Semi-automatic geographic atrophy segmentation for SD-OCT images

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.

Deep-GA-Net for Accurate and Explainable Detection of Geographic Atrophy on OCT Scans

Objective

To propose Deep-GA-Net, a 3-dimensional (3D) deep learning network with 3D attention layer, for the detection of geographic atrophy (GA) on spectral domain OCT (SD-OCT) scans, explain its decision making, and compare it with existing methods.

Design

Deep learning model development.

Participants

Three hundred eleven participants from the Age-Related Eye Disease Study 2 Ancillary SD-OCT Study.

Methods

A dataset of 1284 SD-OCT scans from 311 participants was used to develop Deep-GA-Net. Cross-validation was used to evaluate Deep-GA-Net, where each testing set contained no participant from the corresponding training set. En face heatmaps and important regions at the B-scan level were used to visualize the outputs of Deep-GA-Net, and 3 ophthalmologists graded the presence or absence of GA in them to assess the explainability (i.e., understandability and interpretability) of its detections.

Main Outcome Measures

Accuracy, area under receiver operating characteristic curve (AUC), area under precision-recall curve (APR).

Results

Compared with other networks, Deep-GA-Net achieved the best metrics, with accuracy of 0.93, AUC of 0.94, and APR of 0.91, and received the best gradings of 0.98 and 0.68 on the en face heatmap and B-scan grading tasks, respectively.

Conclusions

Deep-GA-Net was able to detect GA accurately from SD-OCT scans. The visualizations of Deep-GA-Net were more explainable, as suggested by 3 ophthalmologists. The code and pretrained models are publicly available at https://github.com/ncbi/Deep-GA-Net.

Financial Disclosures

The author(s) have no proprietary or commercial interest in any materials discussed in this article.

Compass Fundus-Guided Perimetry in Geographic Atrophy

Purpose

To evaluate compass (CMP), a recently introduced device that combines scanning ophthalmoscopy, automated perimetry, and eye tracking, for fundus-guided perimetry (microperimetry) with the purpose of correlating perimetric retinal sensitivity (PRS) and retinal geographic atrophy (GA) features.

Materials and Methods

A retrospective, cross-sectional study was performed in 56 eyes of 43 patients affected by GA. All patients underwent compass 10-2 perimetry, consisting of a full-threshold visual field on fundus photography and an infrared (IR) image of the central 30° of the retina. Data were exported to an Excel sheet. Binarization with black/white (B/W) variables was applied on the compass photo fundus and matched with visual field scores. Patients underwent autofluorescence (AF) and IR images (Heidelberg, Germany): CMP and Heidelberg IR images were homologated by using GIMP software (https://www.gimp.org), and then atrophic areas were manually measured with the ImageJ program. CMP perimetric grid was overlapped with AF and IR pictures by using GIMP, obtaining composite TIFF images, which were then analyzed with the ImageJ greyscale score (GSS) tool. A hyperautofluorescent halo was identified on the GA edges of some patients. Pearson's correlation between GA size on IR compass and IR Heidelberg and between GSS and PRS values has been calculated; the independent t-test was realized to calculate the correlation between GSS and B/W variables identified on the CMP photo fundus. The Spearman correlation between total deviation and pattern deviation was calculated.

Results

The AUC-ROC score between CMP scores and B/W variables was 93,4%. The Spearman correlation between total deviation and pattern deviation was highly significant (p = 0,00). The correlation between AF GSS values and PRS was significant (p value = 0,00), the correlation between GSS of hyperautofluorescent points and PRS was significant (p value = 0,00), and the correlation between IR GSS and PRS was significant (p value = 0,00). The correlation between AF GSS and B/W variables was significant (p value = 0,002), the correlation between hyperautofluorescent points and B/W was not significant (p value = 0,40), and the correlation between IR GSS and B/W was significant (p = 0,00).

Conclusions

Based on our preliminary results, compass seems to be a reliable, quick, and safe device for the anatomical and functional study of GA. The direct visualization of the visual field on the fundus photography as a background allows a precise assessment and clinical monitoring of this disease.

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.

Clinically relevant deep learning for detection and quantification of geographic atrophy from optical coherence tomography: a model development and external validation study

BACKGROUND

Geographic atrophy is a major vision-threatening manifestation of age-related macular degeneration, one of the leading causes of blindness globally. Geographic atrophy has no proven treatment or method for easy detection. Rapid, reliable, and objective detection and quantification of geographic atrophy from optical coherence tomography (OCT) retinal scans is necessary for disease monitoring, prognostic research, and to serve as clinical endpoints for therapy development. To this end, we aimed to develop and validate a fully automated method to detect and quantify geographic atrophy from OCT.

METHODS

We did a deep-learning model development and external validation study on OCT retinal scans at Moorfields Eye Hospital Reading Centre and Clinical AI Hub (London, UK). A modified U-Net architecture was used to develop four distinct deep-learning models for segmentation of geographic atrophy and its constituent retinal features from OCT scans acquired with Heidelberg Spectralis. A manually segmented clinical dataset for model development comprised 5049 B-scans from 984 OCT volumes selected randomly from 399 eyes of 200 patients with geographic atrophy secondary to age-related macular degeneration, enrolled in a prospective, multicentre, phase 2 clinical trial for the treatment of geographic atrophy (FILLY study). Performance was externally validated on an independently recruited dataset from patients receiving routine care at Moorfields Eye Hospital (London, UK). The primary outcome was segmentation and classification agreement between deep-learning model geographic atrophy prediction and consensus of two independent expert graders on the external validation dataset.

FINDINGS

The external validation cohort included 884 B-scans from 192 OCT volumes taken from 192 eyes of 110 patients as part of real-life clinical care at Moorfields Eye Hospital between Jan 1, 2016, and Dec, 31, 2019 (mean age 78·3 years [SD 11·1], 58 [53%] women). The resultant geographic atrophy deep-learning model produced predictions similar to consensus human specialist grading on the external validation dataset (median Dice similarity coefficient [DSC] 0·96 [IQR 0·10]; intraclass correlation coefficient [ICC] 0·93) and outperformed agreement between human graders (DSC 0·80 [0·28]; ICC 0·79). Similarly, the three independent feature-specific deep-learning models could accurately segment each of the three constituent features of geographic atrophy: retinal pigment epithelium loss (median DSC 0·95 [IQR 0·15]), overlying photoreceptor degeneration (0·96 [0·12]), and hypertransmission (0·97 [0·07]) in the external validation dataset versus consensus grading.

