Hybrid Fusion of High-Resolution and Ultra-Widefield OCTA Acquisitions for the Automatic Diagnosis of Diabetic Retinopathy

Advancing Diabetic Retinopathy Screening: A Systematic Review of Artificial Intelligence and Optical Coherence Tomography Angiography Innovations

Background/Objectives: Diabetic retinopathy (DR) remains a leading cause of preventable blindness, with its global prevalence projected to rise sharply as diabetes incidence increases. Early detection and timely management are critical to reducing DR-related vision loss. Optical Coherence Tomography Angiography (OCTA) now enables non-invasive, layer-specific visualization of the retinal vasculature, facilitating more precise identification of early microvascular changes. Concurrently, advancements in artificial intelligence (AI), particularly deep learning (DL) architectures such as convolutional neural networks (CNNs), attention-based models, and Vision Transformers (ViTs), have revolutionized image analysis. These AI-driven tools substantially enhance the sensitivity, specificity, and interpretability of DR screening. Methods: A systematic review of PubMed, Scopus, WOS, and Embase databases, including quality assessment of published studies, investigating the result of different AI algorithms with OCTA parameters in DR patients was conducted. The variables of interest comprised training databases, type of image, imaging modality, number of images, outcomes, algorithm/model used, and performance metrics. Results: A total of 32 studies were included in this systematic review. In comparison to conventional ML techniques, our results indicated that DL algorithms significantly improve the accuracy, sensitivity, and specificity of DR screening. Multi-branch CNNs, ensemble architectures, and ViTs were among the sophisticated models with remarkable performance metrics. Several studies reported that accuracy and area under the curve (AUC) values were higher than 99%. Conclusions: This systematic review underscores the transformative potential of integrating advanced DL and machine learning (ML) algorithms with OCTA imaging for DR screening. By synthesizing evidence from 32 studies, we highlight the unique capabilities of AI-OCTA systems in improving diagnostic accuracy, enabling early detection, and streamlining clinical workflows. These advancements promise to enhance patient management by facilitating timely interventions and reducing the burden of DR-related vision loss. Furthermore, this review provides critical recommendations for clinical practice, emphasizing the need for robust validation, ethical considerations, and equitable implementation to ensure the widespread adoption of AI-OCTA technologies. Future research should focus on multicenter studies, multimodal integration, and real-world validation to maximize the clinical impact of these innovative tools.

Progression of Capillary Hypoperfusion in Advanced Stages of Nonproliferative Diabetic Retinopathy: 6-month Analysis of RICHARD Study

Purpose

To evaluate the 6-month progression of retinal capillary perfusion in eyes with advanced stages of nonproliferative diabetic retinopathy (NPDR).

Design

RICHARD (NCT05112445), 2-year prospective longitudinal study.

Participants

Sixty eyes with Diabetic Retinopathy Severity Scale (DRSS) levels 43, 47, and 53 from 60 patients with type 2 diabetes. Fifty-one patients completed the 6-month evaluation.

Methods

Eyes were evaluated on Optos California (Optos plc) ultrawidefield fundus fluorescein angiography (UWF-FFA), swept-source OCT angiography (SS-OCTA) (PLEX Elite 9000, ZEISS) and spectral-domain OCTA (SD-OCTA) (CIRRUS HD-OCT 5000 Angioplex, ZEISS). DRSS classification was performed based on 7-field color fundus photographs (CFPs) complemented with Optos California UWF-fundus imaging.

Main Outcome Measures

Ischemic index was obtained from Optos. Vascular quantification metrics, namely foveal avascular zone, skeletonized vessel density (SVD), and perfusion density (PD) metrics, were acquired on OCTA in the superficial and deep capillary plexuses (SCP and DCP). Microaneurysm assessment was automatically performed based on CFP images using the RetmarkerDR (Retmarker SA, Meteda Group).

