Advertisement

in vivo Retinal Pigment Epithelium Imaging using Transscleral OPtical Imaging in healthy eyes

  • Laura Kowalczuk
    Correspondence
    Corresponding author:
    Affiliations
    Laboratory of Applied Photonic Devices (LAPD), School of Engineering, École Polytechnique Fédérale de Lausanne (EPFL), Lausanne, Switzerland

    Faculty of Biology and Medicine, University of Lausanne, Lausanne, Switzerland

    Jules-Gonin Eye Hospital, Fondation Asile des aveugles, Lausanne, Switzerland
    Search for articles by this author
  • Rémy Dornier
    Affiliations
    Laboratory of Applied Photonic Devices (LAPD), School of Engineering, École Polytechnique Fédérale de Lausanne (EPFL), Lausanne, Switzerland
    Search for articles by this author
  • Mathieu Kunzi
    Affiliations
    Laboratory of Applied Photonic Devices (LAPD), School of Engineering, École Polytechnique Fédérale de Lausanne (EPFL), Lausanne, Switzerland
    Search for articles by this author
  • Antonio Iskandar
    Affiliations
    Faculty of Biology and Medicine, University of Lausanne, Lausanne, Switzerland

    Jules-Gonin Eye Hospital, Fondation Asile des aveugles, Lausanne, Switzerland
    Search for articles by this author
  • Zuzana Misutkova
    Affiliations
    Faculty of Biology and Medicine, University of Lausanne, Lausanne, Switzerland

    Jules-Gonin Eye Hospital, Fondation Asile des aveugles, Lausanne, Switzerland
    Search for articles by this author
  • Aurélia Gryczka
    Affiliations
    Faculty of Biology and Medicine, University of Lausanne, Lausanne, Switzerland

    Jules-Gonin Eye Hospital, Fondation Asile des aveugles, Lausanne, Switzerland
    Search for articles by this author
  • Aurélie Navarro
    Affiliations
    Faculty of Biology and Medicine, University of Lausanne, Lausanne, Switzerland

    Jules-Gonin Eye Hospital, Fondation Asile des aveugles, Lausanne, Switzerland
    Search for articles by this author
  • Fanny Jeunet
    Affiliations
    Faculty of Biology and Medicine, University of Lausanne, Lausanne, Switzerland

    Jules-Gonin Eye Hospital, Fondation Asile des aveugles, Lausanne, Switzerland
    Search for articles by this author
  • Irmela Mantel
    Affiliations
    Faculty of Biology and Medicine, University of Lausanne, Lausanne, Switzerland

    Jules-Gonin Eye Hospital, Fondation Asile des aveugles, Lausanne, Switzerland
    Search for articles by this author
  • Francine Behar-Cohen
    Affiliations
    Centre de Recherche des Cordeliers, Université Pierre et Marie Curie - Paris 6, UMRS 1138, Paris, France

    INSERM U1138, USPC, Université de Paris-Cité, Sorbonne Université, From physiopathology of ocular diseases to clinical developments, Paris, France

    Assistance Publique - Hôpitaux de Paris, Ophtalmopôle, Cochin Hospital, Paris, France

    Université Paris Cité, Paris, France

    Hôpital Foch, Suresnes, France
    Search for articles by this author
  • Timothé Laforest
    Affiliations
    Laboratory of Applied Photonic Devices (LAPD), School of Engineering, École Polytechnique Fédérale de Lausanne (EPFL), Lausanne, Switzerland
    Search for articles by this author
  • Christophe Moser
    Affiliations
    Laboratory of Applied Photonic Devices (LAPD), School of Engineering, École Polytechnique Fédérale de Lausanne (EPFL), Lausanne, Switzerland
    Search for articles by this author
Open AccessPublished:October 18, 2022DOI:https://doi.org/10.1016/j.xops.2022.100234

      Abstract

      Objective

      To image healthy retinal pigment epithelial cells (RPE) in vivo using Transscleral OPtical Imaging (TOPI) and to analyze statistics of macular RPE cell features as a function of age, axial length (AL) and eccentricity.

      Design

      Single-center, exploratory, prospective, and descriptive clinical study.

      Participants

      49 eyes (AL: 24.03±0.93 mm; range: 21.88 - 26.7 mm) from 29 participants aged 21 to 70 years (37.1±13.3 years; 19 males, 10 females)

      Methods

      Retinal images, including ultra-wide field fundus photography and spectral-domain optical coherence tomography, AL and refractive error measurements were collected at baseline. For each eye, 6 high resolution RPE images were acquired using TOPI at different locations, one of them being imaged 5 times to evaluate the repeatability of the method. Follow-up ophthalmic examination was repeated 1 to 3 weeks after TOPI to assess safety. RPE images were analyzed with a custom automated software to extract cell parameters. Statistical analysis on the selected high-contrast images included calculation of coefficient of variation (CoV) for each feature at each repetition, Spearman and Mann-Whitney tests to investigate the relationship between cell features and eye and/or subject characteristics.

      Main Outcome Measures.

      RPE cell features such as density, area, center-to-center spacing, number of neighbors, circularity, elongation, solidity and border distance CoV.

      Results

      Macular RPE cell features were extracted from TOPI images at an eccentricity of 1.6° to 16.3° from the fovea. For each feature, the mean CoV was under 4%. Spearman’s test showed correlation within RPE cell features. In the perifovea, the region in which images were selected for all participants, longer AL significantly correlated with decreased RPE cell density (R Spearman, Rs=-0.746; p<0.0001) and increased cell area (Rs=0.668; p<0.0001), without morphological change. Aging also significantly correlated with decreased RPE density (Rs=-0.391; p=0.036) and increased cell area (Rs=0.454; p=0.013). Lower circular, less symmetric, more elongated and larger cells were observed over 50 years.

      Conclusions

      The TOPI technology imaged RPE cells in vivo with repeatability of less than 4% for the CoV and was used to analyze the influence of physiological factors on RPE cell morphometry in the perifovea of healthy volunteers.

      Key Words

      Abbreviations:

      AF (Autofluorescence), AL (Axial Length), AMD (Age-related Macular Degeneration), AO (Adaptive Optics), BCVA (Best-Corrected Visual Acuity), CCS (Center to Center Spacing), CoV (Coefficient of Variation), FFC (Flat-Field Correction), FOV (Field Of View), ICG (Indocyanin Green), IOP (Intraocular pressure), NIR (Near-Infrared), OCT (Optical Coherence Tomography), PRL (Preferred Retinal Locus), QC (Quality Criterion), RE (Refractive Error), RPE (Retinal Pigment Epithelium), Rs (Spearman Coefficient), SD (Standard Deviation), SLO (Scanning Laser Ophthalmoscope), TOPI (Transscleral OPtical Imaging)