INTERPRETATION

We present a fully developed and validated deep-learning composite model for segmentation of geographic atrophy and its subtypes that achieves performance at a similar level to manual specialist assessment. Fully automated analysis of retinal OCT from routine clinical practice could provide a promising horizon for diagnosis and prognosis in both research and real-life patient care, following further clinical validation FUNDING: Apellis Pharmaceuticals.

Improving Interpretability in Machine Diagnosis

Purpose

Manually identifying geographic atrophy (GA) presence and location on OCT volume scans can be challenging and time consuming. This study developed a deep learning model simultaneously (1) to perform automated detection of GA presence or absence from OCT volume scans and (2) to provide interpretability by demonstrating which regions of which B-scans show GA.

Design

Med-XAI-Net, an interpretable deep learning model was developed to detect GA presence or absence from OCT volume scans using only volume scan labels, as well as to interpret the most relevant B-scans and B-scan regions.

Participants

One thousand two hundred eighty-four OCT volume scans (each containing 100 B-scans) from 311 participants, including 321 volumes with GA and 963 volumes without GA.

Methods

Med-XAI-Net simulates the human diagnostic process by using a region-attention module to locate the most relevant region in each B-scan, followed by an image-attention module to select the most relevant B-scans for classifying GA presence or absence in each OCT volume scan. Med-XAI-Net was trained and tested (80% and 20% participants, respectively) using gold standard volume scan labels from human expert graders.

Main Outcome Measures

Accuracy, area under the receiver operating characteristic (ROC) curve, F₁ score, sensitivity, and specificity.

Results

In the detection of GA presence or absence, Med-XAI-Net obtained superior performance (91.5%, 93.5%, 82.3%, 82.8%, and 94.6% on accuracy, area under the ROC curve, F₁ score, sensitivity, and specificity, respectively) to that of 2 other state-of-the-art deep learning methods. The performance of ophthalmologists grading only the 5 B-scans selected by Med-XAI-Net as most relevant (95.7%, 95.4%, 91.2%, and 100%, respectively) was almost identical to that of ophthalmologists grading all volume scans (96.0%, 95.7%, 91.8%, and 100%, respectively). Even grading only 1 region in 1 B-scan, the ophthalmologists demonstrated moderately high performance (89.0%, 87.4%, 77.6%, and 100%, respectively).

Conclusions

Despite using ground truth labels during training at the volume scan level only, Med-XAI-Net was effective in locating GA in B-scans and selecting relevant B-scans within each volume scan for GA diagnosis. These results illustrate the strengths of Med-XAI-Net in interpreting which regions and B-scans contribute to GA detection in the volume scan.

Deep Learning Applied to Automated Segmentation of Geographic Atrophy in Fundus Autofluorescence Images

Purpose

This study describes the development of a deep learning algorithm based on the U-Net architecture for automated segmentation of geographic atrophy (GA) lesions in fundus autofluorescence (FAF) images.

Methods

Image preprocessing and normalization by modified adaptive histogram equalization were used for image standardization to improve effectiveness of deep learning. A U-Net-based deep learning algorithm was developed and trained and tested by fivefold cross-validation using FAF images from clinical datasets. The following metrics were used for evaluating the performance for lesion segmentation in GA: dice similarity coefficient (DSC), DSC loss, sensitivity, specificity, mean absolute error (MAE), accuracy, recall, and precision.

Results

In total, 702 FAF images from 51 patients were analyzed. After fivefold cross-validation for lesion segmentation, the average training and validation scores were found for the most important metric, DSC (0.9874 and 0.9779), for accuracy (0.9912 and 0.9815), for sensitivity (0.9955 and 0.9928), and for specificity (0.8686 and 0.7261). Scores for testing were all similar to the validation scores. The algorithm segmented GA lesions six times more quickly than human performance.

Conclusions

The deep learning algorithm can be implemented using clinical data with a very high level of performance for lesion segmentation. Automation of diagnostics for GA assessment has the potential to provide savings with respect to patient visit duration, operational cost and measurement reliability in routine GA assessments.

Translational Relevance

A deep learning algorithm based on the U-Net architecture and image preprocessing appears to be suitable for automated segmentation of GA lesions on clinical data, producing fast and accurate results.

An integrated time adaptive geographic atrophy prediction model for SD-OCT images

The automated prediction of geographic atrophy (GA) lesion growth can help ophthalmologists understand how the GA progresses, and assess the efficiency of current treatment and the prognosis of the disease. We developed an integrated time adaptive prediction model for identifying the location of future GA growth. The proposed model was comprised of bi-directional long short-term memory (BiLSTM) network-based prediction module and convolutional neural network (CNN)-based refinement module. Considering the discontinuity of time intervals among sequential follow-up visits, we integrated time factors into BiLSTM-based prediction module to control the time attribute expediently. Then, the results from prediction module were refined by a CNN-based strategy to obtain the final locations of future GA growth. The 10 scenarios were designed to evaluate the prediction accuracy of our proposed model. The 1-6th scenarios demonstrated the importance of the prior information similarity, the 7-8th scenarios verified the effect of time factors and refinement methods respectively and the 9th scenario compared the prediction results between those using a single follow-up visit for training and using 2 sequential follow-up visits for training. The 10th scenario showed the model generalization performance across regions. The average dice indexes (DI) of the predicted GA regions in the 1-6th scenarios are 0.86, 0.89, 0.89, 0.92 and 0.88, 0.90, respectively. By integrating time factors to the BiLSTM models, the prediction accuracy was improved by almost 10%. The CNN-based refinement strategy can remove the wrong GA regions effectively to preserve the actual GA regions and improve the prediction accuracy further. The prediction results based on 2 sequential follow-up visits showed higher correlations than that based on single follow-up visit. The proposed model presented a good generalization performance while training patients and testing patients were from different regions. Experimental results demonstrated the importance of prior information to the prediction accuracy. We demonstrate the feasibility of creating a model for disease prediction.

Artificial Intelligence Algorithms for Analysis of Geographic Atrophy: A Review and Evaluation

Purpose

The purpose of this study was to summarize and evaluate artificial intelligence (AI) algorithms used in geographic atrophy (GA) diagnostic processes (e.g. isolating lesions or disease progression).

