31st, 2023. The literature information was then analyzed through CiteSpace.

Results

The United States and China led in the number of publications, with 719 and 609, respectively. The University of London topped the institution list with 139 papers. Tien Yin Wong was the most prolific researcher. Invest. Ophthalmol. Vis. Sci. published the most articles (105). Notable burst keywords were "deep learning," "artificial intelligence," et al.

Conclusion

The United States is at the forefront of DR research, with the University of London as the top institution and Invest. Ophthalmol. Vis. Sci. as the most published journal. Tien Yin Wong is the most influential researcher. Hotspots like "deep learning," and "artificial intelligence," have seen a significant rise, indicating artificial intelligence's growing role in DR imaging.

Empowering Portable Age-Related Macular Degeneration Screening: Evaluation of a Deep Learning Algorithm for a Smartphone Fundus Camera

OBJECTIVES

Despite global research on early detection of age-related macular degeneration (AMD), not enough is being done for large-scale screening. Automated analysis of retinal images captured via smartphone presents a potential solution; however, to our knowledge, such an artificial intelligence (AI) system has not been evaluated. The study aimed to assess the performance of an AI algorithm in detecting referable AMD on images captured on a portable fundus camera.

DESIGN, SETTING

A retrospective image database from the Age-Related Eye Disease Study (AREDS) and target device was used.

PARTICIPANTS

The algorithm was trained on two distinct data sets with macula-centric images: initially on 108,251 images (55% referable AMD) from AREDS and then fine-tuned on 1108 images (33% referable AMD) captured on Asian eyes using the target device. The model was designed to indicate the presence of referable AMD (intermediate and advanced AMD). Following the first training step, the test set consisted of 909 images (49% referable AMD). For the fine-tuning step, the test set consisted of 238 (34% referable AMD) images. The reference standard for the AREDS data set was fundus image grading by the central reading centre, and for the target device, it was consensus image grading by specialists.

OUTCOME MEASURES

Area under receiver operating curve (AUC), sensitivity and specificity of algorithm.

RESULTS

Before fine-tuning, the deep learning (DL) algorithm exhibited a test set (from AREDS) sensitivity of 93.48% (95% CI: 90.8% to 95.6%), specificity of 82.33% (95% CI: 78.6% to 85.7%) and AUC of 0.965 (95% CI:0.95 to 0.98). After fine-tuning, the DL algorithm displayed a test set (from the target device) sensitivity of 91.25% (95% CI: 82.8% to 96.4%), specificity of 84.18% (95% CI: 77.5% to 89.5%) and AUC 0.947 (95% CI: 0.911 to 0.982).

CONCLUSION

The DL algorithm shows promising results in detecting referable AMD from a portable smartphone-based imaging system. This approach can potentially bring effective and affordable AMD screening to underserved areas.

Promoting smartphone-based keratitis screening using meta-learning: A multicenter study

OBJECTIVE

Keratitis is the primary cause of corneal blindness worldwide. Prompt identification and referral of patients with keratitis are fundamental measures to improve patient prognosis. Although deep learning can assist ophthalmologists in automatically detecting keratitis through a slit lamp camera, remote and underserved areas often lack this professional equipment. Smartphones, a widely available device, have recently been found to have potential in keratitis screening. However, given the limited data available from smartphones, employing traditional deep learning algorithms to construct a robust intelligent system presents a significant challenge. This study aimed to propose a meta-learning framework, cosine nearest centroid-based metric learning (CNCML), for developing a smartphone-based keratitis screening model in the case of insufficient smartphone data by leveraging the prior knowledge acquired from slit-lamp photographs.

METHODS

We developed and assessed CNCML based on 13,009 slit-lamp photographs and 4,075 smartphone photographs that were obtained from 3 independent clinical centers. To mimic real-world scenarios with various degrees of sample scarcity, we used training sets of different sizes (0 to 20 photographs per class) from the HUAWEI smartphone to train CNCML. We evaluated the performance of CNCML not only on an internal test dataset but also on two external datasets that were collected by two different brands of smartphones (VIVO and XIAOMI) in another clinical center. Furthermore, we compared the performance of CNCML with that of traditional deep learning models on these smartphone datasets. The accuracy and macro-average area under the curve (macro-AUC) were utilized to evaluate the performance of models.

RESULTS

With merely 15 smartphone photographs per class used for training, CNCML reached accuracies of 84.59%, 83.15%, and 89.99% on three smartphone datasets, with corresponding macro-AUCs of 0.96, 0.95, and 0.98, respectively. The accuracies of CNCML on these datasets were 0.56% to 9.65% higher than those of the most competitive traditional deep learning models.

CONCLUSIONS

CNCML exhibited fast learning capabilities, attaining remarkable performance with a small number of training samples. This approach presents a potential solution for transitioning intelligent keratitis detection from professional devices (e.g., slit-lamp cameras) to more ubiquitous devices (e.g., smartphones), making keratitis screening more convenient and effective.

Experiences of seeking diabetic eye care among patients with diabetes in China: a community-based convergent mixed methods study

OBJECTIVES

This study aimed to characterize the most updated utilization of eye care services and obtain a holistic understanding of barriers among patients with diabetes in China.

STUDY DESIGN

This was a convergent mixed methods study.

METHODS

A convergent triangulation mixed methods approach was used, with a quantitative cross-sectional survey of patients with diabetes and semistructured interviews involving patients and health workers. Following the conceptual framework of the World Health Organization Determinants of Health Behaviours, multivariate logistic regression for quantitative analysis and thematic analysis for qualitative data were used to examine barriers to seeking eye care among patients with diabetes. Triangulation was used to integrate quantitative and qualitative results.

RESULTS

Among 1167 surveyed patients who participated in the quantitative component, 29.1% had undergone eye examinations within the last 12 months, and 9.3% had received eye surgery. Awareness that diabetes causes eye diseases (P < 0.001) and knowing laser treatment can treat diabetic retinopathy (DR; P < 0.001) were associated with higher examination rates. In the qualitative component, involving 20 patients and 11 health workers, barriers were identified from individual, social, and cultural environmental factors. Integration of data highlighted the complex interplay of these factors in shaping care-seeking behaviors and the importance of non-economic factors, including patients' information about costs of DR services and cultural environmental factors.

CONCLUSIONS

Diabetic eye care utilization remains suboptimal in China, emphasizing the impact of cultural and contextual factors. Comprehensive education strategies, along with training for primary health workers and task-shifting, are likely to enhance eye care service utilization in underserved settings.

DeepMonitoring: a deep learning-based monitoring system for assessing the quality of cornea images captured by smartphones

Smartphone-based artificial intelligence (AI) diagnostic systems could assist high-risk patients to self-screen for corneal diseases (e.g., keratitis) instead of detecting them in traditional face-to-face medical practices, enabling the patients to proactively identify their own corneal diseases at an early stage. However, AI diagnostic systems have significantly diminished performance in low-quality images which are unavoidable in real-world environments (especially common in patient-recorded images) due to various factors, hindering the implementation of these systems in clinical practice. Here, we construct a deep learning-based image quality monitoring system (DeepMonitoring) not only to discern low-quality cornea images created by smartphones but also to identify the underlying factors contributing to the generation of such low-quality images, which can guide operators to acquire high-quality images in a timely manner. This system performs well across validation, internal, and external testing sets, with AUCs ranging from 0.984 to 0.999. DeepMonitoring holds the potential to filter out low-quality cornea images produced by smartphones, facilitating the application of smartphone-based AI diagnostic systems in real-world clinical settings, especially in the context of self-screening for corneal diseases.

The AI revolution in glaucoma: Bridging challenges with opportunities

Recent advancements in artificial intelligence (AI) herald transformative potentials for reshaping glaucoma clinical management, improving screening efficacy, sharpening diagnosis precision, and refining the detection of disease progression. However, incorporating AI into healthcare usages faces significant hurdles in terms of developing algorithms and putting them into practice. When creating algorithms, issues arise due to the intensive effort required to label data, inconsistent diagnostic standards, and a lack of thorough testing, which often limits the algorithms' widespread applicability. Additionally, the "black box" nature of AI algorithms may cause doctors to be wary or skeptical. When it comes to using these tools, challenges include dealing with lower-quality images in real situations and the systems' limited ability to work well with diverse ethnic groups and different diagnostic equipment. Looking ahead, new developments aim to protect data privacy through federated learning paradigms, improving algorithm generalizability by diversifying input data modalities, and augmenting datasets with synthetic imagery. The integration of smartphones appears promising for using AI algorithms in both clinical and non-clinical settings. Furthermore, bringing in large language models (LLMs) to act as interactive tool in medicine may signify a significant change in how healthcare will be delivered in the future. By navigating through these challenges and leveraging on these as opportunities, the field of glaucoma AI will not only have improved algorithmic accuracy and optimized data integration but also a paradigmatic shift towards enhanced clinical acceptance and a transformative improvement in glaucoma care.

Impact of Gold-Standard Label Errors on Evaluating Performance of Deep Learning Models in Diabetic Retinopathy Screening: Nationwide Real-World Validation Study

BACKGROUND

For medical artificial intelligence (AI) training and validation, human expert labels are considered the gold standard that represents the correct answers or desired outputs for a given data set. These labels serve as a reference or benchmark against which the model's predictions are compared.

OBJECTIVE

This study aimed to assess the accuracy of a custom deep learning (DL) algorithm on classifying diabetic retinopathy (DR) and further demonstrate how label errors may contribute to this assessment in a nationwide DR-screening program.

