Deep Learning Approaches for Detecting of Nascent Geographic Atrophy in Age-Related Macular Degeneration.

Purpose: Nascent geographic atrophy (nGA) refers to specific features seen on OCT B-scans, which are strongly associated with the future development of geographic atrophy (GA). This study sought to develop a deep learning model to screen OCT B-scans for nGA that warrant further manual review (an artificial intelligence [AI]-assisted approach), and to determine the extent of reduction in OCT B-scan load requiring manual review while maintaining near-perfect nGA detection performance. Design: Development and evaluation of a deep learning model. Participants: One thousand eight hundred and eighty four OCT volume scans (49 B-scans per volume) without neovascular age-related macular degeneration from 280 eyes of 140 participants with bilateral large drusen at baseline, seen at 6-monthly intervals up to a 36-month period (from which 40 eyes developed nGA). Methods: OCT volume and B-scans were labeled for the presence of nGA. Their presence at the volume scan level provided the ground truth for training a deep learning model to identify OCT B-scans that potentially showed nGA requiring manual review. Using a threshold that provided a sensitivity of 0.99, the B-scans identified were assigned the ground truth label with the AI-assisted approach. The performance of this approach for detecting nGA across all visits, or at the visit of nGA onset, was evaluated using fivefold cross-validation. Main Outcome Measures: Sensitivity for detecting nGA, and proportion of OCT B-scans requiring manual review. Results: The AI-assisted approach (utilizing outputs from the deep learning model to guide manual review) had a sensitivity of 0.97 (95% confidence interval [CI] = 0.93-1.00) and 0.95 (95% CI = 0.87-1.00) for detecting nGA across all visits and at the visit of nGA onset, respectively, when requiring manual review of only 2.7% and 1.9% of selected OCT B-scans, respectively. Conclusions: A deep learning model could be used to enable near-perfect detection of nGA onset while reducing the number of OCT B-scans requiring manual review by over 50-fold. This AI-assisted approach shows promise for substantially reducing the current burden of manual review of OCT B-scans to detect this crucial feature that portends future development of GA. Financial Disclosures: Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.

Interreader Agreement and Longitudinal Progression of Incomplete/Complete Retinal Pigment Epithelium and Outer Retinal Atrophy in Age-Related Macular Degeneration.

OBJECTIVE: To analyze the ability to evaluate changes over time of individual lesions of incomplete or complete retinal pigment epithelium (RPE) and outer retinal atrophy (iRORA and cRORA, respectively) in patients with intermediate age-related macular degeneration (iAMD). DESIGN: OCT images from patients enrolled in Proxima B clinical trial (NCT02399072) were utilized. PARTICIPANTS: Patients enrolled in the Proxima B clinical trial, from the cohort with geographic atrophy (GA) in 1 eye and iAMD in the other eye at baseline, were included. METHODS: Junior and senior readers analyzed OCT images for the qualitative presence of 9 distinct early atrophic features (presence of zone of choroidal hypertransmission, attenuation and/or disruption of RPE, disruption of ellipsoid zone [EZ] and external limiting membrane [ELM], outer nuclear layer [ONL] thinning, outer plexiform layer [OPL]/inner nuclear layer [INL] subsidence, and hyporeflective wedge-shaped band). If deemed "present," 7 features were quantified with a predefined tolerance level of 50 µm (diameter for the zone of choroidal hypertransmission, zone of attenuation and/or disruption of the RPE, outer retinal thickness left/right vertical diameter, outer retinal thickness thinnest vertical diameter, annotation of EZ, and ELM disruption). MAIN OUTCOME MEASURES: Interreader agreements for qualitative assessments (κ-type statistics) and quantitative measurements (Bland-Altman statistics) were assessed. Progression of the lesion features over time was described. RESULTS: Moderate agreement was found for presence of choroidal hypertransmission (κ = 0.54), followed by ELM disruption (κ = 0.58), OPL/INL subsidence (κ = 0.46), and a hyporeflective wedge-shaped band (κ = 0.47). Quantification measurements showed that choroidal hypertransmission had the highest agreement, whereas RPE attenuation/disruption had the lowest agreement. Longitudinal adjudicated changes for quantitative measurements of lesion progression showed that choroidal hypertransmission and ELM disruption showed significant progression, whereas EZ disruption and RPE attenuation/disruption did not. CONCLUSIONS: The ability to evaluate changes over time for specific features of iRORA and cRORA was explored. The most robust biomarker was found to be choroidal hypertransmission, followed by ELM disruption and the qualitative markers of OPL/INL subsidence, as well as a wedge-shaped band. Disease progression over time could be assessed by some, but not all, spectral-domain OCT features that were explored. FINANCIAL DISCLOSURE(S): Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.