Methods

The search strategy and selection of publications were both conducted in accordance with the Preferred of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. PubMed and Web of Science were used to extract literary data. The algorithms were summarized by objective, performance, and scope of coverage of GA diagnosis (e.g. lesion automation and GA progression).

Results

Twenty-seven studies were identified for this review. A total of 18 publications focused on lesion segmentation only, 2 were designed to detect and classify GA, 2 were designed to predict future overall GA progression, 3 focused on prediction of future spatial GA progression, and 2 focused on prediction of visual function in GA. GA-related algorithms reported sensitivities from 0.47 to 0.98, specificities from 0.73 to 0.99, accuracies from 0.42 to 0.995, and Dice coefficients from 0.66 to 0.89.

Conclusions

Current GA-AI publications have a predominant focus on lesion segmentation and a minor focus on classification and progression analysis. AI could be applied to other facets of GA diagnoses, such as understanding the role of hyperfluorescent areas in GA. Using AI for GA has several advantages, including improved diagnostic accuracy and faster processing speeds.

Translational Relevance

AI can be used to quantify GA lesions and therefore allows one to impute visual function and quality-of-life. However, there is a need for the development of reliable and objective models and software to predict the rate of GA progression and to quantify improvements due to interventions.

Visualizing and understanding inherent features in SD-OCT for the progression of age-related macular degeneration using deconvolutional neural networks

To develop a convolutional neural network visualization strategy so that optical coherence tomography (OCT) features contributing to the evolution of age-related macular degeneration (AMD) can be better determined. We have trained a U-Net model to utilize baseline OCT to predict the progression of geographic atrophy (GA), a late stage manifestation of AMD. We have augmented the U-Net architecture by attaching deconvolutional neural networks (deconvnets). Deconvnets produce the reconstructed feature maps and provide an indication regarding the inherent baseline OCT features contributing to GA progression. Experiments were conducted on longitudinal spectral domain (SD)-OCT and fundus autofluorescence images collected from 70 eyes with GA. The intensity of Bruch's membrane-outer choroid (BMChoroid) retinal junction exhibited a relative importance of 24%, in the GA progression. The intensity of the inner retinal pigment epithelium (RPE) and BM junction (InRPEBM) showed a relative importance of 22%. BMChoroid (where the AMD feature/damage of choriocapillaris was included) followed by InRPEBM (where the AMD feature/damage of RPE was included) are the layers which appear to be most relevant in predicting the progression of AMD.

Automatic Quantification Software for Geographic Atrophy Associated with Age-Related Macular Degeneration: A Validation Study

Aims

To determine the accuracy and repeatability of new software to automatically quantify GA areas associated to age-related macular degeneration (AMD) by swept-source optical coherence tomography (SS-OCT). Settings and Design. Tertiary referral hospital in Spain. Cross-sectional and noninterventional. Methods and Material. Forty-six eyes from 33 AMD patients with GA, without previous choroidal neovascularization, were scanned using a SS-OCT (Topcon Corporation, Japan), including three consecutive 7 × 7 mm OCT scans. Three independent masked observers manually measured the GA area using FAF images. These measures were compared to the three automatic determinations of the GA. Lesions were classified according to their morphology and number as regular/irregular and single/multiple. Statistical Analysis Used. Intraclass correlation coefficients (ICCs) were estimated to study the agreement between the three physicians in manual measurements. ICC through a two-way mixed effects model was used for the software measures, and Lin's concordance correlation coefficient (CCC) was used to analyse the agreement between the physicians and the software.

Results

The mean age was 76.3 ± 11.7 years. Eighteen cases showed regular lesions, and 30 showed single lesions. The CCC between manual and automatic measures was 0.95 for the whole sample. The CCC for the area according to the lesion type was 0.92 and 0.97; it was 0.99 for single lesions and 0.89 for multiple lesions. The ICC between the three physicians was 0.94 for the whole sample and 0.88 in multiple lesions. The ICC between the three automatic measures for the area was 0.98 for the whole sample, regular or irregular lesions, and single or multiple lesions.

Conclusions

The accuracy of this new software is substantial for the area with a high degree of repeatability agreement, being very precise in single lesions.

MS-CAM: Multi-Scale Class Activation Maps for Weakly-Supervised Segmentation of Geographic Atrophy Lesions in SD-OCT Images

As one of the most critical characteristics in advanced stage of non-exudative Age-related Macular Degeneration (AMD), Geographic Atrophy (GA) is one of the significant causes of sustained visual acuity loss. Automatic localization of retinal regions affected by GA is a fundamental step for clinical diagnosis. In this paper, we present a novel weakly supervised model for GA segmentation in Spectral-Domain Optical Coherence Tomography (SD-OCT) images. A novel Multi-Scale Class Activation Map (MS-CAM) is proposed to highlight the discriminatory significance regions in localization and detail descriptions. To extract available multi-scale features, we design a Scaling and UpSampling (SUS) module to balance the information content between features of different scales. To capture more discriminative features, an Attentional Fully Connected (AFC) module is proposed by introducing the attention mechanism into the fully connected operations to enhance the significant informative features and suppress less useful ones. Based on the location cues, the final GA region prediction is obtained by the projection segmentation of MS-CAM. The experimental results on two independent datasets demonstrate that the proposed weakly supervised model outperforms the conventional GA segmentation methods and can produce similar or superior accuracy comparing with fully supervised approaches. The source code has been released and is available on GitHub: https://github.com/ jizexuan/Multi-Scale-Class-Activation-Map-Tensorflow.

Atlas of Human Retinal Pigment Epithelium Organelles Significant for Clinical Imaging

Purpose

To quantify organelles impacting imaging in the cell body and intact apical processes of human retinal pigment epithelium (RPE), including melanosomes, lipofuscin-melanolipofuscin (LM), mitochondria, and nuclei.

Methods

A normal perifovea of a 21-year-old white male was preserved after rapid organ recovery. An aligned image stack was generated using serial block-face scanning electron microscopy and was annotated by expert readers (TrakEM, ImageJ). Acquired measures included cell body and nuclear volume (n = 17); organelle count in apical processes (n = 17) and cell bodies (n = 8); distance of cell body organelles along a normalized apical-basal axis (n = 8); and dimensions of organelle-bounding boxes in apical processes in selected subsamples of cell bodies and apical processes.