METHODS

Fundus photographs from the Lifeline Express, a nationwide DR-screening program, were analyzed to identify the presence of referable DR using both (1) manual grading by National Health Service England-certificated graders and (2) a DL-based DR-screening algorithm with validated good lab performance. To assess the accuracy of labels, a random sample of images with disagreement between the DL algorithm and the labels was adjudicated by ophthalmologists who were masked to the previous grading results. The error rates of labels in this sample were then used to correct the number of negative and positive cases in the entire data set, serving as postcorrection labels. The DL algorithm's performance was evaluated against both pre- and postcorrection labels.

RESULTS

The analysis included 736,083 images from 237,824 participants. The DL algorithm exhibited a gap between the real-world performance and the lab-reported performance in this nationwide data set, with a sensitivity increase of 12.5% (from 79.6% to 92.5%, P<.001) and a specificity increase of 6.9% (from 91.6% to 98.5%, P<.001). In the random sample, 63.6% (560/880) of negative images and 5.2% (140/2710) of positive images were misclassified in the precorrection human labels. High myopia was the primary reason for misclassifying non-DR images as referable DR images, while laser spots were predominantly responsible for misclassified referable cases. The estimated label error rate for the entire data set was 1.2%. The label correction was estimated to bring about a 12.5% enhancement in the estimated sensitivity of the DL algorithm (P<.001).

CONCLUSIONS

Label errors based on human image grading, although in a small percentage, can significantly affect the performance evaluation of DL algorithms in real-world DR screening.

Two-stage deep neural network for diagnosing fungal keratitis via in vivo confocal microscopy images

Timely and effective diagnosis of fungal keratitis (FK) is necessary for suitable treatment and avoiding irreversible vision loss for patients. In vivo confocal microscopy (IVCM) has been widely adopted to guide the FK diagnosis. We present a deep learning framework for diagnosing fungal keratitis using IVCM images to assist ophthalmologists. Inspired by the real diagnostic process, our method employs a two-stage deep architecture for diagnostic predictions based on both image-level and sequence-level information. To the best of our knowledge, we collected the largest dataset with 96,632 IVCM images in total with expert labeling to train and evaluate our method. The specificity and sensitivity of our method in diagnosing FK on the unseen test set achieved 96.65% and 97.57%, comparable or better than experienced ophthalmologists. The network can provide image-level, sequence-level and patient-level diagnostic suggestions to physicians. The results show great promise for assisting ophthalmologists in FK diagnosis.

Hybrid deep learning models for the screening of Diabetic Macular Edema in optical coherence tomography volumes

Several studies published so far used highly selective image datasets from unclear sources to train computer vision models and that may lead to overestimated results, while those studies conducted in real-life remain scarce. To avoid image selection bias, we stacked convolutional and recurrent neural networks (CNN-RNN) to analyze complete optical coherence tomography (OCT) cubes in a row and predict diabetic macular edema (DME), in a real-world diabetic retinopathy screening program. A retrospective cohort study was carried out. Throughout 4-years, 5314 OCT cubes from 4408 subjects who attended to the diabetic retinopathy (DR) screening program were included. We arranged twenty-two (22) pre-trained CNNs in parallel with a bidirectional RNN layer stacked at the bottom, allowing the model to make a prediction for the whole OCT cube. The staff of retina experts built a ground truth of DME later used to train a set of these CNN-RNN models with different configurations. For each trained CNN-RNN model, we performed threshold tuning to find the optimal cut-off point for binary classification of DME. Finally, the best models were selected according to sensitivity, specificity, and area under the receiver operating characteristics curve (AUROC) with their 95% confidence intervals (95%CI). An ensemble of the best models was also explored. 5188 cubes were non-DME and 126 were DME. Three models achieved an AUROC of 0.94. Among these, sensitivity, and specificity (95%CI) ranged from 84.1-90.5 and 89.7-93.3, respectively, at threshold 1, from 89.7-92.1 and 80-83.1 at threshold 2, and from 80.2-81 and 93.8-97, at threshold 3. The ensemble model improved these results, and lower specificity was observed among subjects with sight-threatening DR. Analysis by age, gender, or grade of DME did not vary the performance of the models. CNN-RNN models showed high diagnostic accuracy for detecting DME in a real-world setting. This engine allowed us to detect extra-foveal DMEs commonly overlooked in other studies, and showed potential for application as the first filter of non-referable patients in an outpatient center within a population-based DR screening program, otherwise ended up in specialized care.

Transparency in Artificial Intelligence Reporting in Ophthalmology-A Scoping Review

Topic

This scoping review summarizes artificial intelligence (AI) reporting in ophthalmology literature in respect to model development and validation. We characterize the state of transparency in reporting of studies prospectively validating models for disease classification.

Clinical Relevance

Understanding what elements authors currently describe regarding their AI models may aid in the future standardization of reporting. This review highlights the need for transparency to facilitate the critical appraisal of models prior to clinical implementation, to minimize bias and inappropriate use. Transparent reporting can improve effective and equitable use in clinical settings.

Methods

Eligible articles (as of January 2022) from PubMed, Embase, Web of Science, and CINAHL were independently screened by 2 reviewers. All observational and clinical trial studies evaluating the performance of an AI model for disease classification of ophthalmic conditions were included. Studies were evaluated for reporting of parameters derived from reporting guidelines (CONSORT-AI, MI-CLAIM) and our previously published editorial on model cards. The reporting of these factors, which included basic model and dataset details (source, demographics), and prospective validation outcomes, were summarized.

Results

Thirty-seven prospective validation studies were included in the scoping review. Eleven additional associated training and/or retrospective validation studies were included if this information could not be determined from the primary articles. These 37 studies validated 27 unique AI models; multiple studies evaluated the same algorithms (EyeArt, IDx-DR, and Medios AI). Details of model development were variably reported; 18 of 27 models described training dataset annotation and 10 of 27 studies reported training data distribution. Demographic information of training data was rarely reported; 7 of the 27 unique models reported age and gender and only 2 reported race and/or ethnicity. At the level of prospective clinical validation, age and gender of populations was more consistently reported (29 and 28 of 37 studies, respectively), but only 9 studies reported race and/or ethnicity data. Scope of use was difficult to discern for the majority of models. Fifteen studies did not state or imply primary users.

Conclusion

Our scoping review demonstrates variable reporting of information related to both model development and validation. The intention of our study was not to assess the quality of the factors we examined, but to characterize what information is, and is not, regularly reported. Our results suggest the need for greater transparency in the reporting of information necessary to determine the appropriateness and fairness of these tools prior to clinical use.

Financial Disclosures

Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.

Deep Learning-Based Vascular Aging Prediction From Retinal Fundus Images

Purpose

The purpose of this study was to establish and validate a deep learning model to screen vascular aging using retinal fundus images. Although vascular aging is considered a novel cardiovascular risk factor, the assessment methods are currently limited and often only available in developed regions.

Methods

We used 8865 retinal fundus images and clinical parameters of 4376 patients from two independent datasets for training a deep learning algorithm. The gold standard for vascular aging was defined as a pulse wave velocity ≥1400 cm/s. The probability of the presence of vascular aging was defined as deep learning retinal vascular aging score, the Reti-aging score. We compared the performance of the deep learning model and clinical parameters by calculating the area under the receiver operating characteristics curve (AUC). We recruited clinical specialists, including ophthalmologists and geriatricians, to assess vascular aging in patients using retinal fundus images, aiming to compare the diagnostic performance between deep learning models and clinical specialists. Finally, the potential of Reti-aging score for identifying new-onset hypertension (NH) and new-onset carotid artery plaque (NCP) in the subsequent three years was examined.

Results

The Reti-aging score model achieved an AUC of 0.826 (95% confidence interval [CI] = 0.793-0.855) and 0.779 (95% CI = 0.765-0.794) in the internal and external dataset. It showed better performance in predicting vascular aging compared with the prediction with clinical parameters. The average accuracy of ophthalmologists (66.3%) was lower than that of the Reti-aging score model, whereas geriatricians were unable to make predictions based on retinal fundus images. The Reti-aging score was associated with the risk of NH and NCP (P < 0.05).

Conclusions

The Reti-aging score model might serve as a novel method to predict vascular aging through analysis of retinal fundus images. Reti-aging score provides a novel indicator to predict new-onset cardiovascular diseases.

Translational Relevance

Given the robust performance of our model, it provides a new and reliable method for screening vascular aging, especially in undeveloped areas.

Latest developments of generative artificial intelligence and applications in ophthalmology

The emergence of generative artificial intelligence (AI) has revolutionized various fields. In ophthalmology, generative AI has the potential to enhance efficiency, accuracy, personalization and innovation in clinical practice and medical research, through processing data, streamlining medical documentation, facilitating patient-doctor communication, aiding in clinical decision-making, and simulating clinical trials. This review focuses on the development and integration of generative AI models into clinical workflows and scientific research of ophthalmology. It outlines the need for development of a standard framework for comprehensive assessments, robust evidence, and exploration of the potential of multimodal capabilities and intelligent agents. Additionally, the review addresses the risks in AI model development and application in clinical service and research of ophthalmology, including data privacy, data bias, adaptation friction, over interdependence, and job replacement, based on which we summarized a risk management framework to mitigate these concerns. This review highlights the transformative potential of generative AI in enhancing patient care, improving operational efficiency in the clinical service and research in ophthalmology. It also advocates for a balanced approach to its adoption.

Artificial Intelligence, Digital Imaging, and Robotics Technologies for Surgical Vitreoretinal Diseases

OBJECTIVE

To review recent technological advancement in imaging, surgical visualization, robotics technology, and the use of artificial intelligence in surgical vitreoretinal (VR) diseases.