      Introduction

      The retinal pigment epithelium (RPE), strategically located between the photoreceptors and the choroidal circulation, forms the outer blood retinal barrier that contributes to the health and function of photoreceptor cells.
      • Lakkaraju A.
      • Umapathy A.
      • Tan L.X.
      • Daniele L.
      • Philp N.J.
      • Boesze-Battaglia K.
      • Williams D.S.
      The cell biology of the retinal pigment epithelium.
      RPE cells are primarily affected in age-related macular degeneration (AMD),
      • Khan K.N.
      • Mahroo O.A.
      • Khan R.S.
      • Mohamed M.D.
      • McKibbin M.
      • Bird A.
      • Michaelides M.
      • Tufail A.
      • Moore A.T.
      Differentiating drusen: Drusen and drusen-like appearances associated with ageing, age-related macular degeneration, inherited eye disease and other pathological processes.
      impaired in the early stages of diabetic retinopathy,
      • Simó R.
      • Villarroel M.
      • Corraliza L.
      • Hernández C.
      • Garcia-Ramírez M.
      The retinal pigment epithelium: something more than a constituent of the blood-retinal barrier--implications for the pathogenesis of diabetic retinopathy.
      ,
      • Rothschild P.R.
      • Salah S.
      • Berdugo M.
      • Gélizé E.
      • Delaunay K.
      • Naud M.C.
      • Klein C.
      • Moulin A.
      • Savoldelli M.
      • Bergin C.
      • Jeanny J.C.
      • Jonet L.
      • Arsenijevic Y.
      • Behar-Cohen F.
      • Crisanti P.
      ROCK-1 mediates diabetes-induced retinal pigment epithelial and endothelial cell blebbing: Contribution to diabetic retinopathy.
      and at the center of pathological processes associated with pachychoroid spectrum diseases.
      • Cheung C.M.G.
      • Lee W.K.
      • Koizumi H.
      • Dansingani K.
      • Lai T.Y.Y.
      • Freund K.B.
      Pachychoroid disease.
      Histopathology studies in human eyes have identified RPE cell features and morphological changes that occur with aging,
      • Feeney-Burns L.
      • Hilderbrand E.S.
      • Eldridge S.
      Aging human RPE: morphometric analysis of macular, equatorial, and peripheral cells.
      • Dorey C.K.
      • Wu G.
      • Ebenstein D.
      • Garsd A.
      • Weiter J.J.
      Cell loss in the aging retina. Relationship to lipofuscin accumulation and macular degeneration.
      • Feeney-Burns L.
      • Burns R.P.
      • Gao C.L.
      Age-related macular changes in humans over 90 years old.
      • Gao H.
      • Hollyfield J.G.
      Aging of the human retina. Differential loss of neurons and retinal pigment epithelial cells.
      • Watzke R.C.
      • Soldevilla J.D.
      • Trune D.R.
      Morphometric analysis of human retinal pigment epithelium: correlation with age and location.
      • Panda-Jonas S.
      • Jonas J.B.
      • Jakobczyk-Zmija M.
      Retinal pigment epithelial cell count, distribution, and correlations in normal human eyes.
      • Harman A.M.
      • Fleming P.A.
      • Hoskins R.V.
      • Moore S.R.
      Development and aging of cell topography in the human retinal pigment epithelium.
      • Ach T.
      • Huisingh C.
      • McGwin Jr., G.
      • Messinger J.D.
      • Zhang T.
      • Bentley M.J.
      • Gutierrez D.B.
      • Ablonczy Z.
      • Smith R.T.
      • Sloan K.R.
      • Curcio C.A.
      Quantitative autofluorescence and cell density maps of the human retinal pigment epithelium.
      • Bhatia S.K.
      • Rashid A.
      • Chrenek M.A.
      • Zhang Q.
      • Bruce B.B.
      • Klein M.
      • Boatright J.H.
      • Jiang Y.
      • Grossniklaus H.E.
      • Nickerson J.M.
      Analysis of RPE morphometry in human eyes.
      and others that are specific to the onset and progression of retinal diseases, such as AMD
      • Ach T.
      • Tolstik E.
      • Messinger J.D.
      • Zarubina A.V.
      • Heintzmann R.
      • Curcio C.A.
      Lipofuscin redistribution and loss accompanied by cytoskeletal stress in retinal pigment epithelium of eyes with age-related macular degeneration.
      • Tarau I.S.
      • Berlin A.
      • Curcio C.A.
      • Ach T.
      The Cytoskeleton of the Retinal Pigment Epithelium: from Normal Aging to Age-Related Macular Degeneration.
      and retinitis pigmentosa.
      • Milam A.H.
      • Li Z.Y.
      • Fariss R.N.
      Histopathology of the human retina in retinitis pigmentosa.
      However, to date, no imaging tool is available in the clinic to image RPE cells in a simple and reliable manner neither for the early diagnosis nor for the follow-up of patients with retinal diseases. Although retinal imaging systems including optical coherence tomography (OCT)
      • Shu X.
      • Beckmann L.
      • Zhang H.
      Visible-light optical coherence tomography: a review.
      and fundus autofluorescence (AF)
      • Calvo-Maroto A.M.
      • Cerviño A.
      Spotlight on fundus autofluorescence.
      have revolutionized diagnosis and management of retinal diseases, allowing for the longitudinal observation of the retinal structure and the patterns of endogenous retinal fluorophores, their low lateral resolution hinders visualization of the RPE at the cellular level.
      The first in vivo images of a human RPE mosaic were obtained with adaptive optics (AO)-scanning laser ophthalmoscope (SLO) in regions where cones were missing in a patient with cone–rod dystrophy.
      • Roorda A.
      • Zhang Y.
      • Duncan J.L.
      High-resolution in vivo imaging of the RPE mosaic in eyes with retinal disease.
      Indeed, even if RPE cells are larger than photoreceptors and have strong light-scattering properties due to their pigments, when using en-face illumination, RPE are masked by the high reflectivity of the cones due to the optical Stiles-Crawford Effect.
      • Gao W.
      • Cense B.
      • Zhang Y.
      • Jonnal R.S.
      • Miller D.T.
      Measuring retinal contributions to the optical Stiles-Crawford effect with optical coherence tomography.
      The only commercially available AO-camera that uses en-face flood-illumination (rtx1TM camera, Imagine Eyes, Orsay, France) provides high resolution retinal images down to the photoreceptors, but poor visualization of the RPE cells. Noninvasive optical imaging methods combining AO-SLO with short wavelength autofluorescence (SW-AF) have been used for RPE cells visualization in the living human eye,
      • Morgan J.I.
      • Dubra A.
      • Wolfe R.
      • Merigan W.H.
      • Williams D.R.
      In vivo autofluorescence imaging of the human and macaque retinal pigment epithelial cell mosaic.
      but their clinical development has been limited by safety concerns associated with AF techniques. Reflectance imaging such as dark-field AO-SLO imaging is safer,
      • Scoles D.
      • Sulai Y.N.
      • Dubra A.
      In vivo dark-field imaging of the retinal pigment epithelium cell mosaic.
      but the quality of RPE images is reduced. AO-OCT, offering micrometer-level axial resolution,
      • Liu Z.
      • Kocaoglu O.P.
      • Miller D.T.
      3D Imaging of Retinal Pigment Epithelial Cells in the Living Human Retina.
      • Baraas R.C.
      • Pedersen H.R.
      • Knoblauch K.
      • Gilson S.J.
      Human Foveal Cone and RPE Cell Topographies and Their Correspondence With Foveal Shape.
      combined with organelle motion was proposed as a new contrast system to visualize RPE cells.
      • Liu Z.
      • Kurokawa K.
      • Hammer D.X.
      • Miller D.T.
      In vivo measurement of organelle motility in human retinal pigment epithelial cells.
      However, the time required for acquisition of these dynamic images makes it uncomfortable for patient and difficult to use. Combined with infrared (IR) or near infrared (NIR) AF, enhanced with indocyanin-green (ICG), or in multimodal setups, AO-SLO successfully identified the RPE mosaic in healthy volunteers,

      Tam J, Liu J, Dubra A, Fariss R. In Vivo Imaging of the Human Retinal Pigment Epithelial Mosaic Using Adaptive Optics Enhanced Indocyanine Green Ophthalmoscopy. Invest Ophthalmol Vis Sci. 2016;57(10):4376-4384.

      • Liu T.
      • Jung H.
      • Liu J.
      • Droettboom M.
      • Tam J.
      Noninvasive near infrared autofluorescence imaging of retinal pigment epithelial cells in the human retina using adaptive optics.
      • Granger C.E.
      • Yang Q.
      • Song H.
      • Saito K.
      • Nozato K.
      • Latchney L.R.
      • Leonard B.T.
      • Chung M.M.
      • Williams D.R.
      • Rossi E.A.
      Human Retinal Pigment Epithelium: In Vivo Cell Morphometry, Multispectral Autofluorescence, and Relationship to Cone Mosaic.
      • Vienola K.V.
      • Zhang M.
      • Snyder V.C.
      • Sahel J.A.
      • Dansingani K.K.
      • Rossi E.A.
      Microstructure of the retinal pigment epithelium near-infrared autofluorescence in healthy young eyes and in patients with AMD.
      • Bower A.J.
      • Liu T.
      • Aguilera N.
      • Li J.
      • Liu J.
      • Lu R.
      • Giannini J.P.
      • Huryn L.A.
      • Dubra A.
      • Liu T.
      • Hammer D.C.
      • Tam J.
      Integrating adaptive optics-SLO and OCT for multimodal visualization of the human retinal pigment epithelial mosaic.
      and patients.
      • Bower A.J.
      • Liu T.
      • Aguilera N.
      • Li J.
      • Liu J.
      • Lu R.
      • Giannini J.P.
      • Huryn L.A.
      • Dubra A.
      • Liu T.
      • Hammer D.C.
      • Tam J.
      Integrating adaptive optics-SLO and OCT for multimodal visualization of the human retinal pigment epithelial mosaic.
      ,
      • Grieve K.
      • Gofas-Salas E.
      • Ferguson R.D.
      • Sahel J.A.
      • Paques M.
      • Rossi E.A.
      In vivo near-infrared autofluorescence imaging of retinal pigment epithelial cells with 757 nm excitation.
      However, AF-based imaging methods generate significantly long lasting autofluorescence reduction at exposures lower than imposed by the safety norms.
      • Masella B.D.
      • Williams D.R.
      • Fischer W.S.
      • Rossi E.A.
      • Hunter J.J.
      Long-term reduction in infrared autofluorescence caused by infrared light below the maximum permissible exposure.
      Hence despite methodological improvements, RPE visualization remains hindered by safety concerns, and relatively long acquisition times.
      • Wynne N.
      • Carroll J.
      • Duncan J.L.
      Promises and pitfalls of evaluating photoreceptor-based retinal disease with adaptive optics scanning light ophthalmoscopy (AOSLO).
      Transscleral OPtical Imaging (TOPI) was developed to overcome these limitations. The unconventional, transscleral illumination of the ocular fundus provides high-contrast images of the retinal layers up to the RPE, at a cellular-resolution and over a large FOV (5°x5°).
      • Laforest T.
      • Künzi M.
      • Kowalczuk L.
      • Carpentras D.
      • Behar-Cohen F.
      • Moser C.
      Transscleral Optical Phase Imaging of the Human Retina.
      Oblique illumination avoids the large background light caused by the high reflectivity of cone photoreceptors providing a unique way to obtain high resolution images of RPE cells.
      • Laforest T.
      • Künzi M.
      • Kowalczuk L.
      • Carpentras D.
      • Behar-Cohen F.
      • Moser C.
      Transscleral Optical Phase Imaging of the Human Retina.
      A prototype for clinical use was designed with acquisition time below 10 seconds. This prototype was used in a single-center, exploratory, prospective, and descriptive clinical study to image RPE cells in healthy eyes with variable axial lengths (AL) from a significant number of participants with different ages, to quantify RPE morphological features, and to assess safety and repeatability of TOPI imaging.

