Quantification of Key Retinal Features in Early and Late Age-Related Macular Degeneration Using Deep Learning

Liefers, Bart; Taylor, Paul; Alsaedi, Abdulrahman; Bailey, Clare; Balaskas, Konstantinos; Dhingra, Narendra; Egan, Catherine; Rodrigues, Filipa Gomes; Gonzalo, Cristina González; Heeren, Tjebo; Lotery, Andrew; Müller, Philipp L.; Olvera-Barrios, Abraham; Paul, Bobby; Schwartz, Roy; Thomas, Darren; Warwick, Alasdair; Tufail, Adnan; Sánchez, Clara I.

doi:10.1016/j.ajo.2020.12.034

Cited by 46 publications

(34 citation statements)

References 27 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In this context, previous studies reported on machine and deep learning (DL)–based methods in eyes with early and intermediate AMD to reliably detect structural biomarkers as lesions of hyperreflective foci (HRF) and RPD, to quantify drusen volumes, and to estimate the individual's eye risk toward disease progression. 27 – 29 In another study, Liefers et al 30 even revealed findings on DL-based models exceeding the human performance on the detection of neovascular AMD-associated structural biomarkers. Interestingly, in the same data set, the accuracy rate for detecting drusen and drusenoid PEDs was slightly higher for readers than for the model.…”

Section: Discussionmentioning

confidence: 97%

“…Interestingly, in the same data set, the accuracy rate for detecting drusen and drusenoid PEDs was slightly higher for readers than for the model. 30 However, further studies validating machine learning approaches to differentiate disease phenotypes in multimodal retinal imaging as well as to detect early signs for disease progression as, for example, lesions of iRORA in eyes with iAMD are currently still ongoing. The availability of standardized gradings on various structural biomarkers as well as multimodal retinal imaging data within the MACUSTAR study can form the basis for the development of further automated grading approaches, which will also help to grade the increasing amount of multimodal retinal imaging data sets available in currently ongoing and future multicenter, observational clinical iAMD trials.…”

Section: Discussionmentioning

confidence: 99%

See 1 more Smart Citation

Intersession Repeatability of Structural Biomarkers in Early and Intermediate Age-Related Macular Degeneration: A MACUSTAR Study Report

Saßmannshausen

Thiele

Behning

et al. 2022

Trans. Vis. Sci. Tech.

View full text Add to dashboard Cite

Purpose To analyze the intersession repeatability of structural biomarkers in eyes with early and intermediate age-related macular degeneration (iAMD) within the cross-sectional part of the observational multicenter MACUSTAR study. Methods Certified site personnel obtained multimodal imaging data at two visits (38 ± 20 [mean ± standard deviation] days apart), including spectral-domain optical coherence tomography (SD-OCT). One junior reader performed systematic and blinded grading at the central reading center, followed by senior reader review. Structural biomarkers included maximum drusen size classification (>63 to ≤125 µm vs. >125 µm), presence of large pigment epithelium detachments (PEDs), reticular pseudodrusen (RPD), vitelliform lesions, and refractile deposits. Intrasession variability was assessed using Cohen's κ statistics. Results At the first visit, 202 study eyes of 202 participants were graded as manifesting with either early ( n = 34) or intermediate ( n = 168) AMD. Grading of imaging data between visits revealed perfect agreement for the maximum drusen size classification (κ = 0.817; 95% confidence interval, 0.70–0.94). In iAMD eyes, perfect to substantial agreement was determined for the presence of large PEDs (0.87; 0.69–1.00) and RPD (0.752; 0.63–0.87), while intersession agreement was lower for the presence of vitelliform lesions (0.649; 0.39–0.65) and refractile deposits (0.342; −0.029–0.713), respectively. Conclusions Multimodal retinal imaging analysis between sessions showed a higher repeatability for structural biomarkers with predefined cutoff values than purely qualitative defined parameters. Translational Relevance A high repeatability of retinal imaging biomarkers will be important to implement automatic grading approaches and to establish robust and meaningful structural clinical endpoints for future interventional clinical trials in patients with iAMD.

show abstract

Section: Discussionmentioning

confidence: 97%

Section: Discussionmentioning

confidence: 99%

Intersession Repeatability of Structural Biomarkers in Early and Intermediate Age-Related Macular Degeneration: A MACUSTAR Study Report

Saßmannshausen

Thiele

Behning

et al. 2022

Trans. Vis. Sci. Tech.