Results

In 2661 sections through 17 cells, apical processes contained 65 ± 24 melanosomes in mononucleate (n = 15) and 131 ± 28 in binucleate cells (n = 2). Cell bodies contained 681 ± 153 LM and 734 ± 170 mitochondria. LM was excluded from the basal quartile, and mitochondria from the apical quartile. Lengths of melanosomes, LM, and mitochondria, respectively were 2305 ± 528, 1320 ± 574, and 1195 ± 294 nm. The ratio of cell body to nucleus volume was 4.6 ± 0.4. LM and mitochondria covered 75% and 63%, respectively, of the retinal imaging plane.

Conclusions

Among RPE signal sources for optical coherence tomography, LM and mitochondria are the most numerous reflective cell body organelles. These and our published data show that most melanosomes are in apical processes. Overlapping LM and previously mitochondria cushions may support multiple reflective bands in cell bodies. This atlas of subcellular reflectivity sources can inform development of advanced optical coherence tomography technologies.

Geographic atrophy segmentation in SD-OCT images using synthesized fundus autofluorescence imaging

BACKGROUND AND OBJECTIVE

Accurate assessment of geographic atrophy (GA) is critical for diagnosis and therapy of non-exudative age-related macular degeneration (AMD). Herein, we propose a novel GA segmentation framework for spectral-domain optical coherence tomography (SD-OCT) images that employs synthesized fundus autofluorescence (FAF) images.

METHODS

An en-face OCT image is created via the restricted sub-volume projection of three-dimensional OCT data. A GA region-aware conditional generative adversarial network is employed to generate a plausible FAF image from the en-face OCT image. The network balances the consistency between the entire synthesize FAF image and the lesion. We use a fully convolutional deep network architecture to segment the GA region using the multimodal images, where the features of the en-face OCT and synthesized FAF images are fused on the front-end of the network.

RESULTS

Experimental results for 56 SD-OCT scans with GA indicate that our synthesis algorithm can generate high-quality synthesized FAF images and that the proposed segmentation network achieves a dice similarity coefficient, an overlap ratio, and an absolute area difference of 87.2%, 77.9%, and 11.0%, respectively.

CONCLUSION

We report an automatic GA segmentation method utilizing synthesized FAF images.

SIGNIFICANCE

Our method is effective for multimodal segmentation of the GA region and can improve AMD treatment.

Use of Fundus Autofluorescence Combined with Optical Coherence Tomography for Diagnose of Geographic Atrophy in Age-Related Macular Degeneration

The aim of this study was to demonstrate the sensitivity of Optical coherence tomography (OCT) in detection of geographic atrophy (GA) secondary to exudative age related macular degeneration (AMD). In this retrospective case series study 77 patients (53% female, with mean ± standard deviation [SD] of 82.6±9.3 years) with 97 eyes (45 OS [left eyes]/52 OD [right eyes]) were included. This was a retrospective review of the charts of patients who presented with exudative AMD at the Pitié Salpetrière Hospital, Paris, France, between December 2016 and August 2017 that received intravitreal injections of anti-vascular endothelial growth factor (anti-VEGF) therapies. At baseline, following biomicroscopy examination, multimodal imaging was performed including, fluorescein angiography (FA), fundus auto-fluorescence (FAF), spectral domain optical coherence tomography (SD-OCT) and indocyanine green angiography (ICGA). During the follow-up, SD-OCT with/without FAF and FA were performed for each patient at 6, 12 and 18 months. For investigation of the prevalence of GA in eyes undergoing intravitreal injections with anti-VEGF therapy, FAF and SD-OCT images were qualitatively reviewed by four independent observers (two graders per group). Kappa coefficient of Cohen was calculated to determine agreement between the graders. The kappa coefficient of Cohen, for inter-rater agreement in the evaluation of FAF images was 0.468, indicating a moderate agreement between the first and second raters. Thus, the sensitivity and specificity of FAF for the diagnosis of GA were 70% and 57%, respectively. If atrophy was assessed with SD-OCT image analysis, the kappa coefficient for inter-rater agreement was 0.846, implying an acceptable agreement between both readers. The sensitivity and specificity of SD-OCT were 93% and 58% respectively. In conclusion, SD-OCT image analysis was more sensitive than FAF for identifying GA in patients treated for exudative AMD.

Semi-automated quantification of geographic atrophy with blue-light autofluorescence and spectral-domain optical coherence tomography: a comparison between the region finder and the advanced retinal pigment epithelium tool in the clinical setting

Purpose

To compare inter‐ and intraobserver reliability and intermodality agreement on quantification of geographic atrophy, using two routinely available quantification tools, based on blue‐light fundus autofluorescence (BAF) and spectral‐domain optical coherence tomography (SD‐OCT).

Methods

Quantifications of atrophic lesions within the central 5 mm of 30 eyes from 30 patients (mean age: 76.1 years) were independently performed by two clinicians on BAF images using the region finder (RF; Heidelberg Engineering) and on SD‐OCT using the advanced retinal pigment epithelium tool (ARPET; Carl Zeiss Meditec) at baseline and follow‐up (mean interval: 336 days). Inter‐ and intraobserver reliability was determined by intraclass correlation coefficients (ICC) and Bland–Altmann plots. Additionally, graders rated the experienced difficulty of each measurement.

Results

Intraclass correlation coefficients (ICC) showed excellent inter‐ and intraobserver reliability with values between 0.994 and 0.998 for RF and slightly higher values for ARPET of 0.997 and 0.999. Bland–Altman plots showed smaller variability for ARPET. Mean interobserver differences (95% CI) for size measurements were −0.11 (−0.27; 0.05) (baseline) and −0.05 mm² (−0.18; 0.08) (follow‐up) for RF and −0.04 (−0.14; 0.06) and −0.06 mm² (−0.14; 0.02) for ARPET. Measurements of lesions were on average 0.57 mm² (0.35; 0.79) or 7.6% larger in ARPET. Lesion size between graders did not differ significantly. There was no statistically significant difference in relative enlargement rates between methods. There was poor to moderate agreement between graders when rating the experienced difficulty of each measurement.

Conclusion

Semi‐automated analysis of geographic atrophy with RF and ARPET is equally reliable and reproducible in clinical settings, despite both algorithms require frequent adjustment by users. The ARPET restricts size measurements to the central 5 mm, which limits its ability to fully track GA progression. Results of both tools are not interchangeable as measurements with ARPET result in larger lesion sizes.