      Methods

      This study (ClinicalTrials.gov: NCT04398394; kofam.ch: SNCTP000003921) was designed in accordance with the tenets of the Declaration of Helsinki, good clinical practice defined by the International Council for the Harmonization of Technical Requirements for the Registration of Pharmaceuticals for Human Use (ICH) or the ISO 14155, as well as all national legal and regulatory requirements. The Ethics Committee of the Swiss Federal Department of Health approved the study (Authorization CER-VD n ° 2019-00429).
      Written informed consent was received from all participants prior to inclusion. Healthy volunteers over 18 years old were included between the period of August 2020 and December 2020 at Jules-Gonin Eye Hospital, Lausanne, Switzerland. Only healthy eyes, emmetropic or ametropic between +3D and -6D, presenting with normal fundus examination, were included. Exclusion criteria included eyes with opacity of the anterior segment or other abnormality preventing good visualization of the fundus, eyes with any ocular disease, or in one of the following clinical situations: less than 3 months post-surgery of the anterior segment (eg cataract), eyes with strong myopia (< −6D), strong hypermetropia (> +5D) and/or strong astigmatism (> +4D). Not included were also individuals unable to follow the procedures of the study (for example due to language problems, psychological disorders, dementia), individuals unable to fix a target at least 20 seconds, individuals not tolerant of being in the dark for 30 minutes, individuals with epilepsy, individuals with albinism, and participants refusing to be informed of the incidental discovery of a clinically significant pathology.

      Clinical examinations

      Baseline examination included a standard ophthalmological examination, with best-corrected visual acuity (BCVA), intraocular pressure (IOP, Icare IC100 TA011, Medilas AG/Icare, Finland), spherical equivalent refractive error (NIDEK RT-6100, NIDEK CO, Japan) and AL (IOL MASTER 700, Carl Zeiss Meditec AG, Germany) measurements, color and autofluorescence fundus photography on the Optos ultra-wide field camera (Optos Daytona P200T, Dunfermline, Scotland), and Spectral-Domain (SD)-OCT (serial SD-OCT, 20°×20° 193-B-scans horizontal and vertical grids) on Spectralis (Heidelberg Engineering, Heidelberg, Germany). During this examination, an ophthalmologist ensured the absence of ophthalmic exclusion criteria.
      Safety evaluation included a follow-up visit scheduled between 1 and 3 weeks after TOPI examination, to exclude any occurrence of adverse event. It included eye examination with BCVA, IOP measurements, as well as eye fundus and posterior retina examinations with OPTOS camera and SD-OCT, using the follow-up mode on Spectralis. In addition to the demographic data (sex, age), the eye-related clinical data were collected, including: BCVA, AL, and spherical equivalent refractive error (RE).

      in vivo RPE imaging

      Principle of the modality. The details of the TOPI are fully described in the article published by Laforest et al.
      • Laforest T.
      • Künzi M.
      • Kowalczuk L.
      • Carpentras D.
      • Behar-Cohen F.
      • Moser C.
      Transscleral Optical Phase Imaging of the Human Retina.
      In brief, transscleral illumination of the retina is performed using two NIR light-emitting diodes (wavelength, λ = 850 nm, pulse peak power = 250 mW per LED, pulse duration = 8 ms, repetition rate = 11 Hz) located on the nasal and temporal side of the eye, coupled with an AO loop including a wavefront sensing (continuous illumination λ = 756 nm, peak power = 70 μW, maximum duration = 1800 seconds) embedded in our retina camera prototype. Transscleral illumination provides a large oblique beam which is scattered by the eye fundus. In short, the infrared light is first transmitted, through the sclera to the posterior segment. After reaching the retina and RPE, light propagates through the choroid and sclera, which are two tissues with strong scattering properties.
      • Hammer M.
      • Roggan A.
      • Schweitzer D.
      • Müller G.
      Optical properties of ocular fundus tissues--an in vitro study using the double-integrating-sphere technique and inverse Monte Carlo simulation.
      Part of the light travels then back towards the RPE, propagates through the neuroretinal translucent cells, the vitreous and the lens, and is collected through the pupil. Between 250 and 350 μm depth, the sclera behaves as an isotropic scatterer.
      • Hammer M.
      • Roggan A.
      • Schweitzer D.
      • Müller G.
      Optical properties of ocular fundus tissues--an in vitro study using the double-integrating-sphere technique and inverse Monte Carlo simulation.
      The light backscattered towards the RPE, which is essentially a secondary light source illuminating the RPE cells from below, contributes greatly to the contrast of the RPE signal. The latter also contains a smaller reflection component. In the ideal case where AO correction implemented in our system provides diffraction limited imaging (Rayleigh criterion),
      • Latychevskaia T.
      Lateral and axial resolution criteria in incoherent and coherent optics and holography, near- and far-field regimes.
      while considering a 6 mm pupil diameter at 850 nm, one gets a theoretical lateral and axial resolution of around 3 μm and 60 μm, respectively. With unavoidably imperfect AO correction the resolution is expected to be lower in our measurements.
      Image acquisition. For each acquisition, 100 raw images are captured in 10 seconds (10 frames per second, 8 ms pulse duration per raw image). The 100 raw images are then aligned and averaged to produce a single high signal-to-noise ratio image. Some features appear more or less contrasted on the dark-field images generated by each illumination beam, therefore the 2 transscleral beams are used simultaneously. On the operating software graphical user interface, five buttons corresponding to the five pre-defined zones are used to systematically select the imaged area. Alternatively, the operator can click anywhere on the drawing representing the wide-field fundus to freely choose an area to image. These two actions lead to moving the internal fixation target to a specific position (Fig S1). The foveal center is thus defined as the preferred retinal locus (PRL) of fixation, and then the internal fixation target guides the subject's gaze to select the imaged area at the defined eccentricities. When the target is centered, the imaging zone coincides with the PRL. When the target is moved, the eye rotates and remains centered on the fovea. As a result, the imaging zone is shifted away from the fovea.
      RPE imaging was performed the day of the screening visit. Three different operators were involved in all the image acquisition. The complete acquisition of an eye (Z1 to Z6) was performed by the same operator. In each included eye, five high resolution 5°x 5° RPE images were acquired (Fig 1A-B): four images equidistant from the fovea at an eccentricity of 5.37° (Z1: Inferonasal, Z2: Inferotemporal; Z3: superonasal; Z4: Superotemporal) and one image centered on the fovea (Z5). In some eyes, one additional image (Z6) was taken at the discretion of the investigator. One of the 6 zones was imaged 5 times to evaluate repeatability for a total of 10 images per eye. For each acquisition, the software records the iris image to check the alignment of the eye and a low-resolution oblique-illuminated 30°x30° IR fundus to locate the high resolution RPE image (Fig 1C). The system also records the internal fixation target coordinates as described in Fig S2. The full TOPI examination, including all the steps performed from the time the participant was greeted to the time he or she left after both eyes were imaged lasted between 30 and 60 minutes.
      Figure thumbnail gr1
      Fig 1Locations of the in vivo RPE images in the right eye of P029 (M, 29 years). (A) Spectralis infrared right eye fundus showing the 6 imaged zones. (B) in vivo RPE images in 6 areas (Z1: Inferonasal, Z2: Inferotemporal; Z3: superonasal; Z4: Superotemporal; Z5: Foveal center; Z6 Nasal). For each 5°x5° raw images, one 1.6°x1.6° sub-image with flat-field correction is magnified. (C) Example of iris pictures and low-resolution oblique-illuminated 30°x30° infrared fundus recorded during Z3 examination. Scale bars=200 μm