Results

Swept-source-OCTA metrics showed statistically significant differences between the advanced stages of NPDR. Differences between DRSS levels 47 and 53 were found at baseline in the inner ring (SVD, SCP: P = 0.005 and DCP: P = 0.042 and PD, SCP: P = 0.003) and outer ring (SVD, SCP: P = 0.007 and DCP: P = 0.030 and PD, SCP: P = 0.020 and DCP: P = 0.025). No significant differences were observed at baseline between DRSS levels 43 and 47. In SD-OCTA, the differences were similar but did not reach statistical significance. The total ischemic index showed an increase in association with diabetic retinopathy (DR) severity, but the differences between DRSS levels did not reach statistical significance. The number of microaneurysms also increased significantly with DR severity (P = 0.033). Statistically significant 6-month progression was detected with SS-OCTA in eyes with DRSS levels 47 and 53 but not in DRSS level 43. In eyes with DRSS level 53, 6-month progression was identified using a combination of metrics of capillary nonperfusion and microaneurysm counts.

Conclusions

In a 6-month period, significant microvascular disease progression can be identified in eyes with DRSS levels 47 and 53 by performing OCTA examinations and microaneurysm counting using CFP.

Financial Disclosures

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

Intraretinal Microvascular Abnormalities in Eyes with Advanced Stages of Nonproliferative Diabetic Retinopathy: Comparison Between UWF-FFA, CFP, and OCTA-The RICHARD Study

INTRODUCTION

This study aimed to evaluate intraretinal microvascular abnormalities (IRMA) in eyes with advanced nonproliferative diabetic retinopathy (NPDR) using multimodal approach in co-located areas focusing on central retina (up to 50°) and to look at possible correlations between IRMA and other structural changes, like ischemia and presence of microaneurysms.

METHODS

The RICHARD study (NCT05112445) included 60 eyes from 60 patients with type 2 diabetes with moderate-severe NPDR, diabetic retinopathy severity levels 43, 47, and 53 (DRSS). IRMA were defined as capillary tortuosity covering a minimum circular area of 300 µm (calculated to correspond to the Early Treatment Diabetic Retinopathy Study standard photo 8A) and were identified using multimodal imaging with distinct fields of view (FoV): color fundus photography (CFP) using a Topcon TRC-50DX camera (Topcon Medical Systems, Japan), Optos California ultra wide field fundus fluorescein angiography (UWF-FFA) (Optos plc, UK), and swept-source optical coherence tomography angiography (SS-OCTA) (PLEX® Elite 9000, ZEISS, USA). Different areas of the retina were examined: central macula (up to 20°) and posterior pole (between 20° and 50°).

RESULTS

Multimodal imaging was used to identify IRMA in co-located areas (FoV < 50°) including UWF-FFA, CFP, and SS-OCTA. In eyes with DRSS levels 47 and 53, IRMA were identified in both areas of the retina, while in eyes with DRSS level 43, IRMA were detected only outside of the central macula (FoV > 20°). Our results show that when evaluating the presence of IRMA (FoV < 50°), UWF-FFA detected 203 IRMA, SS-OCTA detected 133 IRMA, and CFP detected 104 IRMA. Our results also show that the presence of IRMA was positively associated with presence of microaneurysms.

CONCLUSIONS

Identification of IRMA in eyes with advanced NPDR is better achieved by UWF-FFA than CFP and SS-OCTA. A statistically significant correlation was found between the presence of IRMA and the increase in number of microaneurysms.

TRIAL REGISTRATION

ClinicalTrials.gov, identifier NCT05112445.

Association of microaneurysms with retinal vascular alterations in patients with retinal vein occlusion

OBJECTIVE

To investigate the localization, distribution, and type of central microaneurysms (MAs) and their relationship with retinal vascular alterations in patients with retinal vein occlusion (RVO).

METHODS

In this cross-sectional study, ultra-widefield color fundus photography (UWF-CF), standard and single-capture 65° widefield (WF) optical coherence tomography angiography (OCTA) were performed in consecutive patients with RVO treated at the Department of Ophthalmology and Optometry, Medical University of Vienna. UWF-CF, en face and B-Scans in 6 mm × 6 mm OCTA were examined for detection of MAs. Nonperfusion areas (NPA) and collateral vessels (CV) were evaluated on WF-OCTA, ghost vessels (GV), and tortuous vessels (TV) on UWF-CF.