Automated detection and classification of early AMD biomarkers using deep learning

Age-related macular degeneration (AMD) affects millions of people and is a leading cause of blindness throughout the world. Ideally, affected individuals would be identified at an early stage before late sequelae such as outer retinal atrophy or exudative neovascular membranes develop, which could produce irreversible visual loss. Early identification could allow patients to be staged and appropriate monitoring intervals to be established. Accurate staging of earlier AMD stages could also facilitate the development of new preventative therapeutics. However, accurate and precise staging of AMD, particularly using newer optical coherence tomography (OCT)-based biomarkers may be time-intensive and requires expert training which may not feasible in many circumstances, particularly in screening settings. In this work we develop deep learning method for automated detection and classification of early AMD OCT biomarker. Deep convolution neural networks (CNN) were explicitly trained for performing automated detection and classification of hyperreflective foci, hyporeflective foci within the drusen, and subretinal drusenoid deposits from OCT B-scans. Numerous experiments were conducted to evaluate the performance of several state-of-the-art CNNs and different transfer learning protocols on an image dataset containing approximately 20000 OCT B-scans from 153 patients. An overall accuracy of 87% for identifying the presence of early AMD biomarkers was achieved.

Beyond Retinal Layers: A Deep Voting Model for Automated Geographic Atrophy Segmentation in SD-OCT Images

PURPOSE

To automatically and accurately segment geographic atrophy (GA) in spectral-domain optical coherence tomography (SD-OCT) images by constructing a voting system with deep neural networks without the use of retinal layer segmentation.

METHODS

An automatic GA segmentation method for SD-OCT images based on the deep network was constructed. The structure of the deep network was composed of five layers, including one input layer, three hidden layers, and one output layer. During the training phase, the labeled A-scans with 1024 features were directly fed into the network as the input layer to obtain the deep representations. Then a soft-max classifier was trained to determine the label of each individual pixel. Finally, a voting decision strategy was used to refine the segmentation results among 10 trained models.

RESULTS

Two image data sets with GA were used to evaluate the model. For the first dataset, our algorithm obtained a mean overlap ratio (OR) 86.94% ± 8.75%, absolute area difference (AAD) 11.49% ± 11.50%, and correlation coefficients (CC) 0.9857; for the second dataset, the mean OR, AAD, and CC of the proposed method were 81.66% ± 10.93%, 8.30% ± 9.09%, and 0.9952, respectively. The proposed algorithm was capable of improving over 5% and 10% segmentation accuracy, respectively, when compared with several state-of-the-art algorithms on two data sets.

CONCLUSIONS

Without retinal layer segmentation, the proposed algorithm could produce higher segmentation accuracy and was more stable when compared with state-of-the-art methods that relied on retinal layer segmentation results. Our model may provide reliable GA segmentations from SD-OCT images and be useful in the clinical diagnosis of advanced nonexudative AMD.

TRANSLATIONAL RELEVANCE

Based on the deep neural networks, this study presents an accurate GA segmentation method for SD-OCT images without using any retinal layer segmentation results, and may contribute to improved understanding of advanced nonexudative AMD.

Algorithms for the Automated Analysis of Age-Related Macular Degeneration Biomarkers on Optical Coherence Tomography: A Systematic Review

PURPOSE

To assess the quality of optical coherence tomography (OCT) grading algorithms for retinal biomarkers of age-related macular degeneration (AMD).

METHODS

Following a systematic review of the literature data on detection and quantification of AMD retinal biomarkers by available algorithms were extracted and descriptively synthesized. Algorithm quality was assessed using a modified version of the Quality Assessment of Diagnostic Accuracy Studies 2 checklist with a focus on accuracy against established reference standards and risk of bias.

RESULTS

Thirty five studies reporting computer-aided diagnosis (CAD) tools for qualitative analysis or algorithms for quantitative analysis were identified. Compared with manual assessment in reference standards correlation coefficients ranged from 0.54 to 0.97 for drusen, 0.80 to 0.98 for geographic atrophy (GA), and 0.30 to 0.98 for intra- or subretinal fluid and pigment epithelial detachment (PED) detection by automated algorithms. CAD tools achieved area under the curve (AUC) values of 0.94 to 0.99, sensitivity of 0.90 to 1.00, and specificity of 0.89 to 0.92.

CONCLUSIONS

Automated analysis of AMD biomarkers on OCT is promising. However, most of the algorithm validation was performed in preselected patients, exhibiting the targeted biomarker only. In addition, type and quality of reported algorithm validation varied substantially.

TRANSLATIONAL RELEVANCE

The development of algorithms for combined, simultaneous analysis of multiple AMD biomarkers including AMD staging and the agreement on standardized validation procedures would be of considerable translational value for the clinician and the clinical researcher.

En Face Spectral-Domain Optical Coherence Tomography for the Monitoring of Lesion Area Progression in Stargardt Disease

Purpose

We investigated the progression of Stargardt disease (STGD1) over a multiyear follow-up by evaluating the macular lesion area as computed by an automatic algorithm from spectral-domain optical coherence tomography (SD-OCT).

Methods

We reviewed medical records of STGD1 patients, with a clinical and molecular diagnosis of STGD1 at a single institution, who underwent best-corrected visual acuity (BCVA), fundus photography, SD-OCT, full-field electroretinography, and, when available, fundus autofluorescence (FAF). Regression models were fitted on the selected clinical parameters; in particular, on the macular lesion area computed by SD-OCT, to evaluate the disease progression over a multiyear follow-up.

Results

The comparison between SD-OCT and FAF, available for 22 patients, showed that macular lesion area, assessed by SD-OCT, significantly correlated with the area of absent FAF (P < 0.001). Moreover, the longitudinal analysis, performed in 98 patients, showed a significant enlargement of macular lesion area at an estimated exponential rate of 4.6% per year (P = 0.046), together with a significant worsening of BCVA (0.06 logMAR per year; P < 0.001) and a significant decrease of macular thickness (1.6% per year; P = 0.005) over the follow-up.

Conclusions

The current study describes, for the first time in literature, a longitudinal analysis of the macular lesion area assessed by SD-OCT in STGD1 disease, showing a significant progression over the follow-up. Our findings suggest that the evaluation of macular lesion area by en face SD-OCT, together with FAF, could drive the choice of the most amenable candidates and the most suitable area to be treated in gene therapy clinical trials.