      RPE image processing and selection

      The image processing pipeline, coded in Matlab language (MATLAB. (2019). version 9.7.0.1216025 (R2019b) Update 1. Natick, Massachusetts: The MathWorks Inc.), is represented in Fig 2.
      Figure thumbnail gr2
      Fig 2Image processing pipeline. (A-B) First step: each 5°x5° raw RPE images (A) is divided in nine 1.6°x1.6° sub-images (B). Each sub-image is then processed by two sub-pipelines running in parallel. (C-E)Segmentation sub-pipeline”: (C) High-pass filtered image using a flat field correction with sigma = 10. (D) Centers found using the Difference of Gaussians filter with sigma1= 5 and sigma2= 10. (E) Voronoi-based cell segmentation. (F-H)Vessel and haze detection sub-pipeline”: (F) High-pass filtered image using a flat field correction with sigma = 3.1. (G) High-pass filtered image using Butterworth filter (cutoff frequency w = 80 pix-1). (H) Vessel and haze mask. Final step: The mask of segmented cells (I) is generated by removing the vessels and haze mask (H) from the segmentation (E). CR, Contrast reversing
      The first step of the image processing consists in dividing each 5°x5° raw image (Fig 2A) into 9 sub-images of FOV 1.6°x1.6° (Fig 2B). The eccentricities from the fovea of the sub-images are calculated using the target coordinates as the center coordinates of these raw images (Fig.S2). Each sub-image is then processed by two different sub-pipelines. In the “segmentation sub-pipeline” (Fig 2C-E), a flat-field correction (FFC) removes background intensity variations in the sub-images to generate a high-pass filtered image (Fig 2C). FFC subtracts from the original image the image transformed with a Gaussian blur, set with sigma = 10 pixels. This filter blurs the image with a Gaussian function that eliminates details and noise with a threshold defined by its standard deviation, the sigma parameter. Cell centers are then found in the resulting high-pass filtered image by a Difference of Gaussians filter (sigma1= 5; sigma2= 10). Local maxima are considered as cell centers (Fig 2D). A Voronoi-based segmentation is finally applied to find cell boundaries, using cell centers as seed points (Fig 2E). In the “vessel and haze detection sub-pipeline” (Fig 2F-H), the algorithm previously described,
      • Caetano Dos Santos F.L.
      • Laforest T.
      • Künzi M.
      • Kowalczuk L.
      • Behar-Cohen F.
      • Moser C.
      Fully automated detection, segmentation, and analysis of in vivo RPE single cells.
      is applied to remove hazy parts and vessels. The sub-image is high-pass filtered using a FFC with sigma = 3.1 (Fig 2F) and a first-order Butterworth filter with cutoff frequency w = 80 pix-1 (Fig 2G). The resulting image is processed to find parts to be filtered in the image. The final step consists in removing the binary mask (Fig 2H) from the segmentation (Fig 2E) to generate the final mask used to compute output parameters (Fig 2I).
      Quality criterion. The quality of the raw images was highly variable, ranging from poor quality due to noise or out-of-focus images to high quality in the in-focus images. In the latter, some regions revealed high-contrast RPE cells next to low-contrast regions. To quantify the quality of the RPE cell contrast, we established a "quality criterion" (QC) based on the signature of the RPE in the Fourier domain (Fig S3).
      On the low-pass Fourier-transformed raw image, the spatial frequencies were radially averaged (Fig S3A). The resulting profile showed a positive distortion, likened to RPE cells signature, which was not always a local maximum (Fig S3B). Since the shape and intensity of the distortion provide information about the contrast of the RPE cells in the image, three parameters were computed to quantify the distortion (Fig S3C): amplitude (ΔI), area (A) and gradient difference (ΔG) between both sides of the distortion. Amplitude and area were computed by interpolating a linear model between distortion’s inflexion points, resulting in the “QC” cost function (Fig S3D, equation). Each parameter ΔI, A and ΔG was normalized to take values in [0,1] without unit and weighted to tune their influence.
      Different RPE cell features were computed after segmentation: cell density (cells/mm2), cell area (μm2), and number of neighbor cells. On each segmented cell, the measured morphological features (Fig S4) were circularity (a.u.), elongation (a.u.), solidity (a.u.) and border distance coefficient of variation (CoV). The center to center spacing (CCS) based on the Fourier signature of the cells was also computed.
      To obtain the quantitative parameters, the pixel size was corrected for AL and spherical equivalent RE of the eye. Pixel size (μm) = -a*RE + b*(AL-23.5) + 0.74, where a and b come from the linear regression based on a simulation of the optical system, with different refractive error values and their corresponding optical magnification values (Supplemental Appendix).
      Image selection. In order to include only images with well-contrasted RPE cells in data analysis, the combination of four restricted conditions on cell features was used to select the best images (Fig 3E-F,H-I): i) measurable “Cell density” (>0); ii) CCS greater than 11.8 μm to avoid selecting images of other cell types or structures (Fig 3A-C); iii) QC greater than 0.076 to remove blurry and noisy images (Fig 3D,G); iv) “Number of neighboring cells” available, to remove images with only isolated cells detected or with cells detected on the border. Different CCS and QC thresholds were tested and the results were compared to the manual selection. The pair of thresholds that gave the closest selection to manual selection was kept. The CCS threshold of 11.8 μm is coherent with the published row-to-row spacing values.
      • Liu Z.
      • Kocaoglu O.P.
      • Miller D.T.
      3D Imaging of Retinal Pigment Epithelial Cells in the Living Human Retina.
      ,
      • Liu T.
      • Jung H.
      • Liu J.
      • Droettboom M.
      • Tam J.
      Noninvasive near infrared autofluorescence imaging of retinal pigment epithelial cells in the human retina using adaptive optics.
      ,
      • Bower A.J.
      • Liu T.
      • Aguilera N.
      • Li J.
      • Liu J.
      • Lu R.
      • Giannini J.P.
      • Huryn L.A.
      • Dubra A.
      • Liu T.
      • Hammer D.C.
      • Tam J.
      Integrating adaptive optics-SLO and OCT for multimodal visualization of the human retinal pigment epithelial mosaic.
      Figure thumbnail gr3
      Fig 3Illustration of the RPE images selected for the statistical analysis, in function of the center-to-center spacing (CCS) and the quality criterion (QC) values. Images for which the CCS was below 12 μm were excluded so as not to consider other cell types or structures (A, CCS=11.66 μm, QC=0.063, Left eye, from P006 (M, 27 years) at 6.6° superotemporal; B, CCS=9.64 μm, QC=0.0761 Right eye from P010 (F, 32 years) at 5.37° inferotemporal; C, CCS=8.95 μm, QC=0.157, Right eye from P010 at 5.83° superotemporal). Images for which the QC was below to 0.076 were excluded to not consider blurry images (D, CCS=11.85 μm, QC=0.0708, Right eye from P039 (M, 33 years) at 3.11° superotemporal; G, CCS=13.95 μm, QC=0.0756, Left eye from P010 at 4.39° inferonasal). The green square shows examples of RPE images selected for the analysis (E, CCS=12.23 μm, QC=0.0817, Right eye from P017 (M, 58 years) at 5.37° inferonasal; F, CCS=12.02 μm, QC=0.128, Left eye from P039 at 4.39° superonasal; H, CCS=13.98 μm, QC=0.0765 Right eye from P040 (F, 31 years) at 5.83° inferonasal; I, CCS=14.55 μm, QC=0.286, Right eye from P014 (F, 29 years) at 5.37° superonasal).

      Data analysis

      To study the repeatability of the measurements, images of the same area were realigned with Fiji plugin “Template Matching” that corrects for translations,
      • Tseng Q.
      • Duchemin-Pelletier E.
      • Deshiere A.
      • Balland M.
      • Guillou H.
      • Filhol O.
      • Théry M.
      Spatial organization of the extracellular matrix regulates cell-cell junction positioning.
      then “TurboReg” to adjust images in rotation and translation.
      • Thévenaz P.
      • Ruttimann U.E.
      • Unser M.
      A pyramid approach to subpixel registration based on intensity.
      The corrected images were then cropped to get the same area on each image, processed with the standard pipeline and selected with the restricted conditions described above after computation of RPE cell features. The CoV of each cell feature was calculated for each repetition.

      Statistical analysis

      The GraphPad Prism software (version 9.1.2 (226), GraphPad Software, LLC) was used to calculate descriptive statistics of the quantitative data (eye characteristics, mean cell features per subject), to test the normal distribution of the variables (age, AL, RE, BCVA, IOP, cell features) by the Kolmogorov-Smirnov normality tests, and to run the Spearman’s correlation test and linear regressions. To investigate the effect of AL, age and eccentricity, multiple comparisons of the variables (participants, eyes and cell features) between eccentricities, and groups of participants (males versus females; aged < 50 years versus > 50 years) was performed with the Multiple Mann-Whitney test, corrected with the Holm- Šídák method. This nonparametric test based on the distribution of ranked values in each group does not require an equal sample size. For all statistical tests, P-value <0.05 was considered significant.

      Results

      Image selection

      Forty-nine eyes presenting with good visual acuity from 29 healthy volunteers, 19 males and 10 females, with a mean age of 37.1 ± 13.3 years (range, 21-70 years) were included in this study. Participant and eye data collected at visit 1, including screening and TOPI examination, and at visit 2 for safety evaluation are summarized in the Supplemental Table S1. During TOPI examinations, no adverse event related to the procedure was reported. Comparisons of BCVA and IOP between both visits revealed no statistical difference (Mann-Whitney p=0.2917 and 0.4419, respectively). Multimodal, slit-lamp, fundoscopy and OCT imaging did not reveal any change on retinal structures before and after TOPI examination (Table S1).
      A total of 580 raw images were acquired, which means that 5,220 sub-images were processed. Comparison of the average QC per eye and per operator, on this initial data set, showed no inter-operator variability. After the automatic image selection that retained 1,553 sub-images, a manual check removed 47 other sub-images, including those at the center of the fovea image (Z5) because no RPE cells were visible at this location (see Fig 1B, Z5 magnification) and to avoid biasing the measurement at (0°, 0°). The final analyzed dataset included 1,506 high quality and well-contrasted RPE cells images, representing 29% of the total raw sub-images.

