Artificial intelligence applications in OCT and OCTA for diabetic retinopathy: A systematic review

Clinical Snapshot

Key Findings

• P Value: Not reported

• Effect Size: AUC values typically ranging from 0.90 to 0.99 for DL-based classification and segmentation tasks; DL consistently outperformed traditional ML methods

• Primary Outcome: Diagnostic performance of AI/DL models for DR detection, classification, and segmentation using OCT and OCTA images across 175 included studies

• Nnt Or Sensitivity: Specific sensitivity and specificity values not reported in abstract; AUC range of 0.90–0.99 implies high discriminative performance but individual study metrics are not synthesised. CNNs, ViTs, and U-Net architectures identified as consistently strong performers across DR biomarker detection tasks

• Confidence Interval: Not reported — no pooled meta-analytic estimates provided

Clinical Application

Clinical implementation remains premature. Key barriers include: lack of prospective real-world validation studies; model interpretability concerns limiting clinician trust; absence of standardised datasets for benchmarking; and significant variability in performance across imaging devices and clinical settings. Integration with existing retinal imaging workflows would require substantial health informatics infrastructure. Australia has a high burden of diabetic retinopathy, with approximately 1.3 million Australians living with diabetes and DR representing a leading cause of preventable blindness. The RACGP recommends annual retinal examination for all people with diabetes. OCT is Medicare-rebatable (MBS item numbers 11215, 11216) and widely available in Australian ophthalmology and optometry practices. OCTA is less widely reimbursed. The TGA would need to evaluate and register any AI-based DR diagnostic software as a Class IIa or IIb medical device under the Software as a Medical Device (SaMD) framework before clinical deployment. No TGA-approved AI DR screening tool using OCT/OCTA is currently listed on the ARTG as of early 2026. Indigenous Australians have disproportionately high rates of DR and limited access to specialist eye care — AI-assisted screening could theoretically address this equity gap, but only if validated in this population, which current evidence does not support. The National Diabetes Services Scheme (NDSS) and KeepSight programme provide relevant infrastructure for future integration. Adults with diabetes mellitus undergoing retinal screening or monitoring for diabetic retinopathy, particularly in settings where OCT or OCTA imaging is available. Most applicable to secondary and tertiary ophthalmology care settings; potential future application in primary care diabetic eye screening programmes.

Abstract

PURPOSE: Diabetic retinopathy (DR) is a leading cause of vision impairment worldwide. Optical coherence tomography (OCT) and OCT angiography (OCTA) provide detailed retinal imaging, enabling early detection of microvascular changes. This study aims to systematically review artificial intelligence (AI), particularly deep learning (DL), applications for DR detection and analysis using OCT and OCTA images. METHODS: A comprehensive literature search was conducted across PubMed, Web of Science, Scopus, IEEE Xplore, and Embase for studies published up to March 2026. A total of 1007 articles were identified, of which 175 studies met the inclusion criteria following the PRISMA study selection process. RESULTS: DL-based approaches consistently demonstrated superior performance compared to traditional machine learning (ML) methods, with reported AUC values typically ranging from 0.90 to 0.99 across classification and segmentation tasks. Convolutional neural networks (CNNs), Vision Transformers (ViTs), and encoder-decoder architectures such as U-Net showed strong performance in detecting key DR biomarkers, including microaneurysms, macular edema, and neovascularization. However, performance variability was observed depending on dataset size, imaging modality, and annotation quality. CONCLUSIONS: AI-driven analysis of OCT and OCTA images offers significant potential for automated DR detection. Despite promising results, challenges such as limited public datasets, lack of cross-institutional validation, and model interpretability remain. Future research should focus on multimodal integration, explainable AI, and large-scale validation to enhance clinical applicability.

References

1. 1.Tavakoli, M., & Ramezanzadeh, E. (2026). Artificial intelligence applications in OCT and OCTA for diabetic retinopathy: A systematic review. Experimental Eye Research. https://doi.org/10.1016/j.exer.2026.111046