Visualizing human photoreceptor and retinal pigment epithelium cell mosaics in a single volume scan over an extended field of view with adaptive optics optical coherence tomography

Image Quality in Adaptive Optics Optical Coherence Tomography of Diabetic Patients

Background/Objectives: An assessment of the retinal image quality in adaptive optics optical coherence tomography (AO-OCT) is challenging. Many factors influence AO-OCT imaging performance, leading to greatly varying imaging results, even in the same subject. The aim of this study is to introduce quantitative means for an assessment of AO-OCT image quality and to compare these with parameters retrieved from the pyramid wavefront sensor of the system. Methods: We used a spectral domain AO-OCT instrument to repetitively image six patients suffering from diabetic retinopathy over a time span of one year. The data evaluation consists of two volume acquisitions with a focus on the photoreceptor layer, each at five different retinal locations per visit; 7-8 visits per patient are included in this data analysis, resulting in a total of ~420 volumes. Results: A large variability in AO-OCT image quality is observed between subjects and between visits of the same subject. On average, the image quality does not depend on the measurement location. The data show a moderate correlation between the axial position of the volume recording and image quality. The correlation between pupil size and AO-OCT image quality is not linear. A weak correlation is found between the signal-to-noise ratio of the wavefront sensor image and the image quality. Conclusions: The introduced AO-OCT image quality metric gives useful insights into the performance of such a system. A longitudinal assessment of this metric, together with wavefront sensor data, is essential to identify factors influencing image quality and, in the next step, to optimize the performance of AO-OCT systems.

On the improvement of schizophrenia detection with optical coherence tomography data using deep neural networks and aggregation functions

Schizophrenia is a serious mental disorder with a complex neurobiological background and a well-defined psychopathological picture. Despite many efforts, a definitive disease biomarker has still not been identified. One of the promising candidates for a disease-related biomarker could involve retinal morphology , given that the retina is a part of the central nervous system that is known to be affected in schizophrenia and related to multiple illness features. In this study Optical Coherence Tomography (OCT) data is applied to assess the different layers of the retina. OCT data were applied in the process of automatic differentiation of schizophrenic patients from healthy controls. Numerical experiments involved applying several individual 1D Convolutional Neural Network-based models as well as further using the aggregation of classification results to improve the initial classification results. The main goal of the study was to check how methods based on the aggregation of classification results work in classifying neuroanatomical features of schizophrenia. Among over 300, 000 different variants of tested aggregation operators, a few versions provided satisfactory results.

Identifying retinal pigment epithelium cells in adaptive optics-optical coherence tomography images with partial annotations and superhuman accuracy

Retinal pigment epithelium (RPE) cells are essential for normal retinal function. Morphological defects in these cells are associated with a number of retinal neurodegenerative diseases. Owing to the cellular resolution and depth-sectioning capabilities, individual RPE cells can be visualized in vivo with adaptive optics-optical coherence tomography (AO-OCT). Rapid, cost-efficient, and objective quantification of the RPE mosaic's structural properties necessitates the development of an automated cell segmentation algorithm. This paper presents a deep learning-based method with partial annotation training for detecting RPE cells in AO-OCT images with accuracy better than human performance. We have made the code, imaging datasets, and the manual expert labels available online.

Improvements on speed, stability and field of view in adaptive optics OCT for anterior retinal imaging using a pyramid wavefront sensor

We present improvements on the adaptive optics (AO) correction method using a pyramid wavefront sensor (P-WFS) and introduce a novel approach for closed-loop focus shifting in retinal imaging. The method's efficacy is validated through in vivo adaptive optics optical coherence tomography (AO-OCT) imaging in both, healthy individuals and patients with diabetic retinopathy. In both study groups, a stable focusing on the anterior retinal layers is achieved. We further report on an improvement in AO loop speed that can be used to expand the imaging area of AO-OCT in the slow scanning direction, largely independent of the eye's isoplanatic patch. Our representative AO-OCT data reveal microstructural details of the neurosensory retina such as vessel walls and microglia cells that are visualized in single volume data and over an extended field of view. The excellent performance of the P-WFS based AO-OCT imaging in patients suggests good clinical applicability of this technology.