RESULTS

One-hundred-and-twelve patients were included in the study, and data from 59 eyes of 59 patients with disease duration longer than 3 months, good image quality, and without relevant ocular comorbidities were eligible for statistical analysis. Fifty-six of 59 (94.9%) patients were previously treated with anti-vascular endothelial growth factor agents for macular edema, 31 of 59 (52.5%) patients presented with MAs in the central 6 mm and 60 MAs were found in total using multimodal imaging. There was no statistically significant difference in the greatest diameter of fluid-associated versus non-fluid-associated MAs (p = 0.53). Eyes with MAs were associated with CV, TV, and GV (χ2-test; p < 0.001, p = 0.0498, and p = 0.001). Median NPA was 27.3 mm2 (quartiles 1.3-62.8 mm2) in eyes with MAs and 0 mm2 (quartiles 0-36.2 mm2) in eyes without MAs (Mann-Whitney-U-test; p = 0.018).

CONCLUSION

MAs were associated with extensive NPA, the presence of CV, GV, and TV. There was no correlation between the diameter of the MA and the adjacent intraretinal fluid in our predominantly pretreated RVO study patients.

A review of deep learning-based information fusion techniques for multimodal medical image classification

Multimodal medical imaging plays a pivotal role in clinical diagnosis and research, as it combines information from various imaging modalities to provide a more comprehensive understanding of the underlying pathology. Recently, deep learning-based multimodal fusion techniques have emerged as powerful tools for improving medical image classification. This review offers a thorough analysis of the developments in deep learning-based multimodal fusion for medical classification tasks. We explore the complementary relationships among prevalent clinical modalities and outline three main fusion schemes for multimodal classification networks: input fusion, intermediate fusion (encompassing single-level fusion, hierarchical fusion, and attention-based fusion), and output fusion. By evaluating the performance of these fusion techniques, we provide insight into the suitability of different network architectures for various multimodal fusion scenarios and application domains. Furthermore, we delve into challenges related to network architecture selection, handling incomplete multimodal data management, and the potential limitations of multimodal fusion. Finally, we spotlight the promising future of Transformer-based multimodal fusion techniques and give recommendations for future research in this rapidly evolving field.

DISCOVER: 2-D multiview summarization of Optical Coherence Tomography Angiography for automatic diabetic retinopathy diagnosis

Diabetic Retinopathy (DR), an ocular complication of diabetes, is a leading cause of blindness worldwide. Traditionally, DR is monitored using Color Fundus Photography (CFP), a widespread 2-D imaging modality. However, DR classifications based on CFP have poor predictive power, resulting in suboptimal DR management. Optical Coherence Tomography Angiography (OCTA) is a recent 3-D imaging modality offering enhanced structural and functional information (blood flow) with a wider field of view. This paper investigates automatic DR severity assessment using 3-D OCTA. A straightforward solution to this task is a 3-D neural network classifier. However, 3-D architectures have numerous parameters and typically require many training samples. A lighter solution consists in using 2-D neural network classifiers processing 2-D en-face (or frontal) projections and/or 2-D cross-sectional slices. Such an approach mimics the way ophthalmologists analyze OCTA acquisitions: (1) en-face flow maps are often used to detect avascular zones and neovascularization, and (2) cross-sectional slices are commonly analyzed to detect macular edemas, for instance. However, arbitrary data reduction or selection might result in information loss. Two complementary strategies are thus proposed to optimally summarize OCTA volumes with 2-D images: (1) a parametric en-face projection optimized through deep learning and (2) a cross-sectional slice selection process controlled through gradient-based attribution. The full summarization and DR classification pipeline is trained from end to end. The automatic 2-D summary can be displayed in a viewer or printed in a report to support the decision. We show that the proposed 2-D summarization and classification pipeline outperforms direct 3-D classification with the advantage of improved interpretability.