      Statistics on average RPE cell features

      Table S2 summarizes the number of sub-images selected per participant and per eccentricity. Three groups of participants were defined on the basis of usable foveal images and frequency of usable perifoveal images. The fovea was sampled between 1.6° and 2.9° in 10 participants (Group 1, Fig4A). The perifovea was consistently sampled, between 3° and 8° in 15 participants (10 in Group 1; 5 in Group 2, Fig4B) and irregularly sampled in 14 participants (Group 3, Fig4C). In addition, images were taken in retinal periphery, over 8° to 16.3°, at the discretion of the investigator, in 10 participants (7 in Group 1, 2 in Group 2, 1 in Group 3).
      Figure thumbnail gr4
      Fig 4Average cell density for participants as a function of sub-image eccentricity, in the nasal and temporal regions. Each point represents the average density of the inferior and superior quadrants in each region. Data collected at X=0 are located at 1.6° in the inferior and/or superior quadrants. (A) Group 1, 10 participants with consistently selected sub-images from the fovea to the perifovea. Cell density decreases with eccentricity for 1 participant (P043 (M, 28 years), orange circle) in the nasal (Y = 31.93*X + 3820; R2=44.6%, p=0.0176) and temporal regions (Y = -42.81*X + 3956; R2=78.8%, p<0.0001) and for 1 participant (P013 (F, 32 years), blue square) in the nasal region. (B) Group 2, 5 participants with consistently selected sub-images in the perifovea. Cell density decreases with eccentricity for 1 participant (P017 (M, 58 years), orange diamond) in the nasal region (Y = 126.9*X + 4211; R2=80.7%, p=0.015) and for 2 participants in the temporal region (P019 (F, 25 years), blue circle: Y = -26.95*X + 4424; R2=34.8%, p=0.0436; P016 (M, 39 years), green dot: Y = -77.96*X + 3986; R2=82.4%, p=0.0124). (C) Group 3, 14 participants with irregularly selected images in the perifovea. Cell density decreases with eccentricity for 1 participant (P025 (M, 52 years), black square) in the temporal region (Y = -68.27*X + 3750; R2=99%, p=0.0119). Raw data are available in Table S4.

      Effect of eccentricity

      Figure 4 represents the average density of each participant, in function of nasal and temporal eccentricities. Individual linear regressions showed significant correlations for six participants.
      To investigate the effect of eccentricity on cell density, only the data collected at shared eccentricities of the 15 participants from Groups 1 and 2 were considered. Multiple comparisons of average densities at each radial eccentricity (inferior versus superior and nasal versus temporal sides) showed no statistical difference, demonstrating that RPE cells densities are constant in the perifovea (Fig S5).
      Given the radial symmetry around the fovea, the effect of eccentricity on RPE cell features in the perifovea was investigated only in function of radial eccentricity. Linear regressions of RPE cell features in function of eccentricity in the perifovea showed no effect of eccentricity on cell density, area, CCS and circularity, and that cells have a number of neighbors close to 5.7, confirming the regularity the RPE cell between 3° to 8° (Fig 5). Morphological features elongation (R2=5.20%, p=0.0316), border distance CoV (R2=5.38%, p=0.0287) and solidity (R2=6.17%, p=0.0189) showed a slight correlation with eccentricity.
      Figure thumbnail gr5
      Fig 5Linear regressions of RPE cell features in function of eccentricity. Each point represents the mean ± SD of average features from the 15 participants of Groups 1 and 2, at each shared radial eccentricity in the perifoveal region. Between 3° and 8°, eccentricity has no influence on RPE cell density (Y = -6.194*X + 3804; R2=0.09%, p=0.7829), area (Y = 0.5961*X + 222.6; R2=0.22%, p=0.6637), center-to-center spacing (Y = 0.1049*X + 14.14; R2=3.59%, p=0.0753), number of neighbors (Y = -0.01145*X + 5.762; R2=1.35%, p=0.2774) and circularity (Y = 0.002182*X + 0.8851; R2=3.93%, p=0.0626) and little influence on the other morphological features: elongation (Y = -0.003524*X + 0.6326; R2=5.20%, p=0.0316; *), border distance CoV (Y = -0.001835*X + 0.1639; R2=5.38%, p=0.0287; *) and solidity (Y = 0.0002037*X + 0.9506; R2=6.17%, p=0.0189; *). Raw data are available in Table S5.

      Correlation between age, eye characteristics and cell features

      Considering that eccentricity has little influence on RPE cell characteristics over this narrow range of eccentricities, the data computed from the 1,397 sub-images from the 29 participants between 3° and 8° were analyzed to investigate the relationship between age, eye characteristics and cell features.
      The normality test revealed that continuous variables related to patient and eye characteristics do not follow a normal distribution. Therefore, the non-parametric Spearman coefficient was used to measure correlation between participant age, eye data and RPE cell features (Table 1).
      Table 1Spearman’s correlation matrix between participant data (age, sex), eye data (axial length, refractive error) and RPE cell features mean and standard deviation (SD). A grayscale legend indicates the corresponding P value, from white (not significant, ns) to dark gray (highly significant, <0.001).
      Table thumbnail fx1
      The Spearman matrix shows that RPE cell morphological features are correlated between them. For example, increases in elongation (mean) and Border distance CoV (mean and SD) are correlated with decreases in circularity (mean and SD), solidity (mean and SD), number of neighboring cells and in density, and with an increase in cell area (mean and SD). The mean number of neighboring cells (5.7 ± 0.11; range 5.3 - 5.8) remains constant whatever the age and sex of the participant and the AL of the eye. The distribution of the number of neighbors confirmed that a large majority of RPE cells have a triangular packing arrangement (Fig S6A).

      Effect of axial length

      As expected, AL was correlated with refractive error (Table 2; Rs=-0.563; p=0.001). Considering the corrected pixel size, the cell density decreased with AL (Rs=-0.746; p<0.0001). This was correlated with an increase in cell area (Rs=0.668; p<0.0001), without change in cell morphological features, such as circularity (illustrated in Fig 5A; Rs=-0.111; p=0.565). Linear regressions were consistent with these findings. Fig 5B-C illustrates one image from the shortest eye (F, 25 years), compared to the longest eye (M, 29 years). The morphological parameters show larger cells in the long eye, with a geometry similar to those of the short eye.
      Table 2Multiple Mann-Whitney comparisons of the RPE cell features computed from the images obtained between 3° and 8°, per sex group: Females (n=10 participants; AL mean=23.44 ± 0.88 mm, median=23.54 mm, range 21.99-24.95 mm) versus Males (n=19 participants; AL mean=24.34 ± 0.80 mm, median=24.30 mm, range 22.7-26.61 mm).
      Table thumbnail fx2
      Mean AL of Females (n=10 participants, 23.44 ± 0.88 mm) was significantly lower than the one of Males (n=19 participants, 24.34 ± 0.80 mm; p=0.0055). Multiple comparisons demonstrated that none of the participant (age), eye (AL) or RPE cell features variables were significantly different between males and females (Table 2).

      Effect of age

      A significant decrease in cell density with age was measured (Fig 6A ; Rs=-0.391; p=0.036), implying an increase in mean cell area (Rs=0.454; p=0.013). The shape of the RPE cells also changed with age, as demonstrated by the calculated Spearman coefficients and their related significant P-values, for circularity (illustrated in Fig 6A; Rs=-0.569; p=0.0013), elongation (Rs=0.562; p=0.001), solidity (Rs=-0.406; p=0.028), border distance CoV mean (Rs=0.580; p=0.001). This result is illustrated in the comparison of one young participant with one of the oldest volunteers. It shows less circular, less symmetric, elongated and larger cells in the image of the older eye (Fig 6B-C).
      Figure thumbnail gr6
      Fig 6Effect of axial length (AL) on RPE cell features. (A) Linear regressions of axial length in function of cell density and circularity. (B-C) Illustration of quantification in a sub-image of the shorter eye (B, AL=21.88mm, left eye from P019 (F, 25 years) at 5.37° superotemporal) versus one of the longer eyes (C, AL=26.51mm, Left eye from P029 (M, 29 years) at 4.39° superotemporal).
      Figure thumbnail gr7
      Fig 7Effect of Age on RPE cell features. (A) Linear regressions of age in function of cell density and circularity. (B-C) Illustration of quantification in a sub-image of one of the younger participants (B, 27 years, left eye from P002 (F) at 6.6° superotemporal) versus one of the older participants (C, 62 years, left eye from P007 (M) at 5.37° superotemporal).
      To assess the effect of age on these independent variables, the cohort was divided into two groups: participants under 50 years and those over 50 years. The multiple non-parametric test demonstrated that the RPE cell features that significantly differed over 50 years are circularity (mean), border distance CoV (mean), elongation (mean), area (mean) and density (Table 3). The mean number of neighbors was not different between groups, but the distribution of the average number of neighbors per sub-image depended on the age group of participants as shown in Fig S6B. An average of 5.8 neighbors was found in 35% of the images for participants <50 years, whereas this figure drops down to only 20% for participants over 50 years. This suggests a rearrangement of the RPE cells with aging.
      Table 3Multiple Mann-Whitney comparison of the RPE cell features computed from the images obtained between 3° and 8°, per age group: “< 50 years” (n=22 participants; Age mean=30.3 ± 4.7 years, median=29 years, range 21 – 41 years) versus “> 50 years” (n=7 participants; Age mean=58.4 ± 7.3 mm, median=58 years, range 51-70 years).
      Table thumbnail fx3