BACKGROUND

Technological advancements in imaging enhance both preoperative and intraoperative management of surgical VR diseases. Widefield imaging in fundal photography and OCT can improve assessment of peripheral retinal disorders such as retinal detachments, degeneration, and tumors. OCT angiography provides a rapid and noninvasive imaging of the retinal and choroidal vasculature. Surgical visualization has also improved with intraoperative OCT providing a detailed real-time assessment of retinal layers to guide surgical decisions. Heads-up display and head-mounted display utilize 3-dimensional technology to provide surgeons with enhanced visual guidance and improved ergonomics during surgery. Intraocular robotics technology allows for greater surgical precision and is shown to be useful in retinal vein cannulation and subretinal drug delivery. In addition, deep learning techniques leverage on diverse data including widefield retinal photography and OCT for better predictive accuracy in classification, segmentation, and prognostication of many surgical VR diseases.

CONCLUSION

This review article summarized the latest updates in these areas and highlights the importance of continuous innovation and improvement in technology within the field. These advancements have the potential to reshape management of surgical VR diseases in the very near future and to ultimately improve patient care.

FINANCIAL DISCLOSURE(S)

Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.

Association of retinal age gap with chronic kidney disease and subsequent cardiovascular disease sequelae: a cross-sectional and longitudinal study from the UK Biobank

Background

Chronic kidney disease (CKD) increases the risk of cardiovascular disease (CVD) and is more prevalent in older adults. Retinal age gap, a biomarker of aging based on fundus images, has been previously developed and validated. This study aimed to investigate the association of retinal age gap with CKD and subsequent CVD complications.

Methods

A deep learning model was trained to predict the retinal age using 19 200 fundus images of 11 052 participants without any medical history at baseline. Retinal age gap, calculated as retinal age predicted minus chronological age, was calculated for the remaining 35 906 participants. Logistic regression models and Cox proportional hazards regression models were used for the association analysis.

Results

A total of 35 906 participants (56.75 ± 8.04 years, 55.68% female) were included in this study. In the cross-sectional analysis, each 1-year increase in retinal age gap was associated with a 2% increase in the risk of CKD prevalence [odds ratio 1.02, 95% confidence interval (CI) 1.01-1.04, P = .012]. A longitudinal analysis of 35 039 participants demonstrated that 2.87% of them developed CKD in follow-up, and each 1-year increase in retinal age gap was associated with a 3% increase in the risk of CKD incidence (hazard ratio 1.03, 95% CI 1.01-1.05, P = .004). In addition, a total of 111 CKD patients (15.81%) developed CVD in follow-up, and each 1-year increase in retinal age gap was associated with a 10% increase in the risk of incident CVD (hazard ratio 1.10, 95% CI 1.03-1.17, P = .005).

Conclusions

We found that retinal age gap was independently associated with the prevalence and incidence of CKD, and also associated with CVD complications in CKD patients. This supports the use of this novel biomarker in identifying individuals at high risk of CKD and CKD patients with increased risk of CVD.

Diagnostic accuracy of artificial intelligence in detecting retinitis pigmentosa: A systematic review and meta-analysis

Retinitis pigmentosa (RP) is often undetected in its early stages. Artificial intelligence (AI) has emerged as a promising tool in medical diagnostics. Therefore, we conducted a systematic review and meta-analysis to evaluate the diagnostic accuracy of AI in detecting RP using various ophthalmic images. We conducted a systematic search on PubMed, Scopus, and Web of Science databases on December 31, 2022. We included studies in the English language that used any ophthalmic imaging modality, such as OCT or fundus photography, used any AI technologies, had at least an expert in ophthalmology as a reference standard, and proposed an AI algorithm able to distinguish between images with and without retinitis pigmentosa features. We considered the sensitivity, specificity, and area under the curve (AUC) as the main measures of accuracy. We had a total of 14 studies in the qualitative analysis and 10 studies in the quantitative analysis. In total, the studies included in the meta-analysis dealt with 920,162 images. Overall, AI showed an excellent performance in detecting RP with pooled sensitivity and specificity of 0.985 [95%CI: 0.948-0.996], 0.993 [95%CI: 0.982-0.997] respectively. The area under the receiver operating characteristic (AUROC), using a random-effect model, was calculated to be 0.999 [95%CI: 0.998-1.000; P < 0.001]. The Zhou and Dendukuri I² test revealed a low level of heterogeneity between the studies, with [I2 = 19.94%] for sensitivity and [I2 = 21.07%] for specificity. The bivariate I² [20.33%] also suggested a low degree of heterogeneity. We found evidence supporting the accuracy of AI in the detection of RP; however, the level of heterogeneity between the studies was low.

Cost-effectiveness and cost-utility of a digital technology-driven hierarchical healthcare screening pattern in China

Utilization of digital technologies for cataract screening in primary care is a potential solution for addressing the dilemma between the growing aging population and unequally distributed resources. Here, we propose a digital technology-driven hierarchical screening (DH screening) pattern implemented in China to promote the equity and accessibility of healthcare. It consists of home-based mobile artificial intelligence (AI) screening, community-based AI diagnosis, and referral to hospitals. We utilize decision-analytic Markov models to evaluate the cost-effectiveness and cost-utility of different cataract screening strategies (no screening, telescreening, AI screening and DH screening). A simulated cohort of 100,000 individuals from age 50 is built through a total of 30 1-year Markov cycles. The primary outcomes are incremental cost-effectiveness ratio and incremental cost-utility ratio. The results show that DH screening dominates no screening, telescreening and AI screening in urban and rural China. Annual DH screening emerges as the most economically effective strategy with 341 (338 to 344) and 1326 (1312 to 1340) years of blindness avoided compared with telescreening, and 37 (35 to 39) and 140 (131 to 148) years compared with AI screening in urban and rural settings, respectively. The findings remain robust across all sensitivity analyses conducted. Here, we report that DH screening is cost-effective in urban and rural China, and the annual screening proves to be the most cost-effective option, providing an economic rationale for policymakers promoting public eye health in low- and middle-income countries.

Ultra-wide field and new wide field composite retinal image registration with AI-enabled pipeline and 3D distortion correction algorithm

PURPOSE

This study aimed to compare a new Artificial Intelligence (AI) method to conventional mathematical warping in accurately overlaying peripheral retinal vessels from two different imaging devices: confocal scanning laser ophthalmoscope (cSLO) wide-field images and SLO ultra-wide field images.

METHODS

Images were captured using the Heidelberg Spectralis 55-degree field-of-view and Optos ultra-wide field. The conventional mathematical warping was performed using Random Sample Consensus-Sample and Consensus sets (RANSAC-SC). This was compared to an AI alignment algorithm based on a one-way forward registration procedure consisting of full Convolutional Neural Networks (CNNs) with Outlier Rejection (OR CNN), as well as an iterative 3D camera pose optimization process (OR CNN + Distortion Correction [DC]). Images were provided in a checkerboard pattern, and peripheral vessels were graded in four quadrants based on alignment to the adjacent box.

RESULTS

A total of 660 boxes were analysed from 55 eyes. Dice scores were compared between the three methods (RANSAC-SC/OR CNN/OR CNN + DC): 0.3341/0.4665/4784 for fold 1-2 and 0.3315/0.4494/4596 for fold 2-1 in composite images. The images composed using the OR CNN + DC have a median rating of 4 (out of 5) versus 2 using RANSAC-SC. The odds of getting a higher grading level are 4.8 times higher using our OR CNN + DC than RANSAC-SC (p < 0.0001).

CONCLUSION

Peripheral retinal vessel alignment performed better using our AI algorithm than RANSAC-SC. This may help improve co-localizing retinal anatomy and pathology with our algorithm.

Prediction of preeclampsia from retinal fundus images via deep learning in singleton pregnancies: a prospective cohort study

INTRODUCTION

Early prediction of preeclampsia (PE) is of universal importance in controlling the disease process. Our study aimed to assess the feasibility of using retinal fundus images to predict preeclampsia via deep learning in singleton pregnancies.

METHODS

This prospective cohort study was conducted at Shanghai First Maternity and Infant Hospital, Tongji University School of Medicine. Eligible participants included singleton pregnancies who presented for prenatal visits before 14 weeks of gestation from September 1, 2020, to February 1, 2022. Retinal fundus images were obtained using a nonmydriatic digital retinal camera during their initial prenatal visit upon admission before 20 weeks of gestation. In addition, we generated fundus scores, which indicated the predictive value of hypertension, using a hypertension detection model. To evaluate the predictive value of the retinal fundus image-based deep learning algorithm for preeclampsia, we conducted stratified analyses and measured the area under the curve (AUC), sensitivity, and specificity. We then conducted sensitivity analyses for validation.

RESULTS

Our study analyzed a total of 1138 women, 92 pregnancies developed into hypertension disorders of pregnancy (HDP), including 26 cases of gestational hypertension and 66 cases of preeclampsia. The adjusted odds ratio (aOR) of the fundus scores was 2.582 (95% CI, 1.883-3.616; P  < 0.001). Otherwise, in the categories of prepregnancy BMI less than 28.0 and at least 28.0, the aORs were 3.073 (95%CI, 2.265-4.244; P  < 0.001) and 5.866 (95% CI, 3.292-11.531; P  < 0.001). In the categories of maternal age less than 35.0 and at least 35.0, the aORs were 2.845 (95% CI, 1.854-4.463; P  < 0.001) and 2.884 (95% CI, 1.794-4.942; P  < 0.001). The AUC of the fundus score combined with risk factors was 0.883 (sensitivity, 0.722; specificity, 0.934; 95% CI, 0.834-0.932) for predicting preeclampsia.