Geographic Atrophy Segmentation Using Multimodal Deep Learning.

Purpose: To examine deep learning (DL)-based methods for accurate segmentation of geographic atrophy (GA) lesions using fundus autofluorescence (FAF) and near-infrared (NIR) images. Methods: This retrospective analysis utilized imaging data from study eyes of patients enrolled in Proxima A and B (NCT02479386; NCT02399072) natural history studies of GA. Two multimodal DL networks (UNet and YNet) were used to automatically segment GA lesions on FAF; segmentation accuracy was compared with annotations by experienced graders. The training data set comprised 940 image pairs (FAF and NIR) from 183 patients in Proxima B; the test data set comprised 497 image pairs from 154 patients in Proxima A. Dice coefficient scores, Bland-Altman plots, and Pearson correlation coefficient (r) were used to assess performance. Results: On the test set, Dice scores for the DL network to grader comparison ranged from 0.89 to 0.92 for screening visit; Dice score between graders was 0.94. GA lesion area correlations (r) for YNet versus grader, UNet versus grader, and between graders were 0.981, 0.959, and 0.995, respectively. Longitudinal GA lesion area enlargement correlations (r) for screening to 12 months (n = 53) were lower (0.741, 0.622, and 0.890, respectively) compared with the cross-sectional results at screening. Longitudinal correlations (r) from screening to 6 months (n = 77) were even lower (0.294, 0.248, and 0.686, respectively). Conclusions: Multimodal DL networks to segment GA lesions can produce accurate results comparable with expert graders. Translational Relevance: DL-based tools may support efficient and individualized assessment of patients with GA in clinical research and practice.

Rare CIDEC coding variants enriched in age-related macular degeneration patients with small low-luminance deficit cause lipid droplet and fat storage defects.

BACKGROUND: The basis of Age-related macular degeneration (AMD) genetic risk has been well documented; however, few studies have looked at genetic biomarkers of disease progression or treatment response within advanced AMD patients. Here we report the first genome-wide analysis of genetic determinants of low-luminance vision deficit (LLD), which is seen as predictive of visual acuity loss and anti-VEGF treatment response in neovascular AMD patients. METHODS: AMD patients were separated into small- and large-LLD groups for comparison and whole genome sequencing was performed. Genetic determinants of LLD were assessed by common and rare variant genetic analysis. Follow-up functional analysis of rare coding variants identified by the burden test was then performed in vitro. RESULTS: We identified four coding variants in the CIDEC gene. These rare variants were only present in patients with a small LLD, which has been previously shown to indicate better prognosis and better anti-VEGF treatment response. Our in vitro functional characterization of these CIDEC alleles revealed that all decrease the binding affinity between CIDEC and the lipid droplet fusion effectors PLIN1, RAB8A and AS160. The rare CIDEC alleles all cause a hypomorphic defect in lipid droplet fusion and enlargement, resulting in a decreased fat storage capability in adipocytes. CONCLUSIONS: As we did not detect CIDEC expression in the ocular tissue affected by AMD, our results suggest that the CIDEC variants do not play a direct role in the eye and influence low-luminance vision deficit via an indirect and systemic effect related to fat storage capacity.

Deep Learning to Predict Geographic Atrophy Area and Growth Rate from Multimodal Imaging.