      Repeatability of the acquisition

      To test the setup repeatability, the same operator, under the same conditions, took images of the same area few minutes apart (Fig S7). After selection of the well-contrasted RPE images, the images taken in 25 eyes from 17 participants (10M, 7F; age 35.9 ± 13.8 years, range 25 – 70 years) were included in this analysis. Repetitions were done on the inferonasal area (Z1) for 24 eyes and on the inferotemporal area (Z2) for 1 eye, with an average number of selected images of 4.7 ± 1.4 per subject. After re-alignment, segmentation and analysis of the images, the cell features were computed. The computed CoV (mean ± SD) of the mean cell features measured in the repeated areas are presented in Table 4. For each feature, the average CoV was less than 4 %.
      Table 4Cell features coefficient of variation (CoV) on 25 areas, imaged 4.7 ±1.4 times
      Cell featuresMean CoVSD CoV
      Cell density (cells/mm2)0.0320.013
      Area (μm2)0.0350.016
      Nb of neighbor cells0.0240.021
      Circularity (a.u.)0.0100.004
      Elongation (a.u.)0.0230.011
      Solidity (a.u.)0.0010.001
      Border distance Cov (a.u.)0.0390.017

      Discussion

      Using TOPI, the RPE could safely be imaged in 49 eyes from 29 healthy volunteers aged 21 to 70, over the macula, between 1.6 and 16.3° eccentricity from the fovea. In all tested eyes, with mean axial length of 24.03 ± 0.93 mm (range 21.88 to 26.7 mm), high resolution images of RPE were obtained for analysis.
      The previously described image analysis tool for RPE cell detection, segmentation and quantification was used with strict criteria to select well-contrasted cells, thus removing not only the vessels area but also the blurred and low-contrast areas on the RPE cells.
      • Caetano Dos Santos F.L.
      • Laforest T.
      • Künzi M.
      • Kowalczuk L.
      • Behar-Cohen F.
      • Moser C.
      Fully automated detection, segmentation, and analysis of in vivo RPE single cells.
      Ideally, with our imaging modality relying on transillumination and melanin absorption, one RPE cell appears as a uniform black central area surrounded by a bright interface. A central area with non-uniform intensity distribution or with an intensity significantly different from that of its neighboring cells may result in poor imaging of the cell borders. This explains why the RPE cell pattern was not well resolved in some areas, even in high-quality images. The reason for such a non-ideal distribution may be twofold: (i) The density of melanin pigments in the granules fluctuates and the latter are randomly distributed in the cell apical layer,
      • Pollreisz A.
      • Messinger J.D.
      • Sloan K.R.
      • Mittermueller T.
      • Weinhandl A.S.
      • Benson E.
      • Kidd G.J.
      • Schmidt-Erfurth U.
      • Curcio C.A.
      Visualizing melanosomes, lipofuscin, and melanolipofuscin in human retinal pigment epithelium using serial block face scanning electron microscopy.
      ,
      • Pollreisz A.
      • Neschi M.
      • Sloan K.R.
      • Pircher M.
      • Mittermueller T.J.
      • Dacey D.M.
      • Schmidt-Erfurth U.
      • Curcio C.A.
      An atlas of human retinal pigment epithelium organelles significant for clinical imaging.
      as well as between neighboring cells. This results in a non-homogeneous random contrast pattern that may give rise to ambiguous discrimination of cell borders; (ii) transillumination may be disturbed or attenuated by vessels or melanin granules in the choroidal layers altering the distribution of light projected to the RPE and distorting image contrast. Our current prototype did not allow imaging the RPE cells in the center of the fovea, which prevented us from performing analyses in this area. The origin of the lower image quality in the foveal center is not yet completely clear. It might simply be due to non-optimal focusing or to a stronger perturbation of the transillumination mechanisms inherent to our method in the foveal area. This hypothesis is supported by the higher vascular density in the choriocapillaris,
      • Bartol-Puyal F.A.
      • Isanta C.
      • Calvo P.
      • Ruiz-Moreno Ó.
      • Pablo L.
      Relationship between vascular densities of choriocapillaris and the whole choroid using OCTA.
      and concentration of melanin granules in the choroid behind the fovea.
      • Weiter J.J.
      • Delori F.C.
      • Wing G.L.
      • Fitch K.A.
      Retinal pigment epithelial lipofuscin and melanin and choroidal melanin in human eyes.
      Only one third of the sub-images were selected due to the low-contrast on RPE cells border in some areas, together with the strict criteria applied in order to keep only high-quality data to use the Voronoi algorithm. Most of these high-quality sub-images were located between 3° and 8° of eccentricity. In our Voronoi diagrams, we assumed a convex shape. The resulting diagrams presented a honeycomb mosaic which is consistent with the mosaic of healthy RPE cells, allowing to the diagrams to estimate RPE’s morphological features. When using the diagrams to compute the number of neighbors around each RPE cell, we found a value of 5.68 ± 0.11 (range 5.3 - 5.8) which is very similar to the value reported by Ortolan et al. in the macular region (5.56 ±0.35) in ex vivo human retina with a method to define the borders of each cell different from the Voronoi method.
      • Ortolan D.
      • Sharma R.
      • Volkov A.
      • Maminishkis A.
      • Hotaling N.A.
      • Huryn L.A.
      • Cukras C.
      • Di Marco S.
      • Bisti S.
      • Bharti K.
      Single-cell-resolution map of human retinal pigment epithelium helps discover subpopulations with differential disease sensitivity.
      The distribution of the number of neighbors is also coherent with the studies in which Voronoi was used to determine cell shape.
      • Ach T.
      • Huisingh C.
      • McGwin Jr., G.
      • Messinger J.D.
      • Zhang T.
      • Bentley M.J.
      • Gutierrez D.B.
      • Ablonczy Z.
      • Smith R.T.
      • Sloan K.R.
      • Curcio C.A.
      Quantitative autofluorescence and cell density maps of the human retinal pigment epithelium.
      ,
      • Liu Z.
      • Kocaoglu O.P.
      • Miller D.T.
      3D Imaging of Retinal Pigment Epithelial Cells in the Living Human Retina.
      The average CoV of all the cell features, measured in areas imaged multiple times, was less than 4%, demonstrating the good repeatability of image acquisition with TOPI.
      • Bartlett J.W.
      • Frost C.
      Reliability, repeatability and reproducibility: analysis of measurement errors in continuous variables.
      The literature analysis shows that depending on the in vivo imaging modality (Table 5A), the histological methods (Table 5B), the populations and the retinal locations, results on human RPE cell density may vary significantly, which makes the comparison with our results not trivial. Taking this into account, in the perifovea, the RPE densities that we have measured are lower than some in vivo
      • Liu Z.
      • Kocaoglu O.P.
      • Miller D.T.
      3D Imaging of Retinal Pigment Epithelial Cells in the Living Human Retina.
      ,
      • Liu T.
      • Jung H.
      • Liu J.
      • Droettboom M.
      • Tam J.
      Noninvasive near infrared autofluorescence imaging of retinal pigment epithelial cells in the human retina using adaptive optics.
      and ex vivo
      • Ach T.
      • Huisingh C.
      • McGwin Jr., G.
      • Messinger J.D.
      • Zhang T.
      • Bentley M.J.
      • Gutierrez D.B.
      • Ablonczy Z.
      • Smith R.T.
      • Sloan K.R.
      • Curcio C.A.
      Quantitative autofluorescence and cell density maps of the human retinal pigment epithelium.
      measurements, but consistent with our previous data
      • Laforest T.
      • Künzi M.
      • Kowalczuk L.
      • Carpentras D.
      • Behar-Cohen F.
      • Moser C.
      Transscleral Optical Phase Imaging of the Human Retina.
      and with those of Morgan et al.
      • Morgan J.I.
      • Dubra A.
      • Wolfe R.
      • Merigan W.H.
      • Williams D.R.
      In vivo autofluorescence imaging of the human and macaque retinal pigment epithelial cell mosaic.
      In the periphery, the measured densities are lower than some in vivo
      • Liu T.
      • Jung H.
      • Liu J.
      • Droettboom M.
      • Tam J.
      Noninvasive near infrared autofluorescence imaging of retinal pigment epithelial cells in the human retina using adaptive optics.
      and ex vivo
      • Watzke R.C.
      • Soldevilla J.D.
      • Trune D.R.
      Morphometric analysis of human retinal pigment epithelium: correlation with age and location.
      measurements, and consistent with those obtained in vivo by Morgan et al.
      • Morgan J.I.
      • Dubra A.
      • Wolfe R.
      • Merigan W.H.
      • Williams D.R.
      In vivo autofluorescence imaging of the human and macaque retinal pigment epithelial cell mosaic.
      and Granger et al.,
      • Granger C.E.
      • Yang Q.
      • Song H.
      • Saito K.
      • Nozato K.
      • Latchney L.R.
      • Leonard B.T.
      • Chung M.M.
      • Williams D.R.
      • Rossi E.A.
      Human Retinal Pigment Epithelium: In Vivo Cell Morphometry, Multispectral Autofluorescence, and Relationship to Cone Mosaic.
      and ex vivo by Panda-Jonas et al.
      • Panda-Jonas S.
      • Jonas J.B.
      • Jakobczyk-Zmija M.
      Retinal pigment epithelial cell count, distribution, and correlations in normal human eyes.
      and Ach et al.
      • Ach T.
      • Huisingh C.
      • McGwin Jr., G.
      • Messinger J.D.
      • Zhang T.
      • Bentley M.J.
      • Gutierrez D.B.
      • Ablonczy Z.
      • Smith R.T.
      • Sloan K.R.
      • Curcio C.A.
      Quantitative autofluorescence and cell density maps of the human retinal pigment epithelium.
      The measured RPE cell spacings are in agreement with the previously reported values.
      • Liu Z.
      • Kocaoglu O.P.
      • Miller D.T.
      3D Imaging of Retinal Pigment Epithelial Cells in the Living Human Retina.
      ,
      • Liu T.
      • Jung H.
      • Liu J.
      • Droettboom M.
      • Tam J.
      Noninvasive near infrared autofluorescence imaging of retinal pigment epithelial cells in the human retina using adaptive optics.
      ,
      • Granger C.E.
      • Yang Q.
      • Song H.
      • Saito K.
      • Nozato K.
      • Latchney L.R.
      • Leonard B.T.
      • Chung M.M.
      • Williams D.R.
      • Rossi E.A.
      Human Retinal Pigment Epithelium: In Vivo Cell Morphometry, Multispectral Autofluorescence, and Relationship to Cone Mosaic.
      ,
      • Laforest T.
      • Künzi M.
      • Kowalczuk L.
      • Carpentras D.
      • Behar-Cohen F.
      • Moser C.
      Transscleral Optical Phase Imaging of the Human Retina.