CONCLUSION

Our study demonstrates that the use of deep learning algorithm-based retinal fundus images offers promising predictive value for the early detection of preeclampsia.

Evaluating the accuracy of the Ophthalmologist Robot for multiple blindness-causing eye diseases: a multicentre, prospective study protocol

INTRODUCTION

Early eye screening and treatment can reduce the incidence of blindness by detecting and addressing eye diseases at an early stage. The Ophthalmologist Robot is an automated device that can simultaneously capture ocular surface and fundus images without the need for ophthalmologists, making it highly suitable for primary application. However, the accuracy of the device's screening capabilities requires further validation. This study aims to evaluate and compare the screening accuracies of ophthalmologists and deep learning models using images captured by the Ophthalmologist Robot, in order to identify a screening method that is both highly accurate and cost-effective. Our findings may provide valuable insights into the potential applications of remote eye screening.

METHODS AND ANALYSIS

This is a multicentre, prospective study that will recruit approximately 1578 participants from 3 hospitals. All participants will undergo ocular surface and fundus images taken by the Ophthalmologist Robot. Additionally, 695 participants will have their ocular surface imaged with a slit lamp. Relevant information from outpatient medical records will be collected. The primary objective is to evaluate the accuracy of ophthalmologists' screening for multiple blindness-causing eye diseases using device images through receiver operating characteristic curve analysis. The targeted diseases include keratitis, corneal scar, cataract, diabetic retinopathy, age-related macular degeneration, glaucomatous optic neuropathy and pathological myopia. The secondary objective is to assess the accuracy of deep learning models in disease screening. Furthermore, the study aims to compare the consistency between the Ophthalmologist Robot and the slit lamp in screening for keratitis and corneal scar using the Kappa test. Additionally, the cost-effectiveness of three eye screening methods, based on non-telemedicine screening, ophthalmologist-telemedicine screening and artificial intelligence-telemedicine screening, will be assessed by constructing Markov models.

ETHICS AND DISSEMINATION

The study has obtained approval from the ethics committee of the Ophthalmology and Optometry Hospital of Wenzhou Medical University (reference: 2023-026 K-21-01). This work will be disseminated by peer-review publications, abstract presentations at national and international conferences and data sharing with other researchers.

TRIAL REGISTRATION NUMBER

ChiCTR2300070082.

Automated detection of nine infantile fundus diseases and conditions in retinal images using a deep learning system

Purpose

We developed an Infant Retinal Intelligent Diagnosis System (IRIDS), an automated system to aid early diagnosis and monitoring of infantile fundus diseases and health conditions to satisfy urgent needs of ophthalmologists.

Methods

We developed IRIDS by combining convolutional neural networks and transformer structures, using a dataset of 7697 retinal images (1089 infants) from four hospitals. It identifies nine fundus diseases and conditions, namely, retinopathy of prematurity (ROP) (mild ROP, moderate ROP, and severe ROP), retinoblastoma (RB), retinitis pigmentosa (RP), Coats disease, coloboma of the choroid, congenital retinal fold (CRF), and normal. IRIDS also includes depth attention modules, ResNet-18 (Res-18), and Multi-Axis Vision Transformer (MaxViT). Performance was compared to that of ophthalmologists using 450 retinal images. The IRIDS employed a five-fold cross-validation approach to generate the classification results.

Results

Several baseline models achieved the following metrics: accuracy, precision, recall, F1-score (F1), kappa, and area under the receiver operating characteristic curve (AUC) with best values of 94.62% (95% CI, 94.34%-94.90%), 94.07% (95% CI, 93.32%-94.82%), 90.56% (95% CI, 88.64%-92.48%), 92.34% (95% CI, 91.87%-92.81%), 91.15% (95% CI, 90.37%-91.93%), and 99.08% (95% CI, 99.07%-99.09%), respectively. In comparison, IRIDS showed promising results compared to ophthalmologists, demonstrating an average accuracy, precision, recall, F1, kappa, and AUC of 96.45% (95% CI, 96.37%-96.53%), 95.86% (95% CI, 94.56%-97.16%), 94.37% (95% CI, 93.95%-94.79%), 95.03% (95% CI, 94.45%-95.61%), 94.43% (95% CI, 93.96%-94.90%), and 99.51% (95% CI, 99.51%-99.51%), respectively, in multi-label classification on the test dataset, utilizing the Res-18 and MaxViT models. These results suggest that, particularly in terms of AUC, IRIDS achieved performance that warrants further investigation for the detection of retinal abnormalities.

Conclusions

IRIDS identifies nine infantile fundus diseases and conditions accurately. It may aid non-ophthalmologist personnel in underserved areas in infantile fundus disease screening. Thus, preventing severe complications. The IRIDS serves as an example of artificial intelligence integration into ophthalmology to achieve better outcomes in predictive, preventive, and personalized medicine (PPPM / 3PM) in the treatment of infantile fundus diseases.

Supplementary Information

The online version contains supplementary material available at 10.1007/s13167-024-00350-y.

Application of artificial intelligence system for screening multiple fundus diseases in Chinese primary healthcare settings: a real-world, multicentre and cross-sectional study of 4795 cases

BACKGROUND/AIMS

This study evaluates the performance of the Airdoc retinal artificial intelligence system (ARAS) for detecting multiple fundus diseases in real-world scenarios in primary healthcare settings and investigates the fundus disease spectrum based on ARAS.

METHODS

This real-world, multicentre, cross-sectional study was conducted in Shanghai and Xinjiang, China. Six primary healthcare settings were included in this study. Colour fundus photographs were taken and graded by ARAS and retinal specialists. The performance of ARAS is described by its accuracy, sensitivity, specificity and positive and negative predictive values. The spectrum of fundus diseases in primary healthcare settings has also been investigated.

RESULTS

A total of 4795 participants were included. The median age was 57.0 (IQR 39.0-66.0) years, and 3175 (66.2%) participants were female. The accuracy, specificity and negative predictive value of ARAS for detecting normal fundus and 14 retinal abnormalities were high, whereas the sensitivity and positive predictive value varied in detecting different abnormalities. The proportion of retinal drusen, pathological myopia and glaucomatous optic neuropathy was significantly higher in Shanghai than in Xinjiang. Moreover, the percentages of referable diabetic retinopathy, retinal vein occlusion and macular oedema in middle-aged and elderly people in Xinjiang were significantly higher than in Shanghai.

CONCLUSION

This study demonstrated the dependability of ARAS for detecting multiple retinal diseases in primary healthcare settings. Implementing the AI-assisted fundus disease screening system in primary healthcare settings might be beneficial in reducing regional disparities in medical resources. However, the ARAS algorithm must be improved to achieve better performance.

TRIAL REGISTRATION NUMBER

NCT04592068.

Economic evaluation for medical artificial intelligence: accuracy vs. cost-effectiveness in a diabetic retinopathy screening case

Artificial intelligence (AI) models have shown great accuracy in health screening. However, for real-world implementation, high accuracy may not guarantee cost-effectiveness. Improving AI's sensitivity finds more high-risk patients but may raise medical costs while increasing specificity reduces unnecessary referrals but may weaken detection capability. To evaluate the trade-off between AI model performance and the long-running cost-effectiveness, we conducted a cost-effectiveness analysis in a nationwide diabetic retinopathy (DR) screening program in China, comprising 251,535 participants with diabetes over 30 years. We tested a validated AI model in 1100 different diagnostic performances (presented as sensitivity/specificity pairs) and modeled annual screening scenarios. The status quo was defined as the scenario with the most accurate AI performance. The incremental cost-effectiveness ratio (ICER) was calculated for other scenarios against the status quo as cost-effectiveness metrics. Compared to the status quo (sensitivity/specificity: 93.3%/87.7%), six scenarios were cost-saving and seven were cost-effective. To achieve cost-saving or cost-effective, the AI model should reach a minimum sensitivity of 88.2% and specificity of 80.4%. The most cost-effective AI model exhibited higher sensitivity (96.3%) and lower specificity (80.4%) than the status quo. In settings with higher DR prevalence and willingness-to-pay levels, the AI needed higher sensitivity for optimal cost-effectiveness. Urban regions and younger patient groups also required higher sensitivity in AI-based screening. In real-world DR screening, the most accurate AI model may not be the most cost-effective. Cost-effectiveness should be independently evaluated, which is most likely to be affected by the AI's sensitivity.

DeepGraFT: A novel semantic segmentation auxiliary ROI-based deep learning framework for effective fundus tessellation classification

Fundus tessellation (FT) is a prevalent clinical feature associated with myopia and has implications in the development of myopic maculopathy, which causes irreversible visual impairment. Accurate classification of FT in color fundus photo can help predict the disease progression and prognosis. However, the lack of precise detection and classification tools has created an unmet medical need, underscoring the importance of exploring the clinical utility of FT. Thus, to address this gap, we introduce an automatic FT grading system (called DeepGraFT) using classification-and-segmentation co-decision models by deep learning. ConvNeXt, utilizing transfer learning from pretrained ImageNet weights, was employed for the classification algorithm, aligning with a region of interest based on the ETDRS grading system to boost performance. A segmentation model was developed to detect FT exits, complementing the classification for improved grading accuracy. The training set of DeepGraFT was from our in-house cohort (MAGIC), and the validation sets consisted of the rest part of in-house cohort and an independent public cohort (UK Biobank). DeepGraFT demonstrated a high performance in the training stage and achieved an impressive accuracy in validation phase (in-house cohort: 86.85 %; public cohort: 81.50 %). Furthermore, our findings demonstrated that DeepGraFT surpasses machine learning-based classification models in FT classification, achieving a 5.57 % increase in accuracy. Ablation analysis revealed that the introduced modules significantly enhanced classification effectiveness and elevated accuracy from 79.85 % to 86.85 %. Further analysis using the results provided by DeepGraFT unveiled a significant negative association between FT and spherical equivalent (SE) in the UK Biobank cohort. In conclusion, DeepGraFT accentuates potential benefits of the deep learning model in automating the grading of FT and allows for potential utility as a clinical-decision support tool for predicting progression of pathological myopia.