Comparison between Two Adaptive Optics Methods for Imaging of Individual Retinal Pigmented Epithelial Cells

The Retinal Pigment Epithelium (RPE) plays a prominent role in diseases such as age-related macular degeneration, but imaging individual RPE cells is challenging due to their high absorption and low autofluorescence emission. The RPE lies beneath the highly reflective photoreceptor layer (PR) and contains absorptive pigments, preventing direct backscattered light detection when the PR layer is intact. Here, we used near-infrared autofluorescence adaptive optics scanning laser ophthalmoscopy (NIRAF AOSLO) and transscleral flood imaging (TFI) in the same healthy eyes to cross-validate these approaches. Both methods revealed a consistent RPE mosaic pattern and appeared to reflect a distribution of fluorophores consistent with findings from histological studies. Interestingly, even in apparently healthy RPE, we observed dynamic changes over months, suggesting ongoing cellular activity or alterations in fluorophore distribution. These findings emphasize the value of NIRAF AOSLO and TFI in understanding RPE morphology and dynamics.

Age-Related Macular Degeneration, a Mathematically Tractable Disease

A progression sequence for age-related macular degeneration onset may be determinable with consensus neuroanatomical nomenclature augmented by drusen biology and eye-tracked clinical imaging. This narrative review proposes to supplement the Early Treatment of Diabetic Retinopathy Study (sETDRS) grid with a ring to capture high rod densities. Published photoreceptor and retinal pigment epithelium (RPE) densities in flat mounted aged-normal donor eyes were recomputed for sETDRS rings including near-periphery rich in rods and cumulatively for circular fovea-centered regions. Literature was reviewed for tissue-level studies of aging outer retina, population-level epidemiology studies regionally assessing risk, vision studies regionally assessing rod-mediated dark adaptation (RMDA), and impact of atrophy on photopic visual acuity. The 3 mm-diameter xanthophyll-rich macula lutea is rod-dominant and loses rods in aging whereas cone and RPE numbers are relatively stable. Across layers, the largest aging effects are accumulation of lipids prominent in drusen, loss of choriocapillary coverage of Bruch's membrane, and loss of rods. Epidemiology shows maximal risk for drusen-related progression in the central subfield with only one third of this risk level in the inner ring. RMDA studies report greatest slowing at the perimeter of this high-risk area. Vision declines precipitously when the cone-rich central subfield is invaded by geographic atrophy. Lifelong sustenance of foveal cone vision within the macula lutea leads to vulnerability in late adulthood that especially impacts rods at its perimeter. Adherence to an sETDRS grid and outer retinal cell populations within it will help dissect mechanisms, prioritize research, and assist in selecting patients for emerging treatments.

Volumetric Reconstruction of a Human Retinal Pigment Epithelial Cell Reveals Specialized Membranes and Polarized Distribution of Organelles

Purpose

Despite the centrality of the retinal pigment epithelium (RPE) in vision and retinopathy our picture of RPE morphology is incomplete. With a volumetric reconstruction of human RPE ultrastructure, we aim to characterize major membranous features including apical processes and their interactions with photoreceptor outer segments, basolateral infoldings, and the distribution of intracellular organelles.

Methods

A parafoveal retinal sample was acquired from a 21-year-old male organ donor. With serial block-face scanning electron microscopy, a tissue volume from the inner-outer segment junction to basal RPE was captured. Surface membranes and complete internal ultrastructure of an individual RPE cell were achieved with a combination of manual and automated segmentation methods.