      Table 5ASummary of in vivo RPE cell imaging studies in healthy eyes. For comparison, the reported data from the fovea to the periphery are summarized. Eccentricity expressed in mm are estimated in degree, assuming a 24 mm emmetropic eye with 0.288 mm/degree.54 # Mean of average density per subject in the specified eccentricity range were calculated. N/A, raw data not available; RRS, Row-to-row spacing. (Level of significance of Unpaired t test with Welch's correctio: ***, p<0.001; ns, non-significant).
      Table thumbnail fx4
      Table 5BSummary of ex vivo RPE cell imaging studies in healthy eyes. For comparison, the reported data from the fovea to the periphery are summarized. Eccentricity expressed in mm are estimated in degree, assuming a 24 mm emmetropic eye with 0.288 mm/degree.54 (Level of significance of Unpaired t test with Welch's correction: *, p<0.05; **, p<0.01; ***, p<0.001; ns, non-significant).
      Table thumbnail fx5
      RPE’s morphological features were then correlated with eccentricity, age, and axial length. The results confirmed the regularity of the RPE mosaic in a narrow range of eccentricities in the macula, and showed variability of RPE features between participants, related to the wide range of axial lengths and ages of the population included in the study. Eye elongation was significantly associated with a decrease in RPE cell density (implying an increase in cell area) without morphological change. These results do not corroborate with the ex vivo studies of Jonas et al. which report that axial elongation is associated with a decrease in RPE cell density from or posterior to the equator, whereas RPE density in the macular region is independent of axial length.
      • Jonas J.B.
      • Ohno-Matsui K.
      • Jiang W.J.
      • Panda-Jonas S.
      BRUCH MEMBRANE AND THE MECHANISM OF MYOPIZATION: A New Theory.
      There are many factors that could explain change in the size of the macular RPE cells including mechanical stress resulting from the activity on sub foveal smooth muscle cells fibers that are involved in the focalization process.
      • Nickla D.L.
      • Wallman J.
      The multifunctional choroid.
      ,
      • Flügel-Koch C.
      • May C.A.
      • Lütjen-Drecoll E.
      Presence of a contractile cell network in the human choroid.
      In pathologic myopia posterior staphyloma develops in the posterior pole and not at the periphery demonstrating that major alterations of all retinal layers might take place also in the macula and not only in the periphery of myopic eyes. The observed changes in RPE cell density in the larger eyes are a new finding that needs to be confirmed in highly myopic eyes (<-6 diopters) and in a larger population.
      Regarding the effect of age, in a previous set of data,
      • Laforest T.
      • Künzi M.
      • Kowalczuk L.
      • Carpentras D.
      • Behar-Cohen F.
      • Moser C.
      Transscleral Optical Phase Imaging of the Human Retina.
      from 11 healthy volunteers of younger age (28 ±9 years), we reported a higher mean RPE density decreasing with age (R2=34%). The present study, conducted on a larger number of older participants, confirms that aging is slightly but statistically associated with a reduction in macular RPE cell density, and further shows that it is associated with a change in cell morphology. The published results on the influence of aging on RPE cell density in the macula of normal human eyes are highly variable. Two in vivo studies reported no change in RPE density with aging, but participants older than 50 years were only 4 out of 23 in Baraas et al.,
      • Baraas R.C.
      • Pedersen H.R.
      • Knoblauch K.
      • Gilson S.J.
      Human Foveal Cone and RPE Cell Topographies and Their Correspondence With Foveal Shape.
      and 3 out of 10 in Granger et al.
      • Granger C.E.
      • Yang Q.
      • Song H.
      • Saito K.
      • Nozato K.
      • Latchney L.R.
      • Leonard B.T.
      • Chung M.M.
      • Williams D.R.
      • Rossi E.A.
      Human Retinal Pigment Epithelium: In Vivo Cell Morphometry, Multispectral Autofluorescence, and Relationship to Cone Mosaic.
      (Table 5A). Data from larger samples, acquired on post-mortem donor eyes, have reported divergent results (Table 5B). One study found unchanged RPE density in the fovea and increased RPE density in the perifovea with age,
      • Ach T.
      • Huisingh C.
      • McGwin Jr., G.
      • Messinger J.D.
      • Zhang T.
      • Bentley M.J.
      • Gutierrez D.B.
      • Ablonczy Z.
      • Smith R.T.
      • Sloan K.R.
      • Curcio C.A.
      Quantitative autofluorescence and cell density maps of the human retinal pigment epithelium.
      whereas others found stable,
      • Gao H.
      • Hollyfield J.G.
      Aging of the human retina. Differential loss of neurons and retinal pigment epithelial cells.
      ,
      • Watzke R.C.
      • Soldevilla J.D.
      • Trune D.R.
      Morphometric analysis of human retinal pigment epithelium: correlation with age and location.
      ,
      • Harman A.M.
      • Fleming P.A.
      • Hoskins R.V.
      • Moore S.R.
      Development and aging of cell topography in the human retinal pigment epithelium.
      ,
      • Del Priore L.V.
      • Kuo Y.H.
      • Tezel T.H.
      Age-Related Changes in Human RPE Cell Density and Apoptosis Proportion In Situ.
      or decreased density in the macula
      • Dorey C.K.
      • Wu G.
      • Ebenstein D.
      • Garsd A.
      • Weiter J.J.
      Cell loss in the aging retina. Relationship to lipofuscin accumulation and macular degeneration.
      ,
      • Feeney-Burns L.
      • Burns R.P.
      • Gao C.L.
      Age-related macular changes in humans over 90 years old.
      ,
      • Panda-Jonas S.
      • Jonas J.B.
      • Jakobczyk-Zmija M.
      Retinal pigment epithelial cell count, distribution, and correlations in normal human eyes.
      ,
      • Bhatia S.K.
      • Rashid A.
      • Chrenek M.A.
      • Zhang Q.
      • Bruce B.B.
      • Klein M.
      • Boatright J.H.
      • Jiang Y.
      • Grossniklaus H.E.
      • Nickerson J.M.
      Analysis of RPE morphometry in human eyes.
      and the posterior pole.
      • Ts'o M.O.M.
      • Friedman E.
      The retinal pigment epithelium: III. Growth and development.
      Finally, since the study by Watzke et al. that reported a loss of hexagonality with age in the fovea,
      • Watzke R.C.
      • Soldevilla J.D.
      • Trune D.R.
      Morphometric analysis of human retinal pigment epithelium: correlation with age and location.
      three studies have explored the morphometry of RPE cells. Ach et al. examined ex vivo the regularity of RPE cells from 20 donors (10 younger than 51 years of age) and showed that the number of cells with six neighbors decreases significantly with age in the fovea and periphery, indicating that rearrangement of RPE cells occurs throughout life.
      • Ach T.
      • Huisingh C.
      • McGwin Jr., G.
      • Messinger J.D.
      • Zhang T.
      • Bentley M.J.
      • Gutierrez D.B.
      • Ablonczy Z.
      • Smith R.T.
      • Sloan K.R.
      • Curcio C.A.
      Quantitative autofluorescence and cell density maps of the human retinal pigment epithelium.
      With the same Voronoi-based metric, a slight decrease in cell regularity with age was also reported in vivo by Liu et al. in 6 participants aged 25 to 61 years.
      • Liu Z.
      • Kocaoglu O.P.
      • Miller D.T.
      3D Imaging of Retinal Pigment Epithelial Cells in the Living Human Retina.
      Our results regarding the distribution of the average number of neighbors per sub-image also showed a decrease in the prevalence of 5.8 nearest neighbors from 35% (under 50 years) to 20% (over 50 years), confirming the rearrangement of RPE cells with aging. Our results are consistent with those of Bhatia et al.
      • Bhatia S.K.
      • Rashid A.
      • Chrenek M.A.
      • Zhang Q.
      • Bruce B.B.
      • Klein M.
      • Boatright J.H.
      • Jiang Y.
      • Grossniklaus H.E.
      • Nickerson J.M.
      Analysis of RPE morphometry in human eyes.
      who described a decrease in RPE cell density with age in healthy macula from 10 donors, aged 29 to 80 years, as well as changes in RPE cell morphometry. They computed shape and eccentricity factors that can be compared with our circularity and border distance CoV factors, respectively, and reported that macular RPE cells had a larger area, a more elongated shape and were less symmetrical with age. The RPE cell features analyzed in vivo in the present study correlate with features analyzed on post mortem human RPE samples. Together with the repeatability measurements, the results constitute evidence towards reliability of the method.
      AO-SLO-based systems can image the RPE with different modalities such as dark-field, OCT, NIRAF and enhanced-ICG. The flexibility offered by these methods can provide multiple perspectives of the RPE mosaic depending on contrast mechanisms based on reflectance or fluorescence.
      • Bower A.J.
      • Liu T.
      • Aguilera N.
      • Li J.
      • Liu J.
      • Lu R.
      • Giannini J.P.
      • Huryn L.A.
      • Dubra A.
      • Liu T.
      • Hammer D.C.
      • Tam J.
      Integrating adaptive optics-SLO and OCT for multimodal visualization of the human retinal pigment epithelial mosaic.
      Oblique flood-illumination used in TOPI exploits a different contrast mechanism, namely transillumination generated by backscattering, to image the RPE cells, thus offering a new perspective from the cells and their environment. Compared to AO-SLO systems, the design of our TOPI instrument has a similar lateral resolution but provides a larger FOV (5° x 5°), corresponding to the largest FOV an AO system can acquire in a single acquisition. Moreover, TOPI needs a lower radiant exposure than AO-SLO systems,
      • Laforest T.
      • Künzi M.
      • Kowalczuk L.
      • Carpentras D.
      • Behar-Cohen F.
      • Moser C.
      Transscleral Optical Phase Imaging of the Human Retina.
      that is three orders of magnitude lower than the imposed by the American National Standards Institute norms on maximum permissible exposure. This is one important advantage of our method given the alarming evidence of hazards when imaging in the NIR at exposures four to five times lower than the current safety limits, as revealed by long-term reduction of NIRAF.
      • Masella B.D.
      • Williams D.R.
      • Fischer W.S.
      • Rossi E.A.
      • Hunter J.J.
      Long-term reduction in infrared autofluorescence caused by infrared light below the maximum permissible exposure.
      A few limitations were observed in our study. First, stability of fixation has not been measured, for example, by microperimetry, and differences between the PRL and anatomical foveal landmarks were not investigated. It has been reported that flood-illumination AO ophthalmoscopes for photoreceptor imaging exhibit, in healthy subjects, deviation of the PRL from 50-μm (equivalent to approximately 0.15°) as compared the location of peak cone density,
      • Putnam N.M.
      • Hofer H.J.
      • Doble N.
      • Chen L.
      • Carroll J.
      • Williams D.R.
      The locus of fixation and the foveal cone mosaic.
      to an average of 0.85° as compared to the foveal pit center.