Classification of Pulmonary Nodules in 2-[18F]FDG PET/CT Images with a 3D Convolutional Neural Network

Purpose

2-[18F]FDG PET/CT plays an important role in the management of pulmonary nodules. Convolutional neural networks (CNNs) automatically learn features from images and have the potential to improve the discrimination between malignant and benign pulmonary nodules. The purpose of this study was to develop and validate a CNN model for classification of pulmonary nodules from 2-[18F]FDG PET images.

Methods

One hundred thirteen participants were retrospectively selected. One nodule per participant. The 2-[18F]FDG PET images were preprocessed and annotated with the reference standard. The deep learning experiment entailed random data splitting in five sets. A test set was held out for evaluation of the final model. Four-fold cross-validation was performed from the remaining sets for training and evaluating a set of candidate models and for selecting the final model. Models of three types of 3D CNNs architectures were trained from random weight initialization (Stacked 3D CNN, VGG-like and Inception-v2-like models) both in original and augmented datasets. Transfer learning, from ImageNet with ResNet-50, was also used.

Results

The final model (Stacked 3D CNN model) obtained an area under the ROC curve of 0.8385 (95% CI: 0.6455-1.0000) in the test set. The model had a sensibility of 80.00%, a specificity of 69.23% and an accuracy of 73.91%, in the test set, for an optimised decision threshold that assigns a higher cost to false negatives.

Conclusion

A 3D CNN model was effective at distinguishing benign from malignant pulmonary nodules in 2-[18F]FDG PET images.

Supplementary Information

The online version contains supplementary material available at 10.1007/s13139-023-00821-6.

From Machine Learning to Patient Outcomes: A Comprehensive Review of AI in Pancreatic Cancer

Pancreatic cancer is a highly aggressive and difficult-to-detect cancer with a poor prognosis. Late diagnosis is common due to a lack of early symptoms, specific markers, and the challenging location of the pancreas. Imaging technologies have improved diagnosis, but there is still room for improvement in standardizing guidelines. Biopsies and histopathological analysis are challenging due to tumor heterogeneity. Artificial Intelligence (AI) revolutionizes healthcare by improving diagnosis, treatment, and patient care. AI algorithms can analyze medical images with precision, aiding in early disease detection. AI also plays a role in personalized medicine by analyzing patient data to tailor treatment plans. It streamlines administrative tasks, such as medical coding and documentation, and provides patient assistance through AI chatbots. However, challenges include data privacy, security, and ethical considerations. This review article focuses on the potential of AI in transforming pancreatic cancer care, offering improved diagnostics, personalized treatments, and operational efficiency, leading to better patient outcomes.

The performance of a deep learning system in assisting junior ophthalmologists in diagnosing 13 major fundus diseases: a prospective multi-center clinical trial

Artificial intelligence (AI)-based diagnostic systems have been reported to improve fundus disease screening in previous studies. This multicenter prospective self-controlled clinical trial aims to evaluate the diagnostic performance of a deep learning system (DLS) in assisting junior ophthalmologists in detecting 13 major fundus diseases. A total of 1493 fundus images from 748 patients were prospectively collected from five tertiary hospitals in China. Nine junior ophthalmologists were trained and annotated the images with or without the suggestions proposed by the DLS. The diagnostic performance was evaluated among three groups: DLS-assisted junior ophthalmologist group (test group), junior ophthalmologist group (control group) and DLS group. The diagnostic consistency was 84.9% (95%CI, 83.0% ~ 86.9%), 72.9% (95%CI, 70.3% ~ 75.6%) and 85.5% (95%CI, 83.5% ~ 87.4%) in the test group, control group and DLS group, respectively. With the help of the proposed DLS, the diagnostic consistency of junior ophthalmologists improved by approximately 12% (95% CI, 9.1% ~ 14.9%) with statistical significance (P < 0.001). For the detection of 13 diseases, the test group achieved significant higher sensitivities (72.2% ~ 100.0%) and comparable specificities (90.8% ~ 98.7%) comparing with the control group (sensitivities, 50% ~ 100%; specificities 96.7 ~ 99.8%). The DLS group presented similar performance to the test group in the detection of any fundus abnormality (sensitivity, 95.7%; specificity, 87.2%) and each of the 13 diseases (sensitivity, 83.3% ~ 100.0%; specificity, 89.0 ~ 98.0%). The proposed DLS provided a novel approach for the automatic detection of 13 major fundus diseases with high diagnostic consistency and assisted to improve the performance of junior ophthalmologists, resulting especially in reducing the risk of missed diagnoses. ClinicalTrials.gov NCT04723160.

Protocol to analyze fundus images for multidimensional quality grading and real-time guidance using deep learning techniques

Data quality issues have been acknowledged as one of the greatest obstacles in medical artificial intelligence research. Here, we present DeepFundus, which employs deep learning techniques to perform multidimensional classification of fundus image quality and provide real-time guidance for on-site image acquisition. We describe steps for data preparation, model training, model inference, model evaluation, and the visualization of results using heatmaps. This protocol can be implemented in Python using either the suggested dataset or a customized dataset. For complete details on the use and execution of this protocol, please refer to Liu et al.1.

SERIES: eHealth in primary care. Part 6: Global perspectives: Learning from eHealth for low-resource primary care settings and across high-, middle- and low-income countries

BACKGROUND

eHealth offers opportunities to improve health and healthcare systems and overcome primary care challenges in low-resource settings (LRS). LRS has been typically associated with low- and middle-income countries (LMIC), but they can be found in high-income countries (HIC) when human, physical or financial resources are constrained. Adopting a concept of LRS that applies to LMIC and HIC can facilitate knowledge interchange between eHealth initiatives while improving healthcare provision for socioeconomically disadvantaged groups across the globe.

OBJECTIVES

To outline the contributions and challenges of eHealth in low-resource primary care settings.

STRATEGY

We adopt a socio-ecological understanding of LRS, making LRS relevant to LMIC and HIC. To assess the potential of eHealth in primary care settings, we discuss four case studies according to the WHO 'building blocks for strengthening healthcare systems'.

RESULTS AND DISCUSSION

The case studies illustrate eHealth's potential to improve the provision of healthcare by i) improving the delivery of healthcare (using AI-generated chats); ii) supporting the workforce (using telemedicine platforms); iii) strengthening the healthcare information system (through patient-centred healthcare information systems), and iv) improving system-related elements of healthcare (through a mobile health financing platform). Nevertheless, we found that development and implementation are hindered by user-related, technical, financial, regulatory and evaluation challenges. We formulated six recommendations to help anticipate or overcome these challenges: 1) evaluate eHealth's appropriateness, 2) know the end users, 3) establish evaluation methods, 4) prioritise the human component, 5) profit from collaborations, ensure sustainable financing and local ownership, 6) and contextualise and evaluate the implementation strategies.

Recent advancement in integrating artificial intelligence and information technology with real-world data for clinical decision-making in China: A scoping review

OBJECTIVE

Striking innovations and advancements have been achieved with the use of artificial intelligence and healthcare information technology being integrated into clinical real-world data. The current scoping review aimed to provide an overview of the current status of artificial intelligence-/information technology-based clinical decision support tools in China.

METHODS

PubMed/MEDLINE, Embase, China National Knowledge Internet, and Wanfang data were searched for both English and Chinese literature. The gray literature search was conducted for commercially available tools. Original studies that focused on clinical decision support tools driven by artificial intelligence or information technology in China and were published between 2010 and February 2022 were included. Information extracted from each article was further synthesized by themes based on three types of clinical decision-making.

RESULTS

A total of 37 peer-reviewed publications and 13 commercially available tools were included in the final analysis. Among them, 32.0% were developed for disease diagnosis, 54.0% for risk prediction and classification, and 14.0% for disease management. Chronic diseases were the most popular therapeutic areas of exploration, with particular emphasis on cardiovascular and cerebrovascular diseases. Single-center electronic medical records were the mainstream data sources leveraged to inform clinical decision-making, with internal validation being predominately used for model evaluation.

CONCLUSIONS

To effectively promote the extensive use of real-world data and drive a paradigm shift in clinical decision-making in China, multidisciplinary collaboration of key stakeholders is urgently needed.

Deep Learning Performance of Ultra-Widefield Fundus Imaging for Screening Retinal Lesions in Rural Locales

Importance

Retinal diseases are the leading cause of irreversible blindness worldwide, and timely detection contributes to prevention of permanent vision loss, especially for patients in rural areas with limited medical resources. Deep learning systems (DLSs) based on fundus images with a 45° field of view have been extensively applied in population screening, while the feasibility of using ultra-widefield (UWF) fundus image-based DLSs to detect retinal lesions in patients in rural areas warrants exploration.

Objective

To explore the performance of a DLS for multiple retinal lesion screening using UWF fundus images from patients in rural areas.

Design, Setting, and Participants

In this diagnostic study, a previously developed DLS based on UWF fundus images was used to screen for 5 retinal lesions (retinal exudates or drusen, glaucomatous optic neuropathy, retinal hemorrhage, lattice degeneration or retinal breaks, and retinal detachment) in 24 villages of Yangxi County, China, between November 17, 2020, and March 30, 2021.

Interventions

The captured images were analyzed by the DLS and ophthalmologists.