Results

In one RPE cell, apical processes constitute 69% of the total cell surface area, through a dense network of over 3000 terminal branches. Single processes contact several photoreceptors. Basolateral infoldings facing the choriocapillaris resemble elongated filopodia and comprise 22% of the cell surface area. Membranous tubules and sacs of endoplasmic reticulum represent 20% of the cell body volume. A dense basal layer of mitochondria extends apically to partly overlap electron-dense pigment granules. Pores in the nuclear envelope form a distinct pattern of rows aligned with chromatin.

Conclusions

Specialized membranes at the apical and basal side of the RPE cell body involved in intercellular uptake and transport represent over 90% of the total surface area. Together with the polarized distribution of organelles within the cell body, these findings are relevant for retinal clinical imaging, therapeutic approaches, and disease pathomechanisms.

Physical compensation method for dispersion of multiple materials in swept source optical coherence tomography

An ophthalmic swept source-optical coherence tomography (SS-OCT) system based on a high-speed scanning laser at 1060 nm with a scanning rate of 100 KHz is constructed. Since the sample arm of the interferometer is comprised of multiple glass materials, the ensuing dispersion severely degrades imaging quality. In this article, second-order dispersion simulation analysis for various materials was performed first, and dispersion equilibrium was implemented utilizing physical compensation methods. After dispersion compensation, an imaging depth in air of 4.013 mm was achieved in model eye experiments, and signal-to-noise ratio was enhanced by 11.6%, with a value of 53.8 dB. In vivo imaging of the human retina was performed to demonstrate structurally distinguishable retinal images, characterized by an axial resolution improvement of 19.8%, with a value of 7.7 μm close to the theoretical value of 7.5 μm. The proposed physical dispersion compensation method enhances imaging performance in SS-OCT systems, enabling visualization of several low scattering mediums.

Twenty-five years of clinical applications using adaptive optics ophthalmoscopy [Invited]

Twenty-five years ago, adaptive optics (AO) was combined with fundus photography, thereby initiating a new era in the field of ophthalmic imaging. Since that time, clinical applications of AO ophthalmoscopy to investigate visual system structure and function in both health and disease abound. To date, AO ophthalmoscopy has enabled visualization of most cell types in the retina, offered insight into retinal and systemic disease pathogenesis, and been integrated into clinical trials. This article reviews clinical applications of AO ophthalmoscopy and addresses remaining challenges for AO ophthalmoscopy to become fully integrated into standard ophthalmic care.

Analysis of mitochondrial dynamics and function in the retinal pigment epithelium by high-speed high-resolution live imaging

Mitochondrial dysfunction is strongly implicated in neurodegenerative diseases including age-related macular degeneration (AMD), which causes irreversible blindness in over 50 million older adults worldwide. A key site of insult in AMD is the retinal pigment epithelium (RPE), a monolayer of postmitotic polarized cells that performs essential functions for photoreceptor health and vision. Recent studies from our group and others have identified several features of mitochondrial dysfunction in AMD including mitochondrial fragmentation and bioenergetic defects. While these studies provide valuable insight at fixed points in time, high-resolution, high-speed live imaging is essential for following mitochondrial injury in real time and identifying disease mechanisms. Here, we demonstrate the advantages of live imaging to investigate RPE mitochondrial dynamics in cell-based and mouse models. We show that mitochondria in the RPE form extensive networks that are destroyed by fixation and discuss important live imaging considerations that can interfere with accurate evaluation of mitochondrial integrity such as RPE differentiation status and acquisition parameters. Our data demonstrate that RPE mitochondria show localized heterogeneities in membrane potential and ATP production that could reflect focal changes in metabolism and oxidative stress. Contacts between the mitochondria and organelles such as the ER and lysosomes mediate calcium flux and mitochondrial fission. Live imaging of mouse RPE flatmounts revealed a striking loss of mitochondrial integrity in albino mouse RPE compared to pigmented mice that could have significant functional consequences for cellular metabolism. Our studies lay a framework to guide experimental design and selection of model systems for evaluating mitochondrial health and function in the RPE.

