Deep learning segmentation of hyperautofluorescent fleck lesions in Stargardt disease

Quantification of Fundus Autofluorescence Features in a Molecularly Characterized Cohort of More Than 3000 Inherited Retinal Disease Patients from the United Kingdom

Purpose

To quantify relevant fundus autofluorescence (FAF) image features cross-sectionally and longitudinally in a large cohort of inherited retinal diseases (IRDs) patients.

Design

Retrospective study of imaging data (55-degree blue-FAF on Heidelberg Spectralis) from patients.

Participants

Patients with a clinical and molecularly confirmed diagnosis of IRD who have undergone FAF 55-degree imaging at Moorfields Eye Hospital (MEH) and the Royal Liverpool Hospital (RLH) between 2004 and 2019.

Methods

Five FAF features of interest were defined: vessels, optic disc, perimacular ring of increased signal (ring), relative hypo-autofluorescence (hypo-AF) and hyper-autofluorescence (hyper-AF). Features were manually annotated by six graders in a subset of patients based on a defined grading protocol to produce segmentation masks to train an AI model, AIRDetect, which was then applied to the entire imaging dataset.

Main Outcome Measures

Quantitative FAF imaging features including area in mm 2 and vessel metrics, were analysed cross-sectionally by gene and age, and longitudinally to determine rate of progression. AIRDetect feature segmentation and detection were validated with Dice score and precision/recall, respectively.

Results

A total of 45,749 FAF images from 3,606 IRD patients from MEH covering 170 genes were automatically segmented using AIRDetect. Model-grader Dice scores for disc, hypo-AF, hyper-AF, ring and vessels were respectively 0.86, 0.72, 0.69, 0.68 and 0.65. The five genes with the largest hypo-AF areas were CHM , ABCC6 , ABCA4 , RDH12 , and RPE65 , with mean per-patient areas of 41.5, 30.0, 21.9, 21.4, and 15.1 mm 2 . The five genes with the largest hyper-AF areas were BEST1 , CDH23 , RDH12 , MYO7A , and NR2E3 , with mean areas of 0.49, 0.45, 0.44, 0.39, and 0.34 mm 2 respectively. The five genes with largest ring areas were CDH23 , NR2E3 , CRX , EYS and MYO7A, with mean areas of 3.63, 3.32, 2.84, 2.39, and 2.16 mm 2 . Vessel density was found to be highest in EFEMP1 , BEST1 , TIMP3 , RS1 , and PRPH2 (10.6%, 10.3%, 9.8%, 9.7%, 8.9%) and was lower in Retinitis Pigmentosa (RP) and Leber Congenital Amaurosis genes. Longitudinal analysis of decreasing ring area in four RP genes ( RPGR, USH2A, RHO, EYS ) found EYS to be the fastest progressor at -0.18 mm 2 /year.

Conclusions

We have conducted the first large-scale cross-sectional and longitudinal quantitative analysis of FAF features across a diverse range of IRDs using a novel AI approach.

Nonlinear Reduction in Hyperautofluorescent Ring Area in Retinitis Pigmentosa

PURPOSE

To report baseline dimension of the autofluorescent (AF) ring in a large cohort of retinitis pigmentosa (RP) patients and to evaluate models of ring progression.

DESIGN

Cohort study.

PARTICIPANTS

Four hundred and forty-five eyes of 224 patients with clinical diagnosis of RP.

METHODS

Autofluorescent rings from near-infrared AF (NIRAF) and short-wavelength AF (SWAF) imaging modalities in RP eyes were segmented with ring area and horizontal extent extracted from each image for cross-sectional and longitudinal analyses. In longitudinal analysis, for each eye, ring area, horizontal extent, and natural logarithm of the ring area were assessed as the best dependent variable for linear regression by evaluating R2 values. Linear mixed-effects modeling was utilized to account for intereye correlation.

MAIN OUTCOME MEASURES

Autofluorescent ring size characteristics at baseline and ring progression rates.