Multimodality analysis of Hyper-reflective Foci and Hard Exudates in Patients with Diabetic Retinopathy

To investigate the correlations between hyper-reflective foci and hard exudates in patients with non-proliferative diabetic retinopathy (NPDR) and proliferative diabetic retinopathy (PDR) by spectral-domain optical coherence tomography (SD OCT) images. Hyper-reflective foci in retinal SD OCT images were automatically detected by the developed algorithm. Then, the cropped CFP images generated by the semi-automatic registration method were automatically segmented for the hard exudates and corrected by the experienced clinical ophthalmologist. Finally, a set of 5 quantitative imaging features were automatically extracted from SD OCT images, which were used for investigating the correlations of hyper-reflective foci and hard exudates and predicting the severity of diabetic retinopathy. Experimental results demonstrated the positive correlations in area and amount between hard exudates and hyper-reflective foci at different stages of diabetic retinopathy, with statistical significance (all p < 0.05). In addition, the area and amount can be taken as potential discriminant indicators of the severity of diabetic retinopathy.

PROGRESSION OF MACULAR ATROPHY IN PATIENTS WITH NEOVASCULAR AGE-RELATED MACULAR DEGENERATION UNDERGOING ANTIVASCULAR ENDOTHELIAL GROWTH FACTOR THERAPY

PURPOSE

To define the frequency and quantify the progression of macular atrophy (MA) in patients with neovascular age-related macular degeneration undergoing treatment with antivascular endothelial growth factor therapy for >2 years.

METHODS

Fifty-four eyes of 46 patients (86.7 ± 6.8 years, 53.7% women) diagnosed with wet age-related macular degeneration were included in this retrospective study. Eyes that received photodynamic therapy or laser treatment were excluded. All eyes were imaged at baseline and after 2 years with the Cirrus spectral domain optical coherence tomography using a 512 × 128 macular cube scan protocol centered on the fovea. Optical coherence tomography en face fundus images were obtained for each 3-dimensional data set using the U.S. Food and Drug Administration-cleared Advanced RPE Analysis software, which automatically identifies atrophic areas by segmenting regions of increased reflectivity in en face choroidal slab images. Segmentation errors were manually corrected by trained Doheny Image Reading Center graders using a standardized grading protocol. The prevalence rates of atrophy at baseline and at 2-years follow-up and enlargement rates were computed. Baseline demographic factors and types and numbers of antivascular endothelial growth factor injections received over time were correlated with the development and enlargement of atrophy.

RESULTS

Macular atrophy was noted at baseline in 32 (59.3%) eyes and progressed in all eyes over the next 2 years. Among the 28 eyes without atrophy at baseline, MA developed by 2 years in 6 eyes (21.4% of eyes without MA at baseline). Of note, 22 eyes (40.7% of overall cohort) never developed atrophy during the course of the study. Among eyes with atrophy at baseline, the annual growth rate of MA was found to be 0.89 ± 0.93 mm. A multiple regression analysis was performed to evaluate the influence of gender, age, smoking status, medication injected, and number of injections on MA. Except for the number of total injections (R = 0.3, P < 0.01), the studied variables could not significantly predict development or progression of MA (F [0.73, 13] = 0.378, P = 0.86, R = 0.05). However, the study was not powered to detect small effects.

CONCLUSION

Macular atrophy is a frequent finding in eyes with wet age-related macular degeneration both before and after antivascular endothelial growth factor therapy. The frequency of new optical coherence tomography-defined atrophy (21% at 2 years) after starting therapy was close to the rates reported in CATT, IVAN, and HARBOR. The rate of MA enlargement was positively correlated with the number of injections, but did not appear to be greater than that reported for atrophy in the absence of choroidal neovascularization.

En Face Optical Coherence Tomography Imaging for the Detection of Nascent Geographic Atrophy

PURPOSE

To determine if en face optical coherence tomography (OCT) imaging can identify nascent geographic atrophy (nGA) in eyes with intermediate age-related macular degeneration (iAMD).

DESIGN

Retrospective observational case series.

METHODS

Patients with iAMD from the COMPLETE study at the Bascom Palmer Eye Institute were evaluated to determine if nGA was present at baseline and at follow-up using high-density Spectralis OCT B-scans and en face OCT images from the Cirrus OCT instrument. If available, additional en face OCT images and B-scans were analyzed at follow-up times beyond the 52-week period.

RESULTS

A total of 37 eyes (27 patients) were evaluated for at least 1 year using both B-scans and en face images. Two drusen suspicious for nGA at baseline were identified, but neither druse developed GA after 24 and 62 months of follow-up, respectively. Another druse displayed hypertransmission into the choroid at week 52 on B-scan imaging and was classified as nGA. En face OCT imaging identified this druse as a focal bright area. Drusen breakdown occurred during a follow-up of 39 months.

CONCLUSIONS

En face OCT imaging appeared to be as useful as routine B-scan imaging for identifying areas suspicious for nGA in this population from the COMPLETE Study. Additional longitudinal follow-up of eyes with drusen is needed to determine if en face OCT imaging can replace the evaluation of individual B-scans for the detection of nGA.

A view of the current and future role of optical coherence tomography in the management of age-related macular degeneration

Optical coherence tomography (OCT) has become an established diagnostic technology in the clinical management of age-related macular degeneration (AMD). OCT is being used for primary diagnosis, evaluation of therapeutic efficacy, and long-term monitoring. Computer-based advances in image analysis provide complementary imaging tools such as OCT angiography, further novel automated analysis methods as well as feature detection and prediction of prognosis in disease and therapy by machine learning. In early AMD, pathognomonic features such as drusen, pseudodrusen, and abnormalities of the retinal pigment epithelium (RPE) can be imaged in a qualitative and quantitative way to identify early signs of disease activity and define the risk of progression. In advanced AMD, disease activity can be monitored clearly by qualitative and quantified analyses of fluid pooling, such as intraretinal cystoid fluid, subretinal fluid, and pigment epithelial detachment (PED). Moreover, machine learning methods detect a large spectrum of new biomarkers. Evaluation of treatment efficacy and definition of optimal therapeutic regimens are an important aim in managing neovascular AMD. In atrophic AMD hallmarked by geographic atrophy (GA), advanced spectral domain (SD)-OCT imaging largely replaces conventional fundus autofluorescence (FAF) as it adds insight into the condition of the neurosensory layers and associated alterations at the level of the RPE and choroid. Exploration of imaging features by computerized methods has just begun but has already opened relevant and reliable horizons for the optimal use of OCT imaging for individualized and population-based management of AMD-the leading retinal epidemic of modern times.