View full text Add to dashboard Cite

show abstract

“… 15 , 18 , 19 However, because the calculation for DSC is highly dependent on the area of interest, 15 , 16 it becomes more difficult to achieve higher DSCs on smaller structures such as the retina. As examples, Deeley et al 20 reported average DSCs of 0.4 to 0.5 when segmenting nerves and the optic chiasm on brain magnetic resonance imaging (MRI) scans, whereas Liefers et al 21 reported a mean DSC of 0.6 when developing an algorithm to segment features of age-related macular degeneration on OCT. Additionally, our segmentation was more challenging as the leakage boundary was more ambiguous compared to boundaries for other forms of imaging.…”

Section: Conclusion/discussionmentioning

confidence: 99%

“…However, we do note that even between expert human graders, we measured considerable, and higher, inter-rater reliability. Others have also observed intra-grader variability in FA 1 and more easily interpretable images such as OCT. 21 This discordance between clinicians could have serious impacts on patient care, and thus a more objective tool to evaluate FAs has great clinical importance.…”

Section: Conclusion/discussionmentioning

confidence: 99%

Automated Detection of Vascular Leakage in Fluorescein Angiography – A Proof of Concept

Young

Kim

Yakin

et al. 2022

Trans. Vis. Sci. Tech.

View full text Add to dashboard Cite

Purpose The purpose of this paper was to develop a deep learning algorithm to detect retinal vascular leakage (leakage) in fluorescein angiography (FA) of patients with uveitis and use the trained algorithm to determine clinically notable leakage changes. Methods An algorithm was trained and tested to detect leakage on a set of 200 FA images (61 patients) and evaluated on a separate 50-image test set (21 patients). The ground truth was leakage segmentation by two clinicians. The Dice Similarity Coefficient (DSC) was used to measure concordance. Results During training, the algorithm achieved a best average DSC of 0.572 (95% confidence interval [CI] = 0.548–0.596). The trained algorithm achieved a DSC of 0.563 (95% CI = 0.543–0.582) when tested on an additional set of 50 images. The trained algorithm was then used to detect leakage on pairs of FA images from longitudinal patient visits. Longitudinal leakage follow-up showed a >2.21% change in the visible retina area covered by leakage (as detected by the algorithm) had a sensitivity and specificity of 90% (area under the curve [AUC] = 0.95) of detecting a clinically notable change compared to the gold standard, an expert clinician's assessment. Conclusions This deep learning algorithm showed modest concordance in identifying vascular leakage compared to ground truth but was able to aid in identifying vascular FA leakage changes over time. Translational Relevance This is a proof-of-concept study that vascular leakage can be detected in a more standardized way and that tools can be developed to help clinicians more objectively compare vascular leakage between FAs.

show abstract

“…This methodology has been applied to investigating imaging biomarkers and visual outcomes [ 24 ▪ , 26 ▪▪ ]. Following this, another group developed a novel automated segmentation model using a convolutional neural network [ 27 ▪ ]. This model was built using a large, real-world electronic medical records-based dataset from the United Kingdom, annotated by clinical experts with 13 of the most common AMD biomarkers on OCT, including IRF, SRF, and pigment epithelial detachment (PED) [ 27 ▪ ].…”

Section: Development and Application Of Artificial Intelligence Models For Neovascular Age-related Macular Degenerationmentioning

confidence: 99%

Artificial intelligence-based predictions in neovascular age-related macular degeneration

Ferrara¹,

Newton²,

Lee

2021

Current Opinion in Ophthalmology

View full text Add to dashboard Cite

Purpose of review Predicting treatment response and optimizing treatment regimen in patients with neovascular age-related macular degeneration (nAMD) remains challenging. Artificial intelligence-based tools have the potential to increase confidence in clinical development of new therapeutics, facilitate individual prognostic predictions, and ultimately inform treatment decisions in clinical practice. Recent findings To date, most advances in applying artificial intelligence to nAMD have focused on facilitating image analysis, particularly for automated segmentation, extraction, and quantification of imaging-based features from optical coherence tomography (OCT) images. No studies in our literature search evaluated whether artificial intelligence could predict the treatment regimen required for an optimal visual response for an individual patient. Challenges identified for developing artificial intelligence-based models for nAMD include the limited number of large datasets with high-quality OCT data, limiting the patient populations included in model development; lack of counterfactual data to inform how individual patients may have fared with an alternative treatment strategy; and absence of OCT data standards, impairing the development of models usable across devices. Summary Artificial intelligence has the potential to enable powerful prognostic tools for a complex nAMD treatment landscape; however, additional work remains before these tools are applicable to informing treatment decisions for nAMD in clinical practice.

show abstract

Quantification of Key Retinal Features in Early and Late Age-Related Macular Degeneration Using Deep Learning

Cited by 46 publications

References 27 publications

Intersession Repeatability of Structural Biomarkers in Early and Intermediate Age-Related Macular Degeneration: A MACUSTAR Study Report

Intersession Repeatability of Structural Biomarkers in Early and Intermediate Age-Related Macular Degeneration: A MACUSTAR Study Report

Automated Detection of Vascular Leakage in Fluorescein Angiography – A Proof of Concept

Artificial intelligence-based predictions in neovascular age-related macular degeneration

Contact Info

Product

Resources

About