OBJECTIVE: To develop deep learning models for annualized geographic atrophy (GA) growth rate prediction using fundus autofluorescence (FAF) images and spectral-domain OCT volumes from baseline visits, which can be used for prognostic covariate adjustment to increase power of clinical trials. DESIGN: This retrospective analysis estimated GA growth rate as the slope of a linear fit on all available measurements of lesion area over a 2-year period. Three multitask deep learning models-FAF-only, OCT-only, and multimodal (FAF and OCT)-were developed to predict concurrent GA area and annualized growth rate. PARTICIPANTS: Patients were from prospective and observational lampalizumab clinical trials. METHODS: The 3 models were trained on the development data set, tested on the holdout set, and further evaluated on the independent test sets. Baseline FAF images and OCT volumes from study eyes of patients with bilateral GA (NCT02247479; NCT02247531; and NCT02479386) were split into development (1279 patients/eyes) and holdout (443 patients/eyes) sets. Baseline FAF images from study eyes of NCT01229215 (106 patients/eyes) and NCT02399072 (169 patients/eyes) were used as independent test sets. MAIN OUTCOME MEASURES: Model performance was evaluated using squared Pearson correlation coefficient (r2) between observed and predicted lesion areas/growth rates. Confidence intervals were calculated by bootstrap resampling (B = 10 000). RESULTS: On the holdout data set, r2 (95% confidence interval) of the FAF-only, OCT-only, and multimodal models for GA lesion area prediction was 0.96 (0.95-0.97), 0.91 (0.87-0.95), and 0.94 (0.92-0.96), respectively, and for GA growth rate prediction was 0.48 (0.41-0.55), 0.36 (0.29-0.43), and 0.47 (0.40-0.54), respectively. On the 2 independent test sets, r2 of the FAF-only model for GA lesion area was 0.98 (0.97-0.99) and 0.95 (0.93-0.96), and for GA growth rate was 0.65 (0.52-0.75) and 0.47 (0.34-0.60). CONCLUSIONS: We show the feasibility of using baseline FAF images and OCT volumes to predict individual GA area and growth rates using a multitask deep learning approach. The deep learning-based growth rate predictions could be used for covariate adjustment to increase power of clinical trials. FINANCIAL DISCLOSURE(S): Proprietary or commercial disclosure may be found after the references.

Conversion from Intermediate Age-Related Macular Degeneration to Geographic Atrophy in a Proxima B Subcohort Using a Multimodal Approach.

INTRODUCTION: This retrospective analysis assessed geographic atrophy (GA) progression in fellow eyes from the Proxima B trial intermediate age-related macular degeneration (iAMD) subcohort using high-resolution multimodal imaging anchored on optical coherence tomography (OCT). METHODS: Thirty-two patients from the Proxima B iAMD subcohort were assessed; all had GA with no macular neovascularization (MNV) in the contralateral eye. Imaging data, including color fundus photography, fluorescein angiography, near-infrared reflectance, fundus autofluorescence (FAF), and spectral-domain OCT, were obtained. Features preceding progression/conversion to advanced AMD (drusen, reticular pseudodrusen [RPD], MNV, incomplete/complete retinal pigment epithelium and outer retinal atrophy [iRORA/cRORA]) were assessed. RESULTS: Of 30 fellow eyes with available follow-up images, 12 converted to GA (FAF), 2 converted to MNV, and 16 were nonconverters during the review period (median: 17.8 months). iRORA/cRORA features (present in all converters at baseline) were identified on OCT in eyes that progressed to GA. Median time interval from iRORA to cRORA and from cRORA to GA was 7 months each. GA development/progression was either drusen- or RPD-associated (n = 6 each). Eyes with baseline RPD showed faster GA progression versus eyes with drusen (1.49 vs. 0.38 mm2/year). CONCLUSIONS: RPD presence was associated with rapid GA lesion enlargement and may provide an early indication of faster GA progression.

Feasibility Study of a Multimodal, Cloud-Based, Diabetic Retinal Screening Program in a Workplace Environment.