Advances in Optical Coherence Tomography Imaging Technology and Techniques for Choroidal and Retinal Disorders

Optical coherence tomography (OCT) imaging has played a pivotal role in the field of retina. This light-based, non-invasive imaging modality provides high-quality, cross-sectional analysis of the retina and has revolutionized the diagnosis and management of retinal and choroidal diseases. Since its introduction in the early 1990s, OCT technology has continued to advance to provide quicker acquisition times and higher resolution. In this manuscript, we discuss some of the most recent advances in OCT technology and techniques for choroidal and retinal diseases. The emerging innovations discussed include wide-field OCT, adaptive optics OCT, polarization sensitive OCT, full-field OCT, hand-held OCT, intraoperative OCT, at-home OCT, and more. The applications of these rising OCT systems and techniques will allow for a closer monitoring of chorioretinal diseases and treatment response, more robust analysis in basic science research, and further insights into surgical management. In addition, these innovations to optimize visualization of the choroid and retina offer a promising future for advancing our understanding of the pathophysiology of chorioretinal diseases.

Multi-modal and multi-scale clinical retinal imaging system with pupil and retinal tracking

We present a compact multi-modal and multi-scale retinal imaging instrument with an angiographic functional extension for clinical use. The system integrates scanning laser ophthalmoscopy (SLO), optical coherence tomography (OCT) and OCT angiography (OCTA) imaging modalities and provides multi-scale fields of view. For high resolution, and high lateral resolution in particular, cellular imaging correction of aberrations by adaptive optics (AO) is employed. The entire instrument has a compact design and the scanning head is mounted on motorized translation stages that enable 3D self-alignment with respect to the subject's eye by tracking the pupil position. Retinal tracking, based on the information provided by SLO, is incorporated in the instrument to compensate for retinal motion during OCT imaging. The imaging capabilities of the multi-modal and multi-scale instrument were tested by imaging healthy volunteers and patients.

Histology and clinical imaging lifecycle of black pigment in fibrosis secondary to neovascular age-related macular degeneration

PURPOSE

Melanotic cells with large spherical melanosomes, thought to originate from retinal pigment epithelium (RPE), are found in eyes with neovascular age-related macular degeneration (nvAMD). To generate hypotheses about RPE participation in fibrosis, we correlate histology to clinical imaging in an eye with prominent black pigment in fibrotic scar secondary to nvAMD.

METHODS

Macular findings in a white woman with untreated inactive subretinal fibrosis due to nvAMD in her right eye were documented over 9 years with color fundus photography (CFP), fundus autofluorescence (FAF) imaging, and optical coherence tomography (OCT). After death (age 90 years), this index eye was prepared for light and electron microscopy to analyze 7 discrete zones of pigmentation in the fibrotic scar. In additional donor eyes with nvAMD, we determined the frequency of black pigment (n = 36 eyes) and immuno-labeled for retinoid, immunologic, and microglial markers (RPE65, CD68, Iba1, TMEM119; n = 3 eyes).

RESULTS

During follow-up of the index eye, black pigment appeared and expanded within a hypoautofluorescent fibrotic scar. The blackest areas correlated to melanotic cells (containing large spherical melanosomes), some in multiple layers. Pale areas had sparse pigmented cells. Gray areas correlated to cells with RPE organelles entombed in the scar and multinucleate cells containing sparse large spherical melanosomes. In 94% of nvAMD donor eyes, hyperpigmentation was visible. Certain melanotic cells expressed some RPE65 and mostly CD68. Iba1 and TMEM119 immunoreactivity, found both in retina and scar, did not co-localize with melanotic cells.

CONCLUSION

Hyperpigmentation in CFP results from both organelle content and optical superimposition effects. Black fundus pigment in nvAMD is common and corresponds to cells containing numerous large spherical melanosomes and superimposition of cells containing sparse large melanosomes, respectively. Melanotic cells are molecularly distinct from RPE, consistent with a process of transdifferentiation. The subcellular source of spherical melanosomes remains to be determined. Detailed histology of nvAMD eyes will inform future studies using technologies for spatially resolved molecular discovery to generate new therapies for fibrosis. The potential of black pigment as a biomarker for fibrosis can be investigated in clinical multimodal imaging datasets.