Main Outcomes and Measures

The performance of the DLS in rural screening was compared with that of the internal validation in the previous model development stage. The image quality, lesion proportion, and complexity of lesion composition were compared between the model development stage and the rural screening stage.

Results

A total of 6222 eyes in 3149 participants (1685 women [53.5%]; mean [SD] age, 70.9 [9.1] years) were screened. The DLS achieved a mean (SD) area under the receiver operating characteristic curve (AUC) of 0.918 (0.021) (95% CI, 0.892-0.944) for detecting 5 retinal lesions in the entire data set when applied for patients in rural areas, which was lower than that reported at the model development stage (AUC, 0.998 [0.002] [95% CI, 0.995-1.000]; P < .001). Compared with the fundus images in the model development stage, the fundus images in this rural screening study had an increased frequency of poor quality (13.8% [860 of 6222] vs 0%), increased variation in lesion proportions (0.1% [6 of 6222]-36.5% [2271 of 6222] vs 14.0% [2793 of 19 891]-21.3% [3433 of 16 138]), and an increased complexity of lesion composition.

Conclusions and Relevance

This diagnostic study suggests that the DLS exhibited excellent performance using UWF fundus images as a screening tool for 5 retinal lesions in patients in a rural setting. However, poor image quality, diverse lesion proportions, and a complex set of lesions may have reduced the performance of the DLS; these factors in targeted screening scenarios should be taken into consideration in the model development stage to ensure good performance.

Real-world post-deployment performance of a novel machine learning-based digital health technology for skin lesion assessment and suggestions for post-market surveillance

Introduction

Deep Ensemble for Recognition of Malignancy (DERM) is an artificial intelligence as a medical device (AIaMD) tool for skin lesion assessment.

Methods

We report prospective real-world performance from its deployment within skin cancer pathways at two National Health Service hospitals (UK) between July 2021 and October 2022.

Results

A total of 14,500 cases were seen, including patients 18-100 years old with Fitzpatrick skin types I-VI represented. Based on 8,571 lesions assessed by DERM with confirmed outcomes, versions A and B demonstrated very high sensitivity for detecting melanoma (95.0-100.0%) or malignancy (96.0-100.0%). Benign lesion specificity was 40.7-49.4% (DERM-vA) and 70.1-73.4% (DERM-vB). DERM identified 15.0-31.0% of cases as eligible for discharge.

Discussion

We show DERM performance in-line with sensitivity targets and pre-marketing authorisation research, and it reduced the caseload for hospital specialists in two pathways. Based on our experience we offer suggestions on key elements of post-market surveillance for AIaMDs.

An artificial intelligence system for the whole process from diagnosis to treatment suggestion of ischemic retinal diseases

Ischemic retinal diseases (IRDs) are a series of common blinding diseases that depend on accurate fundus fluorescein angiography (FFA) image interpretation for diagnosis and treatment. An artificial intelligence system (Ai-Doctor) was developed to interpret FFA images. Ai-Doctor performed well in image phase identification (area under the curve [AUC], 0.991-0.999, range), diabetic retinopathy (DR) and branch retinal vein occlusion (BRVO) diagnosis (AUC, 0.979-0.992), and non-perfusion area segmentation (Dice similarity coefficient [DSC], 89.7%-90.1%) and quantification. The segmentation model was expanded to unencountered IRDs (central RVO and retinal vasculitis), with DSCs of 89.2% and 83.6%, respectively. A clinically applicable ischemia index (CAII) was proposed to evaluate ischemic degree; patients with CAII values exceeding 0.17 in BRVO and 0.08 in DR may be associated with increased possibility for laser therapy. Ai-Doctor is expected to achieve accurate FFA image interpretation for IRDs, potentially reducing the reliance on retinal specialists.

Effectiveness of an image analyzing AI-based Digital Health Technology to identify Non-Melanoma Skin Cancer and other skin lesions: results of the DERM-003 study

Introduction

Identification of skin cancer by an Artificial Intelligence (AI)-based Digital Health Technology could help improve the triage and management of suspicious skin lesions.

Methods

The DERM-003 study (NCT04116983) was a prospective, multi-center, single-arm, masked study that aimed to demonstrate the effectiveness of an AI as a Medical Device (AIaMD) to identify Squamous Cell Carcinoma (SCC), Basal Cell Carcinoma (BCC), pre-malignant and benign lesions from dermoscopic images of suspicious skin lesions. Suspicious skin lesions that were suitable for photography were photographed with 3 smartphone cameras (iPhone 6S, iPhone 11, Samsung 10) with a DL1 dermoscopic lens attachment. Dermatologists provided clinical diagnoses and histopathology results were obtained for biopsied lesions. Each image was assessed by the AIaMD and the output compared to the ground truth diagnosis.

Results

572 patients (49.5% female, mean age 68.5 years, 96.9% Fitzpatrick skin types I-III) were recruited from 4 UK NHS Trusts, providing images of 611 suspicious lesions. 395 (64.6%) lesions were biopsied; 47 (11%) were diagnosed as SCC and 184 (44%) as BCC. The AIaMD AUROC on images taken by iPhone 6S was 0.88 (95% CI: 0.83-0.93) for SCC and 0.87 (95% CI: 0.84-0.91) for BCC. For Samsung 10 the AUROCs were 0.85 (95% CI: 0.79-0.90) and 0.87 (95% CI, 0.83-0.90), and for the iPhone 11 they were 0.88 (95% CI, 0.84-0.93) and 0.89 (95% CI, 0.86-0.92) for SCC and BCC, respectively. Using pre-determined diagnostic thresholds on images taken on the iPhone 6S the AIaMD achieved a sensitivity and specificity of 98% (95% CI, 88-100%) and 38% (95% CI, 33-44%) for SCC; and 94% (95% CI, 90-97%) and 28% (95 CI, 21-35%) for BCC. All 16 lesions diagnosed as melanoma in the study were correctly classified by the AIaMD.

Discussion

The AIaMD has the potential to support the timely diagnosis of malignant and premalignant skin lesions.

Artificial intelligence for diabetic retinopathy in low-income and middle-income countries: a scoping review

Diabetic retinopathy (DR) is a leading cause of blindness globally. There is growing evidence to support the use of artificial intelligence (AI) in diabetic eye care, particularly for screening populations at risk of sight loss from DR in low-income and middle-income countries (LMICs) where resources are most stretched. However, implementation into clinical practice remains limited. We conducted a scoping review to identify what AI tools have been used for DR in LMICs and to report their performance and relevant characteristics. 81 articles were included. The reported sensitivities and specificities were generally high providing evidence to support use in clinical practice. However, the majority of studies focused on sensitivity and specificity only and there was limited information on cost, regulatory approvals and whether the use of AI improved health outcomes. Further research that goes beyond reporting sensitivities and specificities is needed prior to wider implementation.

Deep learning for detecting visually impaired cataracts using fundus images

Purpose: To develop a visual function-based deep learning system (DLS) using fundus images to screen for visually impaired cataracts. Materials and methods: A total of 8,395 fundus images (5,245 subjects) with corresponding visual function parameters collected from three clinical centers were used to develop and evaluate a DLS for classifying non-cataracts, mild cataracts, and visually impaired cataracts. Three deep learning algorithms (DenseNet121, Inception V3, and ResNet50) were leveraged to train models to obtain the best one for the system. The performance of the system was evaluated using the area under the receiver operating characteristic curve (AUC), sensitivity, and specificity. Results: The AUC of the best algorithm (DenseNet121) on the internal test dataset and the two external test datasets were 0.998 (95% CI, 0.996-0.999) to 0.999 (95% CI, 0.998-1.000),0.938 (95% CI, 0.924-0.951) to 0.966 (95% CI, 0.946-0.983) and 0.937 (95% CI, 0.918-0.953) to 0.977 (95% CI, 0.962-0.989), respectively. In the comparison between the system and cataract specialists, better performance was observed in the system for detecting visually impaired cataracts (p < 0.05). Conclusion: Our study shows the potential of a function-focused screening tool to identify visually impaired cataracts from fundus images, enabling timely patient referral to tertiary eye hospitals.

Artificial intelligence in ophthalmology: The path to the real-world clinic

Artificial intelligence (AI) has great potential to transform healthcare by enhancing the workflow and productivity of clinicians, enabling existing staff to serve more patients, improving patient outcomes, and reducing health disparities. In the field of ophthalmology, AI systems have shown performance comparable with or even better than experienced ophthalmologists in tasks such as diabetic retinopathy detection and grading. However, despite these quite good results, very few AI systems have been deployed in real-world clinical settings, challenging the true value of these systems. This review provides an overview of the current main AI applications in ophthalmology, describes the challenges that need to be overcome prior to clinical implementation of the AI systems, and discusses the strategies that may pave the way to the clinical translation of these systems.

Counteracting Data Bias and Class Imbalance-Towards a Useful and Reliable Retinal Disease Recognition System

Multiple studies presented satisfactory performances for the treatment of various ocular diseases. To date, there has been no study that describes a multiclass model, medically accurate, and trained on large diverse dataset. No study has addressed a class imbalance problem in one giant dataset originating from multiple large diverse eye fundus image collections. To ensure a real-life clinical environment and mitigate the problem of biased medical image data, 22 publicly available datasets were merged. To secure medical validity only Diabetic Retinopathy (DR), Age-Related Macular Degeneration (AMD) and Glaucoma (GL) were included. The state-of-the-art models ConvNext, RegNet and ResNet were utilized. In the resulting dataset, there were 86,415 normal, 3787 GL, 632 AMD and 34,379 DR fundus images. ConvNextTiny achieved the best results in terms of recognizing most of the examined eye diseases with the most metrics. The overall accuracy was 80.46 ± 1.48. Specific accuracy values were: 80.01 ± 1.10 for normal eye fundus, 97.20 ± 0.66 for GL, 98.14 ± 0.31 for AMD, 80.66 ± 1.27 for DR. A suitable screening model for the most prevalent retinal diseases in ageing societies was designed. The model was developed on a diverse, combined large dataset which made the obtained results less biased and more generalizable.