Purpose: To evaluate the feasibility of capturing and interpreting retinal images in a workplace environment using a multimodal, cloud-based, diabetic retinal screening program combined with electronic self-reported questionnaires. The burden of diabetic retinopathy (DR) and other retinal conditions, healthcare utilization, and visual function were also assessed. Methods: A cross-sectional feasibility study was conducted at the Genentech, Inc., Campus Health Center. Eyes of participants were imaged using ultra-widefield (UWF) color fundus photography (CFP) and spectral-domain optical coherence tomography (SD-OCT). A cloud-based platform was used for the automated, seamless transfer of images to a remote reading center for evaluation for DR and other retinal pathologies. Electronic surveys collected participants' self-reported medical histories, healthcare utilization, and visual function data. Results: Among 100 participants (mean age, 43.9 years; 44% male), 33% of them self-reported diabetes. Eye examinations within the past 12 months were reported by 71% of all participants (n = 71/100) and by 85% (n = 28/33) of those with self-reported diabetes. Among participants with complete screening images from both UWF-CFP and SD-OCT, 20% (n = 6/30) of those with self-reported diabetes and 8.5% (n = 5/59) of participants with no history of diabetes were unaware they had mild/moderate nonproliferative DR. Among all participants, 20% (20/100) had a retinal finding, on either UWF-CFP or SD-OCT, or both, which prompted a referral for further evaluation. Conclusions: A retinal screening program deployed via a secure, scalable, and interoperable cloud-based platform was feasible and conveniently integrated into the workplace. Translational Relevance: Cloud-based platforms could be used to promote a secure, scalable, and interoperable system for retinal screening in nontraditional environments.

Translational and pharmacokinetic-pharmacodynamic application for the clinical development of GDC-0334, a novel TRPA1 inhibitor.

GDC-0334 is a novel small molecule inhibitor of transient receptor potential cation channel member A1 (TRPA1), a promising therapeutic target for many nervous system and respiratory diseases. The pharmacokinetic (PK) profile and pharmacodynamic (PD) effects of GDC-0334 were evaluated in this first-in-human (FIH) study. A starting single dose of 25 mg was selected based on integrated preclinical PK, PD, and toxicology data following oral administration of GDC-0334 in guinea pigs, rats, dogs, and monkeys. Human PK and PK-PD of GDC-0334 were characterized after single and multiple oral dosing using a population modeling approach. The ability of GDC-0334 to inhibit dermal blood flow (DBF) induced by topical administration of allyl isothiocyanate (AITC) was evaluated as a target-engagement biomarker. Quantitative models were developed iteratively to refine the parameter estimates of the dose-concentration-effect relationships through stepwise estimation and extrapolation. Human PK analyses revealed that bioavailability, absorption rate constant, and lag time increase when GDC-0334 was administered with food. The inhibitory effect of GDC-0334 on the AITC-induced DBF biomarker exhibited a clear sigmoid-Emax relationship with GDC-0334 plasma concentrations in humans. This study leveraged emerging preclinical and clinical data to enable iterative refinement of GDC-0334 mathematical models throughout the FIH study for dose selection in subsequent cohorts throughout the study. Study Highlights WHAT IS THE CURRENT KNOWLEDGE ON THE TOPIC? GDC-0334 is a novel, small molecule TRPA1 inhibitor and a pharmacokinetic-pharmacodynamic (PK-PD) modeling strategy could be implemented in a systematic and step-wise manner to build and learn from emerging data for early clinical development. WHAT QUESTION DID THIS STUDY ADDRESS? Can noncompartmental and population-based analyses be used to describe the PK and PD characteristics of GDC-0334 in preclinical and clinical studies? WHAT DOES THIS STUDY ADD TO OUR KNOWLEDGE? GDC-0334 exposure generally increased with dose in rats, dogs, and monkeys. The starting dose (25 mg) in the clinical study was determined based on the preclinical data. GDC-0334 exhibited linear PK in humans and the bioavailability was increased with food. The inhibitory effect of GDC-0334 on dermal blood flow induced by the TRPA1 agonist allyl isothiocyanate in humans indicates a clear PK-PD relationship. HOW MIGHT THIS CHANGE CLINICAL PHARMACOLOGY OR TRANSLATIONAL SCIENCE? The models developed based on TRPA1 agonist-induced dermal blood flow inhibition data can be used to predict PK-PD relationships in future preclinical and clinical studies evaluating new drug entities that target TRPA1.

A TRPA1 inhibitor suppresses neurogenic inflammation and airway contraction for asthma treatment.