Retinal adaptive optics imaging with a pyramid wavefront sensor

The pyramid wavefront sensor (P-WFS) has replaced the Shack-Hartmann (SH-) WFS as the sensor of choice for high-performance adaptive optics (AO) systems in astronomy. Many advantages of the P-WFS, such as its adjustable pupil sampling and superior sensitivity, are potentially of great benefit for AO-supported imaging in ophthalmology as well. However, so far no high quality ophthalmic AO imaging was achieved using this novel sensor. Usually, a P-WFS requires modulation and high precision optics that lead to high complexity and costs of the sensor. These factors limit the competitiveness of the P-WFS with respect to other WFS devices for AO correction in visual science. Here, we present a cost-effective realization of AO correction with a non-modulated P-WFS based on standard components and apply this technique to human retinal in vivo imaging using optical coherence tomography (OCT). P-WFS based high quality AO imaging was successfully performed in 5 healthy subjects and smallest retinal cells such as central foveal cone photoreceptors are visualized. The robustness and versatility of the sensor is demonstrated in the model eye under various conditions and in vivo by high-resolution imaging of other structures in the retina using standard and extended fields of view. As a quality benchmark, the performance of conventional SH-WFS based AO was used and successfully met. This work may trigger a paradigm shift with respect to the wavefront sensor of choice for AO in ophthalmic imaging.

Human Foveal Cone and Müller Cells Examined by Adaptive Optics Optical Coherence Tomography

Purpose

The purpose of this study was to image and investigate the foveal microstructure of human cone and Müller cells using adaptive optics-optical coherence tomography.

Methods

Six healthy subjects underwent the prototype adaptive optics-optical coherence tomography imaging, which allowed an axial resolution of 3.4 µm and a transverse resolution of approximately 3 µm. The morphological features of the individual retinal cells observed in the foveola were qualitatively and quantitatively evaluated.

Results

In the six healthy subjects, the image B-scans showed hyper-reflective dots that were densely packed in the outer nuclear layer. The mean number, diameter, and density of hyper-reflective dots in the foveola were 250.8 ± 59.6, 12.7 ± 59.6 µm, and 6966 ± 1833/mm2, respectively. These qualitative and quantitative findings regarding the hyper-reflective dots were markedly consistent with the morphological features of the foveal cone cell nuclei. Additionally, the images showed the funnel-shaped hyporeflective bodies running vertically and obliquely between the inner and external limiting membranes, illustrating the cell morphology of the foveal Müller cells.

Conclusions

Using adaptive optics, we succeeded in visualizing cross-sectional images of the individual cone and Müller cells of the human retina in vivo.

Translational Relevance

Adaptive optics-optical coherence tomography would help to improve our understanding of the pathogenesis of macular diseases.

Reflective mirror-based line-scan adaptive optics OCT for imaging retinal structure and function

Line-scan OCT incorporated with adaptive optics (AO) offers high resolution, speed, and sensitivity for imaging retinal structure and function in vivo. Here, we introduce its implementation with reflective mirror-based afocal telescopes, optimized for imaging light-induced retinal activity (optoretinography) and weak retinal reflections at the cellular scale. A non-planar optical design was followed based on previous recommendations with key differences specific to a line-scan geometry. The three beam paths fundamental to an OCT system -illumination/sample, detection, and reference- were modeled in Zemax optical design software to yield theoretically diffraction-limited performance over a 2.2 deg. field-of-view and 1.5 D vergence range at the eye's pupil. The performance for imaging retinal structure was exemplified by cellular-scale visualization of retinal ganglion cells, macrophages, foveal cones, and rods in human observers. The performance for functional imaging was exemplified by resolving the light-evoked optical changes in foveal cone photoreceptors where the spatial resolution was sufficient for cone spectral classification at an eccentricity 0.3 deg. from the foveal center. This enabled the first in vivo demonstration of reduced S-cone (short-wavelength cone) density in the human foveola, thus far observed only in ex vivo histological preparations. Together, the feasibility for high resolution imaging of retinal structure and function demonstrated here holds significant potential for basic science and translational applications.