RESULTS

A total of 439 eyes had SWAF imaging at baseline with the AF ring observed in 206 (46.9%) eyes. Mean (95% confidence interval) of ring area and horizontal extent were 7.85 (6.60 to 9.11) mm2 and 3.35 (3.10 to 3.60) mm, respectively. In NIRAF, the mean ring area and horizontal extent were 7.74 (6.60 to 8.89) mm2 and 3.26 (3.02 to 3.50) mm, respectively in 251 out of 432 eyes. Longitudinal analysis showed mean progression rates of -0.57 mm2/year and -0.12 mm/year in SWAF using area and horizontal extent as the dependent variable, respectively. When ln(Area) was analyzed as the dependent variable, mean progression was -0.07 ln(mm2)/year, which equated to 6.80% decrease in ring area per year. Similar rates were found in NIRAF (area: -0.59 mm2/year, horizontal extent: -0.12 mm/year and ln(Area): -0.08 ln(mm2)/year, equated to 7.75% decrease in area per year). Analysis of R2 showed that the dependent variable ln(Area) provided the best linear model for ring progression in both imaging modalities, especially in eyes with large overall area change.

CONCLUSIONS

Our data suggest that using an exponential model to estimate progression of the AF ring area in RP is more appropriate than the models assuming linear decrease. Hence, the progression estimates provided in this study should provide more accurate reference points in designing clinical trials in RP patients.

FINANCIAL DISCLOSURE(S)

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

Automated identification of fleck lesions in Stargardt disease using deep learning enhances lesion detection sensitivity and enables morphometric analysis of flecks

PURPOSE

To classify fleck lesions and assess artificial intelligence (AI) in identifying flecks in Stargardt disease (STGD).

METHODS

A retrospective study of 170 eyes from 85 consecutive patients with confirmed STGD. Fundus autofluorescence images were extracted, and flecks were manually outlined. A deep learning model was trained, and a hold-out testing subset was used to compare with manually identified flecks and for graders to assess. Flecks were clustered using K-means clustering.

RESULTS

Of the 85 subjects, 45 were female, and the median age was 37 years (IQR 25-59). A subset of subjects (n=41) had clearly identifiable fleck lesions, and an AI was successfully trained to identify these lesions (average Dice score of 0.53, n=18). The AI segmentation had smaller (0.018 compared with 0.034 mm2, p<0.001) but more numerous flecks (75.5 per retina compared with 40.0, p<0.001), but the total size of flecks was not different. The AI model had higher sensitivity to detect flecks but resulted in more false positives. There were two clusters of flecks based on morphology: broadly, one cluster of small round flecks and another of large amorphous flecks. The per cent frequency of small round flecks negatively correlated with subject age (r=-0.31, p<0.005).

CONCLUSIONS

AI-based detection of flecks shows greater sensitivity than human graders but with a higher false-positive rate. With further optimisation to address current shortcomings, this approach could be used to prescreen subjects for clinical research. The feasibility and utility of quantifying fleck morphology in conjunction with AI-based segmentation as a biomarker of progression require further study.

Recurrent and Concurrent Prediction of Longitudinal Progression of Stargardt Atrophy and Geographic Atrophy

Stargardt disease and age-related macular degeneration are the leading causes of blindness in the juvenile and geriatric populations, respectively. The formation of atrophic regions of the macula is a hallmark of the end-stages of both diseases. The progression of these diseases is tracked using various imaging modalities, two of the most common being fundus autofluorescence (FAF) imaging and spectral-domain optical coherence tomography (SD-OCT). This study seeks to investigate the use of longitudinal FAF and SD-OCT imaging (month 0, month 6, month 12, and month 18) data for the predictive modelling of future atrophy in Stargardt and geographic atrophy. To achieve such an objective, we develop a set of novel deep convolutional neural networks enhanced with recurrent network units for longitudinal prediction and concurrent learning of ensemble network units (termed ReConNet) which take advantage of improved retinal layer features beyond the mean intensity features. Using FAF images, the neural network presented in this paper achieved mean (± standard deviation, SD) and median Dice coefficients of 0.895 (± 0.086) and 0.922 for Stargardt atrophy, and 0.864 (± 0.113) and 0.893 for geographic atrophy. Using SD-OCT images for Stargardt atrophy, the neural network achieved mean and median Dice coefficients of 0.882 (± 0.101) and 0.906, respectively. When predicting only the interval growth of the atrophic lesions with FAF images, mean (± SD) and median Dice coefficients of 0.557 (± 0.094) and 0.559 were achieved for Stargardt atrophy, and 0.612 (± 0.089) and 0.601 for geographic atrophy. The prediction performance in OCT images is comparably good to that using FAF which opens a new, more efficient, and practical door in the assessment of atrophy progression for clinical trials and retina clinics, beyond widely used FAF. These results are highly encouraging for a high-performance interval growth prediction when more frequent or longer-term longitudinal data are available in our clinics. This is a pressing task for our next step in ongoing research.