Despite the development of effective therapies, a substantial proportion of asthmatics continue to have uncontrolled symptoms, airflow limitation, and exacerbations. Transient receptor potential cation channel member A1 (TRPA1) agonists are elevated in human asthmatic airways, and in rodents, TRPA1 is involved in the induction of airway inflammation and hyperreactivity. Here, the discovery and early clinical development of GDC-0334, a highly potent, selective, and orally bioavailable TRPA1 antagonist, is described. GDC-0334 inhibited TRPA1 function on airway smooth muscle and sensory neurons, decreasing edema, dermal blood flow (DBF), cough, and allergic airway inflammation in several preclinical species. In a healthy volunteer Phase 1 study, treatment with GDC-0334 reduced TRPA1 agonist-induced DBF, pain, and itch, demonstrating GDC-0334 target engagement in humans. These data provide therapeutic rationale for evaluating TRPA1 inhibition as a clinical therapy for asthma.

Development of AITC-induced dermal blood flow as a translational in vivo biomarker of TRPA1 activity in human and rodent skin.

BACKGROUND AND PURPOSE: Develop a translational assay of Transient Receptor Potential Ankyrin 1 (TRPA1) activity for use as a preclinical and clinical biomarker. EXPERIMENTAL APPROACH: Allyl isothiocyanate (AITC), capsaicin or citric acid were applied to ears of wildtype and Trpa1-knock out (Trpa1 KO) rats, and changes in dermal blood flow (DBF) were measured by laser speckle contrast imaging. In humans, the DBF, pain and itch responses to 5-20% AITC applied to the forearm were measured and safety was evaluated. Reproducibility of the DBF, pain and itch responses to topically applied 10% and 15% AITC were assessed at two visits separated by 13-15 days. DBF changes were summarized at 5-minute intervals as areas under the curve (AUC) and maxima. Intraclass correlation coefficient (ICC) was calculated to assess arm-arm and period-period reproducibility. KEY RESULTS: AITC- and citric acid-induced DBF were significantly reduced in Trpa1 KO rats compared to wildtype (90 ± 2% and 65 ± 11% reduction, respectively), whereas capsaicin response did not differ. In humans, each AITC concentration significantly increased DBF compared to vehicle with the maximal increase occurring 5 minutes post application. Ten percent and 15% AITC were selected as safe and effective stimuli. AUC from 0 to 5 minutes was the most reproducible metric of AITC-induced DBF across arms (ICC = 0.92) and periods (ICC = 0.85). Subject-reported pain was more reproducible than itch across visits (ICC = 0.76 vs 0.17, respectively). CONCLUSION AND IMPLICATIONS: AITC-induced DBF is a suitable target engagement biomarker of TRPA1 activity for preclinical and clinical studies of TRPA1 antagonists.

Development of Deep Learning Models to Predict Best-Corrected Visual Acuity from Optical Coherence Tomography.

Purpose: To develop deep learning (DL) models to predict best-corrected visual acuity (BCVA) from optical coherence tomography (OCT) images from patients with neovascular age-related macular degeneration (nAMD). Methods: Retrospective analysis of OCT images and associated BCVA measurements from the phase 3 HARBOR trial (NCT00891735). DL regression models were developed to predict BCVA at the concurrent visit and 12 months from baseline using OCT images. Binary classification models were developed to predict BCVA of Snellen equivalent of <20/40, <20/60, and ≤20/200 at the concurrent visit and 12 months from baseline. Results: The regression model to predict BCVA at the concurrent visit had R2 = 0.67 (root-mean-square error [RMSE] = 8.60) in study eyes and R2 = 0.84 (RMSE = 9.01) in fellow eyes. The best classification model to predict BCVA at the concurrent visit had an area under the receiver operating characteristic curve (AUC) of 0.92 in study eyes and 0.98 in fellow eyes. The regression model to predict BCVA at month 12 using baseline OCT had R2 = 0.33 (RMSE = 14.16) in study eyes and R2 = 0.75 (RMSE = 11.27) in fellow eyes. The best classification model to predict BCVA at month 12 had AUC = 0.84 in study eyes and AUC = 0.96 in fellow eyes. Conclusions: DL shows promise in predicting BCVA from OCTs in nAMD. Further research should elucidate the utility of models in clinical settings. Translational Relevance: DL models predicting BCVA could be used to enhance understanding of structure-function relationships and develop more efficient clinical trials.

