A fundus image dataset for intelligent retinopathy of prematurity system.

Image-based artificial intelligence (AI) systems stand as the major modality for evaluating ophthalmic conditions. However, most of the currently available AI systems are designed for experimental research using single-central datasets. Most of them fell short of application in real-world clinical settings. In this study, we collected a dataset of 1,099 fundus images in both normal and pathologic eyes from 483 premature infants for intelligent retinopathy of prematurity (ROP) system development and validation. Dataset diversity was visualized with a spatial scatter plot. Image classification was conducted by three annotators. To the best of our knowledge, this is one of the largest fundus datasets on ROP, and we believe it is conducive to the real-world application of AI systems.

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.

Multiplatform tear proteomic profiling reveals novel non-invasive biomarkers for diabetic retinopathy.

OBJECTIVES: To investigate a comprehensive proteomic profile of the tear fluid in patients with diabetic retinopathy (DR) and further define non-invasive biomarkers. METHODS: A cross-sectional, multicentre study that includes 46 patients with DR, 28 patients with diabetes mellitus (DM), and 30 healthy controls (HC). Tear samples were collected with Schirmer strips. As for the discovery set, data-independent acquisition mass spectrometry was used to characterize the tear proteomic profile. Differentially expressed proteins between groups were identified, with gene ontology enrichment analysis and Kyoto Encyclopedia of Genes and Genomes enrichment analysis further developed. Classifying performance of biomarkers for distinguishing DR from DM was compared by the combination of three machine-learning algorithms. The selected biomarker panel was tested in the validation cohort using parallel reaction monitoring mass spectrometry. RESULTS: Among 3364 proteins quantified, 235 and 88 differentially expressed proteins were identified for DR when compared to HC and DM, respectively, which were fundamentally related to retina homeostasis, inflammation and immunity, oxidative stress, angiogenesis and coagulation, metabolism, and cellular adhesion processes. The biomarker panel consisting of NAD-dependent protein deacetylase sirtuin-2 (SIR2), amine oxidase [flavin-containing] B (AOFB), and U8 snoRNA-decapping enzyme (NUD16) exhibited the best diagnostic performance in discriminating DR from DM, with AUCs of 0.933 and 0.881 in the discovery and validation set, respectively. CONCLUSIONS: Tear protein dysregulation is comprehensively revealed to be associated with DR onset. The combination of tear SIR2, AOFB, and NUD16 can be a novel potential approach for non-invasive detection or pre-screening of DR. CLINICAL TRIAL REGISTRATION: Chinese Clinical Trial Registry Identifier: ChiCTR2100054263. https://www.chictr.org.cn/showproj.html?proj=143177 . Date of registration: 2021/12/12.

Refractive Status and Biometric Characteristics of Children With Familial Exudative Vitreoretinopathy.

Purpose: To compare biometric characteristics between patients with early-stage familial exudative vitreoretinopathy (FEVR) and healthy controls. Methods: This case-control study included 50 FEVR eyes in stage 1-2 and 50 control eyes matched by age, gender and spherical equivalent (SE). Biometric parameters including axial length (AL), white-to-white diameter (WTW), central corneal thickness (CCT), anterior chamber depth (ACD), lens thickness (LT), pupil diameter, vitreous chamber depth, anterior and posterior corneal surface curvature radius (ACR and PCR), anterior lens surface curvature radius (ALR) and posterior lens surface curvature radius were measured using IOLMaster 700 and compared between cases and controls using paired t-test. Correlations between SE and biometric measures were assessed using Pearson correlation coefficient (r) in cases and controls. Results: Both FEVR cases and matched controls had a mean age of 7.6 years, 48% female and mean SE of -5.3 D (80% myopia). Compared to controls, FEVR eyes had smaller AL (P = 0.009), WTW (P = 0.001), ACD (P < 0.001), and ALR (P = 0.03), but larger CCT (P = 0.02) and LT (P = 0.01). In FEVR eyes, SE was negatively correlated with AL (r = -0.79, P < 0.001), positively correlated with ACR (r = 0.29, P = 0.04) and PCR (r = 0.33, P = 0.02), whereas in controls, SE was negatively correlated with AL (r = -0.82, P < 0.001) and LT (r = -0.34, P = 0.02), positively correlated with ALR (r = 0.29, P = 0.04). Conclusions: Patients at early stage of FEVR exhibited a unique eye morphology resembling ocular development arrest, which may help to develop screening and early detection tools for FEVR. In FEVR patients, myopia is very prevalent and significantly associated with corneal curvature increase.