Multimodal Phenomap of Stargardt Disease Integrating Structural, Psychophysical, and Electrophysiologic Measures of Retinal Degeneration

Objective

To cluster the diverse phenotypic features of Stargardt disease (STGD) using unsupervised clustering of multimodal retinal structure and function data.

Design

Retrospective cross-sectional study.

Subjects

Eyes of subjects with STGD and fundus autofluorescence (FAF), OCT, electroretinography (ERG), and microperimetry (MP) data available within 1 year of the baseline evaluation.

Methods

A total of 46 variables from FAF, OCT, ERG, and MP results were recorded for subjects with STGD as defined per published criteria. Factor analysis of mixed data identified the most informative variables. Unsupervised hierarchical clustering and silhouette analysis identified the optimal number of clusters to classify multimodal phenotypes.

Main Outcome Measures

Phenotypic clusters of STGD subjects and the corresponding cluster features.

Results

We included 52 subjects and 102 eyes with a mean visual acuity (VA) at the time of multimodal testing of 0.69 ± 0.494 logarithm of minimum angle of resolution (20/63 Snellen). We identified 4 clusters of eyes. Compared to the other clusters, cluster 1 (n = 16) included younger subjects, VA greater than that of clusters 2 and 3, normal or moderately low total macular volume (TMV), greater preservation of scotopic and photopic ERG responses and fixation stability, less atrophy, and fewer flecks. Cluster 2 (n = 49) differed from cluster 1 mainly with less atrophy and relatively stable fixation. Cluster 3 (n = 10) included older subjects than clusters 1 and 2 and showed the lowest VA, TMV, ERG responses, and fixation stability, with extensive atrophy. Cluster 4 (n = 27) showed better VA, TMV similar to clusters 1 and 2, moderate ERG activity, stable fixation, and moderate-high atrophy and flecks.

Conclusions

Reflecting the phenotypic complexity of STGD, an unsupervised clustering approach incorporating phenotypic measures can be used to categorize STGD eyes into distinct clusters. The clusters exhibit differences in structural and functional measures including quantity of flecks, extent of retinal atrophy, visual fixation accuracy, and ERG responses, among other features. If novel pharmacologic, gene, or cell therapy modalities become available in the future, the multimodal phenomap approach may be useful to individualize treatment decisions, and its utility in aiding prognostication requires further evaluation.

Financial Disclosures

Proprietary or commercial disclosure may be found after the references.

Artificial intelligence in retinal disease: clinical application, challenges, and future directions

Retinal diseases are a leading cause of blindness in developed countries, accounting for the largest share of visually impaired children, working-age adults (inherited retinal disease), and elderly individuals (age-related macular degeneration). These conditions need specialised clinicians to interpret multimodal retinal imaging, with diagnosis and intervention potentially delayed. With an increasing and ageing population, this is becoming a global health priority. One solution is the development of artificial intelligence (AI) software to facilitate rapid data processing. Herein, we review research offering decision support for the diagnosis, classification, monitoring, and treatment of retinal disease using AI. We have prioritised diabetic retinopathy, age-related macular degeneration, inherited retinal disease, and retinopathy of prematurity. There is cautious optimism that these algorithms will be integrated into routine clinical practice to facilitate access to vision-saving treatments, improve efficiency of healthcare systems, and assist clinicians in processing the ever-increasing volume of multimodal data, thereby also liberating time for doctor-patient interaction and co-development of personalised management plans.

Automated Segmentation of Autofluorescence Lesions in Stargardt Disease

OBJECTIVE

To train a deep learning (DL) algorithm to perform fully automated semantic segmentation of multiple autofluorescence lesion types in Stargardt disease.

DESIGN

Cross-sectional study with retrospective imaging data.

SUBJECTS

The study included 193 images from 193 eyes of 97 patients with Stargardt disease.

METHODS

Fundus autofluorescence images obtained from patient visits between 2013 and 2020 were annotated with ground-truth labels. Model training and evaluation were performed using fivefold cross-validation.

MAIN OUTCOMES MEASURES

Dice similarity coefficients, intraclass correlation coefficients, and Bland-Altman analyses comparing algorithm-predicted and grader-labeled segmentations.