Signal Strength Reduction Effects in OCT Angiography.

PURPOSE: To elucidate the relationship between vessel density (VD) measurements and signal strength in OCT angiography (OCTA). DESIGN: Cross-sectional study. PARTICIPANTS: Healthy volunteers. METHODS: OCT angiography images obtained from healthy volunteers were analyzed to demonstrate the relationship between signal strength index (SSI) and VD. Experiments were performed to determine the effects of signal strength reduction on VD measurements on the Optovue/AngioVue (Optovue, Inc, Fremont, CA) and Cirrus/AngioPlex OCTA (Carl Zeiss Meditec, Inc, Dublin, CA) systems. Signal strength reduction was generated by either neutral density filters (NDFs) or defocus. MAIN OUTCOME MEASURES: Regression analysis of signal strength effects on VD. RESULTS: Vessel density decreased linearly with signal strength with high statistical significance on both OCTA systems tested and for all analyzed sources of variation in signal strength. The slope of VD versus SSI was greatest when signal strength was adjusted by NDFs, followed by defocus, interscan difference, interindividual variation, and left-right eye difference. Multivariate analysis revealed that both SSI and age had a significant effect on the interindividual variation in VD. CONCLUSIONS: Vessel density measurements using OCTA were affected significantly by OCT signal strengths on 2 OCTA platforms. Investigators should exercise caution when interpreting VD data from OCTA scans. Quantification algorithms for OCTA should ideally remove the signal strength bias.

Correlation of Outer Retinal Degeneration and Choriocapillaris Loss in Stargardt Disease Using En Face Optical Coherence Tomography and Optical Coherence Tomography Angiography.

PURPOSE: This study measured and correlated degeneration of the junction between the inner and outer segments (IS/OS), the retinal pigment epithelium (RPE), and the choriocapillaris (CC) in Stargardt disease (STGD). DESIGN: Prospective cross-sectional study. METHODS: This study was conducted at the Casey Eye Institute. A total of 23 patients with STGD were enrolled and underwent optical coherence tomography angiography (OCTA). Scans were centered on the fovea. OCT slab projections and en face boundary maps were used to create masks to measure total IS/OS loss or RPE atrophy as well as regions of isolated IS/OS loss, isolated RPE atrophy, and matched IS/OS and RPE degeneration or intact IS/OS junction and RPE. CC vascular density (CCVD) was quantified from the CC angiogram. Outcomes included the area of loss, and the CCVD of degeneration in different areas was quantified and correlated. RESULTS: The total area of IS/OS loss was strongly correlated with the total area of RPE atrophy (r = 0.96; P < 0.0001) by a 1.6:1 ratio (r2 = 0.90). CCVD within regions of matched degeneration (85.6% ± 2.7%; P < 0.0001), isolated IS/OS junction loss (93.6% ± 1.0%; P = 0.0011), and isolated RPE atrophy (94.1% ± 1.1%; P = 0.0065) were all significantly lower than normal (99.0% ± 0.17%). There was a trend for CCVD within intact areas (97.6% ± 0.38%) to decline as the area diminished (r = 0.68). CONCLUSIONS: Photoreceptor and RPE degeneration exhibited a strong relationship wherein the IS/OS loss was 1.6-fold greater than that of RPE atrophy, supporting the theory that photoreceptor degeneration precedes RPE in STGD. Both the photoreceptors and the RPE degeneration contributed synergistically to CCVD attenuation, but extralesional CCVD also tended to be abnormal. The findings and techniques in this study may be of utility in developing endpoints for clinical trials.

Plexus-Specific Detection of Retinal Vascular Pathologic Conditions with Projection-Resolved OCT Angiography.