Automatic Identification of Pathology-Distorted Retinal Layer Boundaries Using SD-OCT Imaging

OBJECTIVE

We propose an effective automatic method for identification of four retinal layer boundaries from the spectral domain optical coherence tomography images in the presence and absence of pathologies and morphological changes due to disease.

METHODS

The approach first finds an approximate location of three reference layers and then uses these to bound the search space for the actual layers, which is achieved by modeling the problem as a graph and applying Dijkstra's shortest path algorithm. The edge weight between nodes is determined using pixel distance, slope similarity to a reference, and nonassociativity of the layers, which is designed to overcome the distorting effects that pathology can play in the boundary determination.

RESULTS

The accuracy of our method was evaluated on three different datasets. It outperforms the current five state-of-the-art methods. On average, the mean and standard deviation of the root-mean-square error in the form of mean ± standard deviation in pixels for our method is 1.57 ± 0.69, which is lower than compared to the existing top five methods of 16.17 ± 22.64, 6.66 ± 9.11, 5.70 ± 10.54, 3.69 ± 2.04, and 2.29 ± 1.54.

CONCLUSION

Our method is highly accurate, robust, reliable, and consistent. This identification can enable to quantify the biomarkers of the retina in large-scale study for assessing, monitoring disease progression, as well as early detection of retinal diseases.

SIGNIFICANCE

Identification of these boundaries can help to determine the loss of neuroretinal cells or layers and the presence of retinal pathology, which can be used as features for the automatic determination of the stages of retinal diseases.

Automated geographic atrophy segmentation for SD-OCT images using region-based C-V model via local similarity factor

Age-related macular degeneration (AMD) is the leading cause of blindness among elderly individuals. Geographic atrophy (GA) is a phenotypic manifestation of the advanced stages of non-exudative AMD. Determination of GA extent in SD-OCT scans allows the quantification of GA-related features, such as radius or area, which could be of important value to monitor AMD progression and possibly identify regions of future GA involvement. The purpose of this work is to develop an automated algorithm to segment GA regions in SD-OCT images. An en face GA fundus image is generated by averaging the axial intensity within an automatically detected sub-volume of the three dimensional SD-OCT data, where an initial coarse GA region is estimated by an iterative threshold segmentation method and an intensity profile set, and subsequently refined by a region-based Chan-Vese model with a local similarity factor. Two image data sets, consisting on 55 SD-OCT scans from twelve eyes in eight patients with GA and 56 SD-OCT scans from 56 eyes in 56 patients with GA, respectively, were utilized to quantitatively evaluate the automated segmentation algorithm. We compared results obtained by the proposed algorithm, manual segmentation by graders, a previously proposed method, and experimental commercial software. When compared to a manually determined gold standard, our algorithm presented a mean overlap ratio (OR) of 81.86% and 70% for the first and second data sets, respectively, while the previously proposed method OR was 72.60% and 65.88% for the first and second data sets, respectively, and the experimental commercial software OR was 62.40% for the second data set.

Drusen Volume and Retinal Pigment Epithelium Abnormal Thinning Volume Predict 2-Year Progression of Age-Related Macular Degeneration

PURPOSE

To analyze the value of novel measures of retinal pigment epithelium-drusen complex (RPEDC) volume to predict 2-year disease progression of intermediate age-related macular degeneration (AMD).

DESIGN

Prospective, observational study.

PARTICIPANTS

Three hundred forty-five AMD and 122 non-AMD participants enrolled in the Age Related Eye Disease Study 2 Ancillary Spectral-Domain (SD) Optical Coherence Tomography (OCT) study.

METHODS

High-density SD OCT macular volumes were obtained at yearly study visits. The RPEDC abnormal thickening (henceforth, OCT drusen) and RPEDC abnormal thinning (RAT) volumes were generated by semiautomated segmentation of total RPEDC within a 5-mm-diameter macular field.

MAIN OUTCOME MEASURES

Volume change and odds ratio (OR) with 95% confidence intervals (CI) for progression to advanced AMD with choroidal neovascularization (CNV) or central geographic atrophy (GA).

RESULTS

Complete volumes were obtained in 265 and 266 AMD eyes and in 115 and 97 control eyes at baseline and at year 2, respectively. In AMD eyes, mean (standard deviation) OCT drusen volume increased from 0.08 mm(3) (0.16 mm(3)) to 0.10 mm(3) (0.23 mm(3); P < 0.001), and RAT volume increased from 8.3 × 10(-4) mm(3) (20.8 × 10(-4) mm(3)) to 18.4 × 10(-4) mm(3) (46.6 × 10(-4) mm(3); P < 0.001). Greater baseline OCT drusen volume was associated with 2-year progression to CNV (P = 0.002). Odds of developing CNV increased by 31% for every 0.1-mm(3) increase in baseline OCT drusen volume (OR, 1.31; 95% CI, 1.06-1.63; P = 0.013). Greater baseline RAT volume was associated with significant 2-year increase in RAT volume (P < 0.001), noncentral GA (P < 0.001), and progression to central GA (P < 0.001). Odds of developing central GA increased by 32% for every 0.001-mm(3) increase in baseline RAT volume (OR, 1.32; 95% CI, 1.14-1.53; P < 0.001). In non-AMD eyes, all volumes were significantly lower than AMD eyes and showed no significant 2-year change.

CONCLUSIONS

Macular OCT drusen and RAT volumes increased significantly in AMD eyes over 2 years. These quantitative SD OCT biomarkers predict 2-year AMD progression and may serve as useful biomarkers for future clinical trials.

Restricted Summed-Area Projection for Geographic Atrophy Visualization in SD-OCT Images

PURPOSE

To enhance the rapid assessment of geographic atrophy (GA) across the macula in a single projection image generated from three-dimensional (3D) spectral-domain optical coherence tomography (SD-OCT) scans by introducing a novel restricted summed-area projection (RSAP) technique.

METHODS

We describe a novel en face GA visualization technique, the RSAP, by restricting the axial projection of SD-OCT images to the regions beneath the Bruch's membrane (BM) boundary and also considering the choroidal vasculature's influence on GA visualization. The technique analyzes the intensity distribution beneath the retinal pigment epithelium (RPE) layer to fit a cross-sectional surface in the sub-RPE region. The area is taken as the primary GA projection. A median filter is then adopted to smooth the generated GA projection image. The RSAP technique was evaluated in 99 3D SD-OCT data sets from 27 eyes of 21 patients presenting with advanced nonexudative age-related macular degeneration and GA. We used the mean difference between GA and background regions and GA separability metric to measure GA contrast and distinction in the generated images, respectively. We compared our results with two existing GA projection techniques, the summed-voxel projection (SVP) and Sub-RPE Slab techniques.