RESULTS

The overall Dice similarity coefficient across all lesion classes was 0.78 (95% confidence interval [CI], 0.69-0.86). Dice coefficients were 0.90 (95% CI, 0.85-0.94) for areas of definitely decreased autofluorescence (DDAF), 0.55 (95% CI, 0.35-0.76) for areas of questionably decreased autofluorescence (QDAF), and 0.88 (95% CI, 0.73-1.00) for areas of abnormal background autofluorescence (ABAF). Intraclass correlation coefficients comparing the ground-truth and automated methods were 0.997 (95% CI, 0.996-0.998) for DDAF, 0.863 (95% CI, 0.823-0.895) for QDAF, and 0.974 (95% CI, 0.966-0.980) for ABAF.

CONCLUSIONS

A DL algorithm performed accurate segmentation of autofluorescence lesions in Stargardt disease, demonstrating the feasibility of fully automated segmentation as an alternative to manual or semiautomated labeling methods.

Automated segmentation and feature discovery of age-related macular degeneration and Stargardt disease via self-attended neural networks

Age-related macular degeneration (AMD) and Stargardt disease are the leading causes of blindness for the elderly and young adults respectively. Geographic atrophy (GA) of AMD and Stargardt atrophy are their end-stage outcomes. Efficient methods for segmentation and quantification of these atrophic lesions are critical for clinical research. In this study, we developed a deep convolutional neural network (CNN) with a trainable self-attended mechanism for accurate GA and Stargardt atrophy segmentation. Compared with traditional post-hoc attention mechanisms which can only visualize CNN features, our self-attended mechanism is embedded in a fully convolutional network and directly involved in training the CNN to actively attend key features for enhanced algorithm performance. We applied the self-attended CNN on the segmentation of AMD and Stargardt atrophic lesions on fundus autofluorescence (FAF) images. Compared with a preexisting regular fully convolutional network (the U-Net), our self-attended CNN achieved 10.6% higher Dice coefficient and 17% higher IoU (intersection over union) for AMD GA segmentation, and a 22% higher Dice coefficient and a 32% higher IoU for Stargardt atrophy segmentation. With longitudinal image data having over a longer time, the developed self-attended mechanism can also be applied on the visual discovery of early AMD and Stargardt features.

Deep learning: applications in retinal and optic nerve diseases

Deep learning (DL) represents a paradigm-shifting, burgeoning field of research with emerging clinical applications in optometry. Unlike traditional programming, which relies on human-set specific rules, DL works by exposing the algorithm to a large amount of annotated data and allowing the software to develop its own set of rules (i.e. learn) by adjusting the parameters inside the model (network) during a training process in order to complete the task on its own. One major limitation of traditional programming is that, with complex tasks, it may require an extensive set of rules to accurately complete the assignment. Additionally, traditional programming can be susceptible to human bias from programmer experience. With the dramatic increase in the amount and the complexity of clinical data, DL has been utilised to automate data analysis and thus to assist clinicians in patient management. This review will present the latest advances in DL, for managing posterior eye diseases as well as DL-based solutions for patients with vision loss.

A Systematic Review of Artificial Intelligence Applications Used for Inherited Retinal Disease Management

Nowadays, Artificial Intelligence (AI) and its subfields, Machine Learning (ML) and Deep Learning (DL), are used for a variety of medical applications. It can help clinicians track the patient’s illness cycle, assist with diagnosis, and offer appropriate therapy alternatives. Each approach employed may address one or more AI problems, such as segmentation, prediction, recognition, classification, and regression. However, the amount of AI-featured research on Inherited Retinal Diseases (IRDs) is currently limited. Thus, this study aims to examine artificial intelligence approaches used in managing Inherited Retinal Disorders, from diagnosis to treatment. A total of 20,906 articles were identified using the Natural Language Processing (NLP) method from the IEEE Xplore, Springer, Elsevier, MDPI, and PubMed databases, and papers submitted from 2010 to 30 October 2021 are included in this systematic review. The resultant study demonstrates the AI approaches utilized on images from different IRD patient categories and the most utilized AI architectures and models with their imaging modalities, identifying the main benefits and challenges of using such methods.

Kinetics of Heterogeneous Background in Stargardt’s Disease over Time

Stargardt’s disease (STGD1) is caused by mutations in the ABCA4 gene. Different lesions characterised by decreased autofluorescence levels are found in fundus autofluorescence (FAF) from STGD1 patients and could be used as outcome indicators for disease progression. We investigated the fate of foci with reduced autofluorescence (FRA) within the heterogeneous background of STGD1 patients using FAF imaging. Genetically confirmed STGD1 patients presenting heterogeneous background autofluorescence on high-quality FAF images at a minimum of two visits at least 12 months apart were chosen. A grid centred on the fovea was used to define five different zones. Within each zone, five FRA were randomly selected for each eye. The eccentricity of foci was determined at different time points for each patient. Analysis of 175 randomly chosen FRA showed consistent centrifugal displacement over time, most notably in eyes showing areas with definitely decreased autofluorescence. Interestingly, FRA did not leave an area of hypo-autofluorescence on FAF in locations where they were previously located. These findings may help to better understand STGD1 progression, improve FAF interpretation, and shed light on the nature of heterogeneous background.