Hyperreflective Foci, Optical Coherence Tomography Progression Indicators in Age-Related Macular Degeneration, Include Transdifferentiated Retinal Pigment Epithelium

Purpose

By optical coherence tomography (OCT) imaging, hyperreflective foci (HRF) indicate progression risk for advanced age-related macular degeneration (AMD) and are in part attributable to ectopic retinal pigment epithelium (RPE). We hypothesized that ectopic RPE are molecularly distinct from in-layer cells and that their cross-retinal course follows Müller glia.

Methods

In clinical OCT (61 eyes, 44 patients with AMD, 79.4 ± 7.7 years; 29 female; follow-up = 4.7 ± 0.9 years), one HRF type, RPE plume (n = 129 in 4 morphologies), was reviewed. Twenty eyes of 20 donors characterized by ex vivo OCT were analyzed by histology (normal, 4; early/intermediate AMD, 7; geographic atrophy, 6; neovascular AMD, 3). Cryosections were stained with antibodies to retinoid (RPE65, CRALPB) and immune (CD68, CD163) markers. In published RPE cellular phenotypes, red immunoreactivity was assessed semiquantitatively by one observer (none, some cells, all cells).

Results

Plume morphology evolved over time and many resolved (40%). Trajectories of RPE plume and cellular debris paralleled Müller glia, including near atrophy borders. RPE corresponding to HRF lost immunoreactivity for retinoid markers and gained immunoreactivity for immune markers. Aberrant immunoreactivity appeared in individual in-layer RPE cells and extended to all abnormal phenotypes. Müller glia remained CRALBP positive. Plume cells approached and contacted retinal capillaries.

Conclusions

HRF are indicators not predictors of overall disease activity. Gain and loss of function starts with individual in-layer RPE cells and extends to all abnormal phenotypes. Evidence for RPE transdifferentiation, possibly due to ischemia, supports a proposed process of epithelial–mesenchyme transition. Data can propel new biomarkers and therapeutic strategies for AMD.

Measuring the Contributions of Basal Laminar Deposit and Bruch's Membrane in Age-Related Macular Degeneration

Purpose

Basal laminar deposit (BLamD) is a consistent finding in age-related macular degeneration (AMD). We quantified BLamD thickness, appearance, and topography in eyes of aged donors with and without AMD and evaluated its relationship to other components of the retinal pigment epithelium-basal lamina/Bruch's membrane (RPE-BL-BrM) complex.

Methods

Donor eyes (n = 132) were classified as normal (n = 54), early to intermediate AMD (n = 24), geographic atrophy (GA; n = 13), and neovascular AMD (NV; n = 41). In high-resolution histology, we assessed RPE, BLamD, and BrM thicknesses and phenotypes at 3309 predefined locations in the central (foveal and perifovea) and superior (perifoveal) sections. Pre-mortem optical coherence tomography (OCT) imaging of a 90-year-old woman was compared to postmortem histopathology.

Results

In non-atrophic areas of AMD eyes, the RPE-BLamD is thick (normal = 13.7 µm, early-intermediate = 16.8 µm, GA = 17.4 µm, NV = 18.7 µm), because the BLamD is thick (normal = 0.3 µm, early-intermediate = 5.5 µm, GA = 4.1 µm, NV = 5.3 µm). RPE layer thickness is similar across these stages. Disease-associated variants of BLamD (thick, late, basal mounds) cluster subfoveally. A thick BLamD is visible on OCT as a hyporeflective split in the RPE-BL-BrM complex. BrM is thin (3.5 µm) in NV (normal = 4.2 µm, early to intermediate = 4.4 µm, and GA = 4.2 µm).