RESULTS

Comparative results demonstrate that the RSAP technique is more effective in displaying GA than the SVP and Sub-RPE Slab. The average of the mean difference between GA and background regions and the GA separability based on SVP, Sub-RPE Slab, and RSAP were 0.129/0.880, 0.238/0.919, and 0.276/0.938, respectively.

CONCLUSIONS

The RSAP technique was more effective for GA visualization than the conventional SVP and Sub-RPE Slab techniques. Our technique decreases choroidal vasculature influence on GA projection images by analyzing the intensity distribution characteristics in sub-RPE regions. The generated GA projection image with the RSAP technique has improved contrast and distinction.

TRANSLATIONAL RELEVANCE

Our method for automated generation of GA projection images from SD-OCT images may improve the visualization of the macular abnormalities and the management of GA.

Polarisation-sensitive OCT is useful for evaluating retinal pigment epithelial lesions in patients with neovascular AMD

Background/aims

To examine the reproducibility of lesion dimensions of the retinal pigment epithelium (RPE) in neovascular age-related macular degeneration (AMD) with polarisation-sensitive optical coherence tomography (PS-OCT), specifically imaging the RPE.

Methods

Twenty-six patients (28 eyes) with neovascular AMD were included in this study, and examined by a PS-OCT prototype. Each patient was scanned five times at a 1-day visit. The PS-OCT B-scan located closest to the macular centre presenting with RPE atrophy was identified, and the longitudinal diameter of the lesion was quantified manually using AutoCAD 2008. This procedure was followed for the identical B-scan position in all five scans per eye and patient. Reproducibility of qualitative changes in PS-OCT was evaluated. Interobserver variability was assessed. Results were compared with intensity-based spectral-domain OCT (SD-OCT) imaging.

Results

Mean variability of all atrophy lesion dimensions was 0.10 mm (SD±=0.06 mm). Coefficient of variation (SD±/mean) was 0.06 on average (SD±=0.03). Interobserver variability assessment showed a mean difference of 0.02 mm across all patients regarding RPE lesion size evaluation (paired t test: p=0.38). Spearman correlation coefficient was r=0.98, p<0.001. Results revealed a good overall reproducibility of ∼90%. PS-OCT specifically detected the RPE in all eyes compared with conventional intensity-based SD-OCT that was not capable to clearly identify RPE atrophy in 25 eyes (89.3%, p<0.01).

Conclusions

PS-OCT offers good reproducibility of RPE atrophy assessment in neovascular AMD, and may be suitable for precise RPE evaluation in clinical practice. PS-OCT unambiguously identifies RPE changes in choroidal neovascularisation compared with intensity-based SD-OCT that does not identify the RPE status reliably.

Automated segmentation of geographic atrophy of the retinal epithelium via random forests in AREDS color fundus images

BACKGROUND

Age-related macular degeneration (AMD), left untreated, is the leading cause of vision loss in people older than 55. Severe central vision loss occurs in the advanced stage of the disease, characterized by either the in growth of choroidal neovascularization (CNV), termed the "wet" form, or by geographic atrophy (GA) of the retinal pigment epithelium (RPE) involving the center of the macula, termed the "dry" form. Tracking the change in GA area over time is important since it allows for the characterization of the effectiveness of GA treatments. Tracking GA evolution can be achieved by physicians performing manual delineation of GA area on retinal fundus images. However, manual GA delineation is time-consuming and subject to inter-and intra-observer variability.

METHODS

We have developed a fully automated GA segmentation algorithm in color fundus images that uses a supervised machine learning approach employing a random forest classifier. This algorithm is developed and tested using a dataset of images from the NIH-sponsored Age Related Eye Disease Study (AREDS). GA segmentation output was compared against a manual delineation by a retina specialist.

RESULTS

Using 143 color fundus images from 55 different patient eyes, our algorithm achieved PPV of 0.82±0.19, and NPV of 0:95±0.07.

DISCUSSION

This is the first study, to our knowledge, applying machine learning methods to GA segmentation on color fundus images and using AREDS imagery for testing. These preliminary results show promising evidence that machine learning methods may have utility in automated characterization of GA from color fundus images.

Feasibility of level-set analysis of enface OCT retinal images in diabetic retinopathy

Pathology segmentation in retinal images of patients with diabetic retinopathy is important to help better understand disease processes. We propose an automated level-set method with Fourier descriptor-based shape priors. A cost function measures the difference between the current and expected output. We applied our method to enface images generated for seven retinal layers and determined correspondence of pathologies between retinal layers. We compared our method to a distance-regularized level set method and show the advantages of using well-defined shape priors. Results obtained allow us to observe pathologies across multiple layers and to obtain metrics that measure the co-localization of pathologies in different layers.

Automated segmentation of geographic atrophy in fundus autofluorescence images using supervised pixel classification

Geographic atrophy (GA) is a manifestation of the advanced or late stage of age-related macular degeneration (AMD). AMD is the leading cause of blindness in people over the age of 65 in the western world. The purpose of this study is to develop a fully automated supervised pixel classification approach for segmenting GA, including uni- and multifocal patches in fundus autofluorescene (FAF) images. The image features include region-wise intensity measures, gray-level co-occurrence matrix measures, and Gaussian filter banks. A [Formula: see text]-nearest-neighbor pixel classifier is applied to obtain a GA probability map, representing the likelihood that the image pixel belongs to GA. Sixteen randomly chosen FAF images were obtained from 16 subjects with GA. The algorithm-defined GA regions are compared with manual delineation performed by a certified image reading center grader. Eight-fold cross-validation is applied to evaluate the algorithm performance. The mean overlap ratio (OR), area correlation (Pearson's [Formula: see text]), accuracy (ACC), true positive rate (TPR), specificity (SPC), positive predictive value (PPV), and false discovery rate (FDR) between the algorithm- and manually defined GA regions are [Formula: see text], [Formula: see text], [Formula: see text], [Formula: see text], [Formula: see text], [Formula: see text], and [Formula: see text], respectively.