Development and evaluation of an artificial intelligence system for children intussusception diagnosis using ultrasound images

Accurate identification of intussusception in children is critical for timely non-surgical management. We propose an end-to-end artificial intelligence algorithm, the Children Intussusception Diagnosis Network (CIDNet) system, that utilizes ultrasound images to rapidly diagnose intussusception. 9999 ultrasound images of 4154 pediatric patients were divided into training, validation, test, and independent reader study datasets. The independent reader study cohort was used to compare the diagnostic performance of the CIDNet system to six radiologists. Performance was evaluated using, among others, balance accuracy (BACC) and area under the receiver operating characteristic curve (AUC). The CIDNet system performed the best in diagnosing intussusception with a BACC of 0.8464 and AUC of 0.9716 in the test dataset compared to other deep learning algorithms. The CIDNet system compared favorably with expert radiologists by outstanding identification performance and robustness (BACC:0.9297; AUC:0.9769). CIDNet is a stable and precise technological tool for identifying intussusception in ultrasound scans of children.

Deep Learning for the Detection of Multiple Fundus Diseases Using Ultra-widefield Images

INTRODUCTION

To design and evaluate a deep learning model based on ultra-widefield images (UWFIs) that can detect several common fundus diseases.

METHODS

Based on 4574 UWFIs, a deep learning model was trained and validated that can identify normal fundus and eight common fundus diseases, namely referable diabetic retinopathy, retinal vein occlusion, pathologic myopia, retinal detachment, retinitis pigmentosa, age-related macular degeneration, vitreous opacity, and optic neuropathy. The model was tested on three test sets with data volumes of 465, 979, and 525. The performance of the three deep learning networks, EfficientNet-B7, DenseNet, and ResNet-101, was evaluated on the internal test set. Additionally, we compared the performance of the deep learning model with that of doctors in a tertiary referral hospital.

RESULTS

Compared to the other two deep learning models, EfficientNet-B7 achieved the best performance. The area under the receiver operating characteristic curves of the EfficientNet-B7 model on the internal test set, external test set A and external test set B were 0.9708 (0.8772, 0.9849) to 1.0000 (1.0000, 1.0000), 0.9683 (0.8829, 0.9770) to 1.0000 (0.9975, 1.0000), and 0.8919 (0.7150, 0.9055) to 0.9977 (0.9165, 1.0000), respectively. On a data set of 100 images, the total accuracy of the deep learning model was 93.00%, the average accuracy of three ophthalmologists who had been working for 2 years and three ophthalmologists who had been working in fundus imaging for more than 5 years was 88.00% and 94.00%, respectively.

CONCLUSION

High performance was achieved on all three test sets using our UWFI multidisease classification model with a small sample size and fast model inference. The performance of the artificial intelligence model was comparable to that of a physician with 2-5 years of experience in fundus diseases at a tertiary referral hospital. The model is expected to be used as an effective aid for fundus disease screening.

Economic evaluation of combined population-based screening for multiple blindness-causing eye diseases in China: a cost-effectiveness analysis

BACKGROUND

More than 90% of vision impairment is avoidable. However, in China, a routine screening programme is currently unavailable in primary health care. With the dearth of economic evidence on screening programmes for multiple blindness-causing eye diseases, delivery options, and screening frequencies, we aimed to evaluate the costs and benefits of a population-based screening programme for multiple eye diseases in China.

METHODS

We developed a decision-analytic Markov model for a cohort of individuals aged 50 years and older with a total of 30 1-year cycles. We calculated the cost-effectiveness and cost-utility of screening programmes for multiple major blindness-causing eye diseases in China, including age-related macular degeneration, glaucoma, diabetic retinopathy, cataracts, and pathological myopia, from a societal perspective (including direct and indirect costs). We analysed rural and urban settings separately by different screening delivery options (non-telemedicine [ie, face-to-face] screening, artificial intelligence [AI] telemedicine screening, and non-AI telemedicine screening) and frequencies. We calculated incremental cost-utility ratios (ICURs) using quality-adjusted life-years and incremental cost-effectiveness ratios (ICERs) in terms of the cost per blindness year avoided. One-way deterministic and simulated probabilistic sensitivity analyses were used to assess the robustness of the main outcomes.

FINDINGS

Compared with no screening, non-telemedicine combined screening of multiple eye diseases satisfied the criterion for a highly cost-effective health intervention, with an ICUR of US$2494 (95% CI 1130 to 2716) and an ICER of $12 487 (8773 to 18 791) in rural settings. In urban areas, the ICUR was $624 (395 to 907), and the ICER was $7251 (4238 to 13 501). Non-AI telemedicine screening could result in fewer costs and greater gains in health benefits (ICUR $2326 [1064 to 2538] and ICER $11 766 [8200 to 18 000] in rural settings; ICUR $581 [368 to 864] and ICER $6920 [3926 to 13 231] in urban settings). AI telemedicine screening dominated no screening in rural settings, and in urban settings the ICUR was $244 (-315 to 1073) and the ICER was $2567 (-4111 to 15 389). Sensitivity analyses showed all results to be robust. By further comparison, annual AI telemedicine screening was the most cost-effective strategy in both rural and urban areas.

INTERPRETATION

Combined screening of multiple eye diseases is cost-effective in both rural and urban China. AI coupled with teleophthalmology presents an opportunity to promote equity in eye health.

FUNDING

National Natural Science Foundation of China.

Artificial Intelligence in Community-Based Diabetic Retinopathy Telemedicine Screening in Urban China: Cost-effectiveness and Cost-Utility Analyses With Real-world Data

BACKGROUND

Community-based telemedicine screening for diabetic retinopathy (DR) has been highly recommended worldwide. However, evidence from low- and middle-income countries (LMICs) on the choice between artificial intelligence (AI)-based and manual grading-based telemedicine screening is inadequate for policy making.

OBJECTIVE

The aim of this study was to test whether the AI model is more worthwhile than manual grading in community-based telemedicine screening for DR in the context of labor costs in urban China.

METHODS

We conducted cost-effectiveness and cost-utility analyses by using decision-analytic Markov models with 30 one-year cycles from a societal perspective to compare the cost, effectiveness, and utility of 2 scenarios in telemedicine screening for DR: manual grading and an AI model. Sensitivity analyses were performed. Real-world data were obtained mainly from the Shanghai Digital Eye Disease Screening Program. The main outcomes were the incremental cost-effectiveness ratio (ICER) and the incremental cost-utility ratio (ICUR). The ICUR thresholds were set as 1 and 3 times the local gross domestic product per capita.

RESULTS

The total expected costs for a 65-year-old resident were US $3182.50 and US $3265.40, while the total expected years without blindness were 9.80 years and 9.83 years, and the utilities were 6.748 quality-adjusted life years (QALYs) and 6.753 QALYs in the AI model and manual grading, respectively. The ICER for the AI-assisted model was US $2553.39 per year without blindness, and the ICUR was US $15,216.96 per QALY, which indicated that AI-assisted model was not cost-effective. The sensitivity analysis suggested that if there is an increase in compliance with referrals after the adoption of AI by 7.5%, an increase in on-site screening costs in manual grading by 50%, or a decrease in on-site screening costs in the AI model by 50%, then the AI model could be the dominant strategy.

CONCLUSIONS

Our study may provide a reference for policy making in planning community-based telemedicine screening for DR in LMICs. Our findings indicate that unless the referral compliance of patients with suspected DR increases, the adoption of the AI model may not improve the value of telemedicine screening compared to that of manual grading in LMICs. The main reason is that in the context of the low labor costs in LMICs, the direct health care costs saved by replacing manual grading with AI are less, and the screening effectiveness (QALYs and years without blindness) decreases. Our study suggests that the magnitude of the value generated by this technology replacement depends primarily on 2 aspects. The first is the extent of direct health care costs reduced by AI, and the second is the change in health care service utilization caused by AI. Therefore, our research can also provide analytical ideas for other health care sectors in their decision to use AI.

DeepFundus: A flow-cytometry-like image quality classifier for boosting the whole life cycle of medical artificial intelligence

Medical artificial intelligence (AI) has been moving from the research phase to clinical implementation. However, most AI-based models are mainly built using high-quality images preprocessed in the laboratory, which is not representative of real-world settings. This dataset bias proves a major driver of AI system dysfunction. Inspired by the design of flow cytometry, DeepFundus, a deep-learning-based fundus image classifier, is developed to provide automated and multidimensional image sorting to address this data quality gap. DeepFundus achieves areas under the receiver operating characteristic curves (AUCs) over 0.9 in image classification concerning overall quality, clinical quality factors, and structural quality analysis on both the internal test and national validation datasets. Additionally, DeepFundus can be integrated into both model development and clinical application of AI diagnostics to significantly enhance model performance for detecting multiple retinopathies. DeepFundus can be used to construct a data-driven paradigm for improving the entire life cycle of medical AI practice.