Conclusions

The RPE-BL-BrM complex is thick in AMD, driven by the accumulation and expansion of BLamD rather than expansion of either three-layer BrM, RPE-BL, or RPE. BLamD is clinically appreciable by OCT in some patients as a non-neovascular "split RPE-BL-BrM complex" or "double-layer sign." BLamD may contribute toward the formation and progression of high-risk drusen yet also exhibit protective properties.

Biometrics, Impact, and Significance of Basal Linear Deposit and Subretinal Drusenoid Deposit in Age-Related Macular Degeneration

Purpose

Basal linear deposit (BLinD) is a thin layer of soft drusen material. To elucidate the biology of extracellular deposits conferring age-related macular degeneration (AMD) progression risk and inform multimodal clinical imaging based on optical coherence tomography (OCT), we examined lipid content and regional prevalence of BLinD, soft drusen, pre-BLinD, and subretinal drusenoid deposit (SDD) in AMD and non-AMD aged eyes. We estimated BLinD volume and illustrated its relation to type 1 macular neovascularization (MNV).

Methods

Donor eyes were classified as early to intermediate AMD (n = 25) and age-matched controls (n = 54). In high-resolution histology, we assessed BLinD/soft drusen thickness at 836 and 1716 locations in AMD and control eyes, respectively. BLinD volume was estimated using solid geometry in donor eyes, one clinically characterized.

Results

BLinD, drusen, type 1 MNV, and fluid occupy the sub-RPE-basal laminar space. BLinD volume in a 3-mm diameter circle may be as much as 0.0315 mm3. Osmophilic lipid was more concentrated in BLinD/drusen than SDD. In the fovea, BLinD/drusen was prevalent in AMD eyes; pre-BLinD was prevalent in control eyes. SDD was low in the fovea and high in perifovea, especially in AMD eyes.

Conclusions

Although invisible, BLinD may presage type 1 MNV. BLinD volume approaches the criterion OCT drusen volume of 0.03 mm3 for AMD progression risk. BLinD culminates years of subfoveal lipid accumulation. SDD is detected relatively late in life, with currently unknown precursors. Deposit topography suggests one outer retinal lipid recycling system serving specialized cone and rod physiology, and its dysregulation in AMD is due to impaired transfer to the circulation.

Exploring a Structural Basis for Delayed Rod-Mediated Dark Adaptation in Age-Related Macular Degeneration Via Deep Learning

Purpose

Delayed rod-mediated dark adaptation (RMDA) is a functional biomarker for incipient age-related macular degeneration (AMD). We used anatomically restricted spectral domain optical coherence tomography (SD-OCT) imaging data to localize de novo imaging features associated with and to test hypotheses about delayed RMDA.

Methods

Rod intercept time (RIT) was measured in participants with and without AMD at 5 degrees from the fovea, and macular SD-OCT images were obtained. A deep learning model was trained with anatomically restricted information using a single representative B-scan through the fovea of each eye. Mean-occlusion masking was utilized to isolate the relevant imaging features.

Results

The model identified hyporeflective outer retinal bands on macular SD-OCT associated with delayed RMDA. The validation mean standard error (MSE) registered to the foveal B-scan localized the lowest error to 0.5 mm temporal to the fovea center, within an overall low-error region across the rod-free zone and adjoining parafovea. Mean absolute error (MAE) on the test set was 4.71 minutes (8.8% of the dynamic range).

Conclusions

We report a novel framework for imaging biomarker discovery using deep learning and demonstrate its ability to identify and localize a previously undescribed biomarker in retinal imaging. The hyporeflective outer retinal bands in central macula on SD-OCT demonstrate a structural basis for dysfunctional rod vision that correlates to published histopathologic findings.

Translational Relevance

This agnostic approach to anatomic biomarker discovery strengthens the rationale for RMDA as an outcome measure in early AMD clinical trials, and also expands the utility of deep learning beyond automated diagnosis to fundamental discovery.