Stargardt disease and progress in therapeutic strategies

Background: Stargardt disease (STGD1) is an autosomal recessive retinal dystrophy due to mutations in ABCA4, characterized by subretinal deposition of lipofuscin-like substances and bilateral centrifugal vision loss. Despite the tremendous progress made in the understanding of STGD1, there are no approved treatments to date. This review examines the challenges in the development of an effective STGD1 therapy.Materials and Methods: A literature review was performed through to June 2021 summarizing the spectrum of retinal phenotypes in STGD1, the molecular biology of ABCA4 protein, the in vivo and in vitro models used to investigate the mechanisms of ABCA4 mutations and current clinical trials.Results: STGD1 phenotypic variability remains an challenge for clinical trial design and patient selection. Pre-clinical development of therapeutic options has been limited by the lack of animal models reflecting the diverse phenotypic spectrum of STDG1. Patient-derived cell lines have facilitated the characterization of splice mutations but the clinical presentation is not always predicted by the effect of specific mutations on retinoid metabolism in cellular models. Current therapies primarily aim to delay vision loss whilst strategies to restore vision are less well developed.Conclusions: STGD1 therapy development can be accelerated by a deeper understanding of genotype-phenotype correlations.

Stargardt disease: Multimodal imaging: A review

Stargardt disease (STGD1) is an autosomal recessive retinal dystrophy, characterised by bilateral progressive central vision loss and subretinal deposition of lipofuscin-like substances. Recent advances in molecular diagnosis and therapeutic options are complemented by the increasing recognition of new multimodal imaging biomarkers that may predict genotype and disease progression. Unique non-invasive imaging features of STDG1 are useful for gene variant interpretation and may even provide insight into the underlying molecular pathophysiology. In addition, pathognomonic imaging features of STGD1 have been used to train neural networks to improve time efficiency in lesion segmentation and disease progression measurements. This review will discuss the role of key imaging modalities, correlate imaging signs across varied STGD1 presentations and illustrate the use of multimodal imaging as an outcome measure in determining the efficacy of emerging STGD1 specific therapies.

Next-Generation Sequencing Applications for Inherited Retinal Diseases

Inherited retinal diseases (IRDs) represent a collection of phenotypically and genetically diverse conditions. IRDs phenotype(s) can be isolated to the eye or can involve multiple tissues. These conditions are associated with diverse forms of inheritance, and variants within the same gene often can be associated with multiple distinct phenotypes. Such aspects of the IRDs highlight the difficulty met when establishing a genetic diagnosis in patients. Here we provide an overview of cutting-edge next-generation sequencing techniques and strategies currently in use to maximise the effectivity of IRD gene screening. These techniques have helped researchers globally to find elusive causes of IRDs, including copy number variants, structural variants, new IRD genes and deep intronic variants, among others. Resolving a genetic diagnosis with thorough testing enables a more accurate diagnosis and more informed prognosis and should also provide information on inheritance patterns which may be of particular interest to patients of a child-bearing age. Given that IRDs are heritable conditions, genetic counselling may be offered to help inform family planning, carrier testing and prenatal screening. Additionally, a verified genetic diagnosis may enable access to appropriate clinical trials or approved medications that may be available for the condition.

Artificial Intelligence (AI) Applications for Age-Related Macular Degeneration (AMD) and Other Retinal Dystrophies

Artificial intelligence (AI), with its subdivisions (machine and deep learning), is a new branch of computer science that has shown impressive results across a variety of domains. The applications of AI to medicine and biology are being widely investigated. Medical specialties that rely heavily on images, including radiology, dermatology, oncology and ophthalmology, were the first to explore AI approaches in analysis and diagnosis. Applications of AI in ophthalmology have concentrated on diseases with high prevalence, such as diabetic retinopathy, retinopathy of prematurity, age-related macular degeneration (AMD), and glaucoma. Here we provide an overview of AI applications for diagnosis, classification, and clinical management of AMD and other macular dystrophies.