Towards precision medicine based on a continuous deep learning optimization and ensemble approach

We developed a continuous learning system (CLS) based on deep learning and optimization and ensemble approach, and conducted a retrospective data simulated prospective study using ultrasound images of breast masses for precise diagnoses. We extracted 629 breast masses and 2235 images from 561 cases in the institution to train the model in six stages to diagnose benign and malignant tumors, pathological types, and diseases. We randomly selected 180 out of 3098 cases from two external institutions. The CLS was tested with seven independent datasets and compared with 21 physicians, and the system's diagnostic ability exceeded 20 physicians by training stage six. The optimal integrated method we developed is expected accurately diagnose breast masses. This method can also be extended to the intelligent diagnosis of masses in other organs. Overall, our findings have potential value in further promoting the application of AI diagnosis in precision medicine.

An AI-Aided Diagnostic Framework for Hematologic Neoplasms Based on Morphologic Features and Medical Expertise

A morphologic examination is essential for the diagnosis of hematological diseases. However, its conventional manual operation is time-consuming and laborious. Herein, we attempt to establish an artificial intelligence (AI)-aided diagnostic framework integrating medical expertise. This framework acts as a virtual hematological morphologist (VHM) for diagnosing hematological neoplasms. Two datasets were established as follows: An image dataset was used to train the Faster Region-based Convolutional Neural Network to develop an image-based morphologic feature extraction model. A case dataset containing retrospective morphologic diagnostic data was used to train a support vector machine algorithm to develop a feature-based case identification model based on diagnostic criteria. Integrating these 2 models established a whole-process AI-aided diagnostic framework, namely, VHM, and a 2-stage strategy was applied to practice case diagnosis. The recall and precision of VHM in bone marrow cell classification were 94.65% and 93.95%, respectively. The balanced accuracy, sensitivity, and specificity of VHM were 97.16%, 99.09%, and 92%, respectively, in the differential diagnosis of normal and abnormal cases, and 99.23%, 97.96%, and 100%, respectively, in the precise diagnosis of chronic myelogenous leukemia in chronic phase. This work represents the first attempt, to our knowledge, to extract multimodal morphologic features and to integrate a feature-based case diagnosis model for designing a comprehensive AI-aided morphologic diagnostic framework. The performance of our knowledge-based framework was superior to that of the widely used end-to-end AI-based diagnostic framework in terms of testing accuracy (96.88% vs 68.75%) or generalization ability (97.11% vs 68.75%) in differentiating normal and abnormal cases. The remarkable advantage of VHM is that it follows the logic of clinical diagnostic procedures, making it a reliable and interpretable hematological diagnostic tool.

Deep-learning-based personalized prediction of absolute neutrophil count recovery and comparison with clinicians for validation

Neutropenia and its complications are major adverse effects of cytotoxic chemotherapy. The time to recovery from neutropenia varies from patient to patient, and cannot be easily predicted even by experts. Therefore, we trained a deep learning model using data from 525 pediatric patients with solid tumors to predict the day when patients recover from severe neutropenia after high-dose chemotherapy. We validated the model with data from 99 patients and compared its performance to those of clinicians. The accuracy of the model at predicting the recovery day, with a 1-day error, was 76%; its performance was better than those of the specialist group (58.59%) and the resident group (32.33%). In addition, 80% of clinicians changed their initial predictions at least once after the model's prediction was conveyed to them. In total, 86 prediction changes (90.53%) improved the recovery day estimate.

Identification of ocular refraction based on deep learning algorithm as a novel retinoscopy method

BACKGROUND

The evaluation of refraction is indispensable in ophthalmic clinics, generally requiring a refractor or retinoscopy under cycloplegia. Retinal fundus photographs (RFPs) supply a wealth of information related to the human eye and might provide a promising approach that is more convenient and objective. Here, we aimed to develop and validate a fusion model-based deep learning system (FMDLS) to identify ocular refraction via RFPs and compare with the cycloplegic refraction. In this population-based comparative study, we retrospectively collected 11,973 RFPs from May 1, 2020 to November 20, 2021. The performance of the regression models for sphere and cylinder was evaluated using mean absolute error (MAE). The accuracy, sensitivity, specificity, area under the receiver operating characteristic curve, and F1-score were used to evaluate the classification model of the cylinder axis.

RESULTS

Overall, 7873 RFPs were retained for analysis. For sphere and cylinder, the MAE values between the FMDLS and cycloplegic refraction were 0.50 D and 0.31 D, representing an increase of 29.41% and 26.67%, respectively, when compared with the single models. The correlation coefficients (r) were 0.949 and 0.807, respectively. For axis analysis, the accuracy, specificity, sensitivity, and area under the curve value of the classification model were 0.89, 0.941, 0.882, and 0.814, respectively, and the F1-score was 0.88.

CONCLUSIONS

The FMDLS successfully identified the ocular refraction in sphere, cylinder, and axis, and showed good agreement with the cycloplegic refraction. The RFPs can provide not only comprehensive fundus information but also the refractive state of the eye, highlighting their potential clinical value.

Medical Staff and Resident Preferences for Using Deep Learning in Eye Disease Screening: Discrete Choice Experiment

Background

Deep learning–assisted eye disease diagnosis technology is increasingly applied in eye disease screening. However, no research has suggested the prerequisites for health care service providers and residents willing to use it.

Objective

The aim of this paper is to reveal the preferences of health care service providers and residents for using artificial intelligence (AI) in community-based eye disease screening, particularly their preference for accuracy.

Methods

Discrete choice experiments for health care providers and residents were conducted in Shanghai, China. In total, 34 medical institutions with adequate AI-assisted screening experience participated. A total of 39 medical staff and 318 residents were asked to answer the questionnaire and make a trade-off among alternative screening strategies with different attributes, including missed diagnosis rate, overdiagnosis rate, screening result feedback efficiency, level of ophthalmologist involvement, organizational form, cost, and screening result feedback form. Conditional logit models with the stepwise selection method were used to estimate the preferences.

Results

Medical staff preferred high accuracy: The specificity of deep learning models should be more than 90% (odds ratio [OR]=0.61 for 10% overdiagnosis; P<.001), which was much higher than the Food and Drug Administration standards. However, accuracy was not the residents’ preference. Rather, they preferred to have the doctors involved in the screening process. In addition, when compared with a fully manual diagnosis, AI technology was more favored by the medical staff (OR=2.08 for semiautomated AI model and OR=2.39 for fully automated AI model; P<.001), while the residents were in disfavor of the AI technology without doctors’ supervision (OR=0.24; P<.001).

Conclusions

Deep learning model under doctors’ supervision is strongly recommended, and the specificity of the model should be more than 90%. In addition, digital transformation should help medical staff move away from heavy and repetitive work and spend more time on communicating with residents.

Multi-label classification of fundus images with graph convolutional network and LightGBM

Early detection and treatment of retinal disorders are critical for avoiding irreversible visual impairment. Given that patients in the clinical setting may have various types of retinal illness, the development of multi-label fundus disease detection models capable of screening for multiple diseases is more in line with clinical needs. This article presented a composite model based on hybrid graph convolution for patient-level multi-label fundus illness identification. The composite model comprised a backbone module, a hybrid graph convolution module, and a classifier module. This article established the relationship between labels via graph convolution and then employed a self-attention mechanism to design a hybrid graph convolution structure. The backbone module extracted features using EfficientNet-B4, whereas the classifier module output multi-label using LightGBM. Additionally, this work investigated the input pattern of binocular images and the influence of label correlation on the model's identification performance. The proposed model MCGL-Net outperformed all other state-of-the-art methods on the publicly available ODIR dataset, with F1 reaching 91.60% on the test set. Ablation experiments were also performed in this paper. Experiments showed that the idea of hybrid graph convolutional structure and composite model designed in this paper promotes the model performance under any backbone CNN. The adoption of hybrid graph convolution can increase the F1 by 2.39% in trials using EfficientNet-B4 as the backbone. The composite model had a higher F1 index by 5.42% than the single EfficientNet-B4 model.

A Multicenter Clinical Study of the Automated Fundus Screening Algorithm

Purpose

To evaluate the effectiveness of automated fundus screening software in detecting eye diseases by comparing the reported results against those given by human experts.

Results

There were 1585 subjects who completed the procedure and yielded qualified images. The prevalence of referable diabetic retinopathy (RDR), glaucoma suspect (GCS), and referable macular diseases (RMD) were 20.4%, 23.2%, and 49.0%, respectively. The overall sensitivity values for RDR, GCS, and RMD diagnosis are 0.948 (95% confidence interval [CI], 0.918-0.967), 0.891 (95% CI, 0.855-0.919), and 0.901 (95% CI-0.878, 0.920), respectively. The overall specificity values for RDR, GCS, and RMD diagnosis are 0.954 (95% CI, 0.915-0.965), 0.993 (95% CI-0.986, 0.996), and 0.955 (95% CI-0.939, 0.968), respectively.

Methods

We prospectively enrolled 1743 subjects at seven hospitals throughout China. At each hospital, an operator records the subjects' information, takes fundus images, and submits the images to the Image Reading Center of Zhongshan Ophthalmic Center, Sun Yat-Sen University (IRC). The IRC grades the images according to the study protocol. Meanwhile, these images will also be automatically screened by the artificial intelligence algorithm. Then, the analysis results of automated screening algorithm are compared against the grading results of IRC. The end point goals are lower bounds of 95% CI of sensitivity values that are greater than 0.85 for all three target diseases, and lower bounds of 95% CI of specificity values that are greater than 0.90 for RDR and 0.85 for GCS and RMD.

Conclusions

Automated fundus screening software demonstrated a high sensitivity and specificity in detecting RDR, GCS, and RMD from color fundus imaged captured using various cameras.

Translational Relevance

These findings suggest that automated software can improve the screening effectiveness for eye diseases, especially in a primary care context, where experienced ophthalmologists are scarce.