OBJECTIVE: To evaluate the projection-resolved (PR) optical coherence tomography angiography (OCTA) algorithm in detecting plexus-specific vascular abnormalities in retinal pathologies. DESIGN: Cross-sectional observational clinical study. PARTICIPANTS: Patients diagnosed with retinal vascular diseases and healthy volunteers. METHODS: Eyes were imaged using an OCT system operating at 840 nm and employing the split-spectrum amplitude decorrelation algorithm. A novel algorithm suppressed projection artifacts inherent to OCTA. The volumetric scans were segmented and visualized on different plexuses. MAIN OUTCOME MEASURES: Qualitative observation of vascular abnormalities on both cross-sectional and en face PR-OCTA images. RESULTS: Eight illustrative cases are reported. In cases of diabetic retinopathy, retinal vessel occlusion, and retinitis pigmentosa, PR-OCTA detected retinal nonperfusion regions within deeper retinal plexuses not visualized by conventional OCTA. In age-related macular degeneration, cross-sectional PR-OCTA permitted the classification of choroidal neovascularization, and, in a case of retinal angiomatous proliferation, identified a vertical vessel contiguous with the deep capillary plexus. In macular telangiectasia, PR-OCTA detected a diving perifoveal vein and delineated subretinal neovascularization. CONCLUSIONS: Application of PR-OCTA promises to improve sensitive, accurate evaluation of individual vascular plexuses in multiple retinal diseases.

Sensitivity and Specificity of OCT Angiography to Detect Choroidal Neovascularization.

PURPOSE: To determine the sensitivity and specificity of optical coherence tomography angiography (OCTA) in the detection of choroidal neovascularization (CNV) in age-related macular degeneration (AMD). DESIGN: Prospective case series. SUBJECTS: Prospective series of seventy-two eyes were studied, which included eyes with treatment-naive CNV due to AMD, non-neovascular AMD, and normal controls. METHODS: All eyes underwent OCTA with a spectral domain (SD) OCT (Optovue, Inc.). The 3D angiogram was segmented into separate en face views including the inner retinal angiogram, outer retinal angiogram, and choriocapillaris angiogram. Detection of abnormal flow in the outer retina served as candidate CNV with OCTA. Masked graders reviewed structural OCT alone, en face OCTA alone, and en face OCTA combined with cross-sectional OCTA for the presence of CNV. MAIN OUTCOME MEASURE: The sensitivity and specificity of CNV detection compared to the gold standard of fluorescein angiography (FA) and OCT was determined for structural SD-OCT alone, en face OCTA alone, and with en face OCTA combined with cross-sectional OCTA. RESULTS: Of 32 eyes with CNV, both graders identified 26 true positives with en face OCTA alone, resulting in a sensitivity of 81.3%. Four of the 6 false negatives had large subretinal hemorrhage (SRH) and sensitivity improved to 94% for both graders if eyes with SRH were excluded. The addition of cross-sectional OCTA along with en face OCTA improved the sensitivity to 100% for both graders. Structural OCT alone also had a sensitivity of 100%. The specificity of en face OCTA alone was 92.5% for grader A and 97.5% for grader B. The specificity of structural OCT alone was 97.5% for grader A and 85% for grader B. Cross-sectional OCTA combined with en face OCTA had a specificity of 97.5% for grader A and 100% for grader B. CONCLUSIONS: Sensitivity and specificity for CNV detection with en face OCTA combined with cross-sectional OCTA approaches that of the gold standard of FA with OCT, and it is better than en face OCTA alone. Structural OCT alone has excellent sensitivity for CNV detection. False positives from structural OCT can be mitigated with the addition of flow information with OCTA.

Optical coherence tomography angiography: Technical principles and clinical applications in ophthalmology.

Optical coherence tomography angiography (OCTA) is a functional extension of OCT that provides information on retinal and choroidal circulations without the need for dye injections. With the recent development of high-speed OCT systems and efficient algorithms, OCTA has become clinically feasible. In this review article, we discuss the technical principles of OCTA, including image processing and artifacts, and its clinical applications in ophthalmology. We summarize recent studies which qualitatively or quantitatively assess disease presentation, progression, and/or response to treatment.

Projection-Resolved Optical Coherence Tomography Angiography of Macular Retinal Circulation in Glaucoma.

