Automated segmentation of ultra-widefield fluorescein angiography of diabetic retinopathy using deep learning

When Two Eyes Don't Suffice-Learning Difficult Hyperfluorescence Segmentations in Retinal Fundus Autofluorescence Images via Ensemble Learning

Hyperfluorescence (HF) and reduced autofluorescence (RA) are important biomarkers in fundus autofluorescence images (FAF) for the assessment of health of the retinal pigment epithelium (RPE), an important indicator of disease progression in geographic atrophy (GA) or central serous chorioretinopathy (CSCR). Autofluorescence images have been annotated by human raters, but distinguishing biomarkers (whether signals are increased or decreased) from the normal background proves challenging, with borders being particularly open to interpretation. Consequently, significant variations emerge among different graders, and even within the same grader during repeated annotations. Tests on in-house FAF data show that even highly skilled medical experts, despite previously discussing and settling on precise annotation guidelines, reach a pair-wise agreement measured in a Dice score of no more than 63-80% for HF segmentations and only 14-52% for RA. The data further show that the agreement of our primary annotation expert with herself is a 72% Dice score for HF and 51% for RA. Given these numbers, the task of automated HF and RA segmentation cannot simply be refined to the improvement in a segmentation score. Instead, we propose the use of a segmentation ensemble. Learning from images with a single annotation, the ensemble reaches expert-like performance with an agreement of a 64-81% Dice score for HF and 21-41% for RA with all our experts. In addition, utilizing the mean predictions of the ensemble networks and their variance, we devise ternary segmentations where FAF image areas are labeled either as confident background, confident HF, or potential HF, ensuring that predictions are reliable where they are confident (97% Precision), while detecting all instances of HF (99% Recall) annotated by all experts.

Imaging Modalities for Assessing the Vascular Component of Diabetic Retinal Disease: Review and Consensus for an Updated Staging System

Purpose

To review the evidence for imaging modalities in assessing the vascular component of diabetic retinal disease (DRD), to inform updates to the DRD staging system.

Design

Standardized narrative review of the literature by an international expert workgroup, as part of the DRD Staging System Update Effort, a project of the Mary Tyler Moore Vision Initiative. Overall, there were 6 workgroups: Vascular Retina, Neural Retina, Systemic Health, Basic and Cellular Mechanisms, Visual Function, and Quality of Life.

Participants

The Vascular Retina workgroup, including 16 participants from 4 countries.

Methods

Literature review was conducted using standardized evidence grids for 5 modalities: standard color fundus photography (CFP), widefield color photography (WFCP), standard fluorescein angiography (FA), widefield FA (WFFA), and OCT angiography (OCTA). Summary levels of evidence were determined on a validated scale from I (highest) to V (lowest). Five virtual workshops were held for discussion and consensus.

Main Outcome Measures

Level of evidence for each modality.

Results

Levels of evidence for standard CFP, WFCP, standard FA, WFFA, and OCTA were I, II, I, I, and II respectively. Traditional vascular lesions on standard CFP should continue to be included in an updated staging system, but more studies are required before they can be used in posttreatment eyes. Widefield color photographs can be used for severity grading within the area covered by standard CFPs, although these gradings may not be directly interchangeable with each other. Evaluation of the peripheral retina on WFCP can be considered, but the method of grading needs to be clarified and validated. Standard FA and WFFA provide independent prognostic value, but the need for dye administration should be considered. OCT angiography has significant potential for inclusion in the DRD staging system, but various barriers need to be addressed first.

Conclusions

This study provides evidence-based recommendations on the utility of various imaging modalities for assessment of the vascular component of DRD, which can inform future updates to the DRD staging system. Although new imaging modalities offer a wealth of information, there are still major gaps and unmet research needs that need to be addressed before this potential can be realized.

Financial Disclosures

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

Using Deep Learning to Segment Retinal Vascular Leakage and Occlusion in Retinal Vasculitis

PURPOSE

Retinal vasculitis (RV) is characterised by retinal vascular leakage, occlusion or both on fluorescein angiography (FA). There is no standard scheme available to segment RV features. We aimed to develop a deep learning model to segment both vascular leakage and occlusion in RV.

METHODS

Four hundred and sixty-three FA images from 82 patients with retinal vasculitis were used to develop a deep learning model, in 60:20:20 ratio for training:validation:testing. Parameters, including deep learning architectures (DeeplabV3+, UNet++ and UNet), were altered to find the best binary segmentation model separately for retinal vascular leakage and occlusion, using a Dice score to determine the reliability of each model.

RESULTS

Our best model for vascular leakage had a Dice score of 0.6279 (95% confidence interval (CI) 0.5584-0.6974). For occlusion, the best model achieved a Dice score of 0.6992 (95% CI 0.6109-0.7874).

CONCLUSION

Our RV segmentation models could perform reliable segmentation for retinal vascular leakage and occlusion in FAs of RV patients.

Clinical utility of ultra-widefield fluorescein angiography and optical coherence tomography angiography for retinal vein occlusions

Retinal vein occlusions (RVOs) are the second most common retinal vascular disease after diabetic retinopathy, and are a significant cause of visual impairment, especially in the elderly population. RVOs result in visual loss due to macular ischemia, cystoid macular edema (CME), and complications related to neovascularization. Vascular assessment in RVOs traditionally relies on standard fluorescein angiography (FA) for assessment of macular and retinal ischemia, which aids in prognostication and guides intervention. Standard FA has significant limitations-it is time-consuming, requires invasive dye administration, allows for limited assessment of the peripheral retina, and is usually evaluated semi-qualitatively, by ophthalmologists with tertiary expertise. More recently, the introduction of ultra-widefield FA (UWF FA) and optical coherence tomography angiography (OCTA) into clinical practice has changed the tools available for vascular evaluation in RVOs. UWF FA allows for evaluation of peripheral retinal perfusion, and OCTA is non-invasive, rapidly-acquired, and provides more information on capillary perfusion. Both modalities can be used to provide more quantitative parameters related to retinal perfusion. In this article, we review the clinical utility and impact of UWF FA and OCTA in the evaluation and management of patients with RVOs.