PURPOSE: To detect macular perfusion defects in glaucoma using projection-resolved optical coherence tomography (OCT) angiography. DESIGN: Prospective observation study. PARTICIPANTS: A total of 30 perimetric glaucoma and 30 age-matched normal participants were included. METHODS: One eye of each participant was imaged using 6×6-mm macular OCT angiography (OCTA) scan pattern by 70-kHz 840-nm spectral-domain OCT. Flow signal was calculated by the split-spectrum amplitude-decorrelation angiography algorithm. A projection-resolved OCTA (PR-OCTA) algorithm was used to remove flow projection artifacts. Four en face OCTA slabs were analyzed: the superficial vascular complex (SVC), intermediate capillary plexus (ICP), deep capillary plexus (DCP), and all-plexus retina (SVC + ICP + DCP). The vessel density (VD), defined as the percentage area occupied by flow pixels, was calculated from en face OCTA. A novel algorithm was used to adjust the vessel density to compensate for local variations in OCT signal strength. MAIN OUTCOME MEASURES: Macular retinal VD, ganglion cell complex (GCC) thickness, and visual field (VF) sensitivity. RESULTS: Focal capillary dropout could be visualized in the SVC, but not the ICP and DVP, in glaucomatous eyes. In the glaucoma group, the SVC and all-plexus retinal VD (mean ± standard deviation: 47.2%±7.1% and 73.5%±6.6%) were lower than in the normal group (60.5%±4.0% and 83.2%±4.2%, both P < 0.001, t test). The ICP and DCP VD were not significantly lower in the glaucoma group. Among the overall macular VD parameters, the SVC VD had the best diagnostic accuracy as measured by the area under the receiver operating characteristic curve (AROC). The accuracy was even better when the worse hemisphere (inferior or superior) was used, achieving an AROC of 0.983 and a sensitivity of 96.7% at a specificity of 95%. Among the glaucoma participants, the hemispheric SVC VD values were highly correlated with the corresponding GCC thickness and VF sensitivity (P < 0.003). The reflectance compensation step in VD calculation significantly improved repeatability, normal population variation, and correlation with VF and GCC thickness. CONCLUSIONS: On the basis of PR-OCTA, glaucoma preferentially affects perfusion in the SVC in the macula more than the deeper plexuses. Reflectance-compensated SVC VD measurement by PR-OCTA detected glaucoma with high accuracy and could be useful in the clinical evaluation of glaucoma.

Automated boundary detection of the optic disc and layer segmentation of the peripapillary retina in volumetric structural and angiographic optical coherence tomography.

To improve optic disc boundary detection and peripapillary retinal layer segmentation, we propose an automated approach for structural and angiographic optical coherence tomography. The algorithm was performed on radial cross-sectional B-scans. The disc boundary was detected by searching for the position of Bruch's membrane opening, and retinal layer boundaries were detected using a dynamic programming-based graph search algorithm on each B-scan without the disc region. A comparison of the disc boundary using our method with that determined by manual delineation showed good accuracy, with an average Dice similarity coefficient ≥0.90 in healthy eyes and eyes with diabetic retinopathy and glaucoma. The layer segmentation accuracy in the same cases was on average less than one pixel (3.13 µm).

Automated detection of dilated capillaries on optical coherence tomography angiography.

Automated detection and grading of angiographic high-risk features in diabetic retinopathy can potentially enhance screening and clinical care. We have previously identified capillary dilation in angiograms of the deep plexus in optical coherence tomography angiography as a feature associated with severe diabetic retinopathy. In this study, we present an automated algorithm that uses hybrid contrast to distinguish angiograms with dilated capillaries from healthy controls and then applies saliency measurement to map the extent of the dilated capillary networks. The proposed algorithm agreed well with human grading.

Choriocapillaris evaluation in choroideremia using optical coherence tomography angiography.

The choriocapillaris plays an important role in supporting the metabolic demands of the retina. Studies of the choriocapillaris in disease states with optical coherence tomography angiography (OCTA) have proven insightful. However, image artifacts complicate the identification and quantification of the choriocapillaris in degenerative diseases such as choroideremia. Here, we demonstrate a supervised machine learning approach to detect intact choriocapillaris based on training with results from an expert grader. We trained a random forest classifier to evaluate en face structural OCT and OCTA information along with spatial image features. Evaluation of the trained classifier using previously unseen data showed good agreement with manual grading.