      Roshandel D, Sampson DM, Mackey DA, Chen FK. Impact of Reference Center Choice on Adaptive Optics Imaging Cone Mosaic Analysis. Invest Ophthalmol Vis Sci. 2022;63(4):12.

      In our study, participant fixation and low-resolution fundus visualization in our prototype allowed to localize the fovea and check the approximate eccentricity accordingly. Second, the measurements performed with the prototype tested in this study acquired only one-third high-contrast images on the RPE cells and did not image RPE foveal cells with sufficient accuracy. Improvements in the hardware and software of the next version of the camera, the details of which are beyond the scope of this study, are expected to improve the contrast on the RPE cells and to acquire images of the cells in the foveal center. Finally, another limitation is the relatively low number of participants. Although the current study provides a large quantitative and qualitative analysis of human RPE cells, performed in vivo in healthy volunteers, larger homogeneous patient populations are necessary to enrich the normative quantitative data and subsequently describe pathological situations.
      Since TOPI allows for fast image acquisition and is user-friendly, it should provide new insights into RPE cell in normal conditions and identify early RPE changes that might predispose to retinal diseases and objective surrogate morphological RPE markers to test new therapies for degenerative retinal diseases.

      Uncited reference

      Kolb H. Facts and Figures Concerning the Human Retina. 2005 May 1 [updated 2007 Jul 5]. In: Kolb H, Fernandez E, Nelson R, editors. Webvision: The Organization of the Retina and Visual System [Internet]. Salt Lake City (UT): University of Utah Health Sciences Center; 1995–. PMID: 21413409.

      .

      Acknowledgement

      This study is part of the project ASSESS [retinAI phase contrast imaging for Early diagnoSiS) that has received funding from EIT Health. EIT Health is supported by the European Institute of Innovation and Technology (EIT), a body of the European Union receives support from the European Union´s Horizon 2020 Research and innovation programme.

      References

        • Lakkaraju A.
        • Umapathy A.
        • Tan L.X.
        • Daniele L.
        • Philp N.J.
        • Boesze-Battaglia K.
        • Williams D.S.
        The cell biology of the retinal pigment epithelium.
        Prog Retin Eye Res. 2020; 100846
        • Khan K.N.
        • Mahroo O.A.
        • Khan R.S.
        • Mohamed M.D.
        • McKibbin M.
        • Bird A.
        • Michaelides M.
        • Tufail A.
        • Moore A.T.
        Differentiating drusen: Drusen and drusen-like appearances associated with ageing, age-related macular degeneration, inherited eye disease and other pathological processes.
        Prog Retin Eye Res. 2016; 53: 70-106
        • Simó R.
        • Villarroel M.
        • Corraliza L.
        • Hernández C.
        • Garcia-Ramírez M.
        The retinal pigment epithelium: something more than a constituent of the blood-retinal barrier--implications for the pathogenesis of diabetic retinopathy.
        J Biomed Biotechnol. 2010; 190724
        • Rothschild P.R.
        • Salah S.
        • Berdugo M.
        • Gélizé E.
        • Delaunay K.
        • Naud M.C.
        • Klein C.
        • Moulin A.
        • Savoldelli M.
        • Bergin C.
        • Jeanny J.C.
        • Jonet L.
        • Arsenijevic Y.
        • Behar-Cohen F.
        • Crisanti P.
        ROCK-1 mediates diabetes-induced retinal pigment epithelial and endothelial cell blebbing: Contribution to diabetic retinopathy.
        Sci Rep. 2017; 7: 8834
        • Cheung C.M.G.
        • Lee W.K.
        • Koizumi H.
        • Dansingani K.
        • Lai T.Y.Y.
        • Freund K.B.
        Pachychoroid disease.
        Eye. 2019; 33: 14-33
        • Feeney-Burns L.
        • Hilderbrand E.S.
        • Eldridge S.
        Aging human RPE: morphometric analysis of macular, equatorial, and peripheral cells.
        Invest Ophthalmol Vis Sci. 1984 Feb; 25: 195-200
        • Dorey C.K.
        • Wu G.
        • Ebenstein D.
        • Garsd A.
        • Weiter J.J.
        Cell loss in the aging retina. Relationship to lipofuscin accumulation and macular degeneration.
        Invest. Ophthalmol. Vis. Sci. 1989; 30: 1691-1699
        • Feeney-Burns L.
        • Burns R.P.
        • Gao C.L.
        Age-related macular changes in humans over 90 years old.
        Am J Ophthalmol. 1990; 109: 265-278
        • Gao H.
        • Hollyfield J.G.
        Aging of the human retina. Differential loss of neurons and retinal pigment epithelial cells.
        Invest Ophthalmol Vis Sci. 1992; 33: 1-17
        • Watzke R.C.
        • Soldevilla J.D.
        • Trune D.R.
        Morphometric analysis of human retinal pigment epithelium: correlation with age and location.
        Curr Eye Res. 1993; 12: 133-142
        • Panda-Jonas S.
        • Jonas J.B.
        • Jakobczyk-Zmija M.
        Retinal pigment epithelial cell count, distribution, and correlations in normal human eyes.
        Am J Ophthalmol. 1996; 121: 181-189
        • Harman A.M.
        • Fleming P.A.
        • Hoskins R.V.
        • Moore S.R.
        Development and aging of cell topography in the human retinal pigment epithelium.
        Invest Ophthalmol Vis Sci. 1997; 38: 2016-2026
        • Ach T.
        • Huisingh C.
        • McGwin Jr., G.
        • Messinger J.D.
        • Zhang T.
        • Bentley M.J.
        • Gutierrez D.B.
        • Ablonczy Z.
        • Smith R.T.
        • Sloan K.R.
        • Curcio C.A.
        Quantitative autofluorescence and cell density maps of the human retinal pigment epithelium.
        Invest Ophthalmol Vis Sci. 2014; 55: 4832-4841
        • Bhatia S.K.
        • Rashid A.
        • Chrenek M.A.
        • Zhang Q.
        • Bruce B.B.
        • Klein M.
        • Boatright J.H.
        • Jiang Y.
        • Grossniklaus H.E.
        • Nickerson J.M.
        Analysis of RPE morphometry in human eyes.
        Mol Vis. 2016; 22: 898-916
        • Ach T.
        • Tolstik E.
        • Messinger J.D.
        • Zarubina A.V.
        • Heintzmann R.
        • Curcio C.A.
        Lipofuscin redistribution and loss accompanied by cytoskeletal stress in retinal pigment epithelium of eyes with age-related macular degeneration.
        Invest Ophthalmol Vis Sci. 2015; 56: 3242-3252
        • Tarau I.S.
        • Berlin A.
        • Curcio C.A.
        • Ach T.
        The Cytoskeleton of the Retinal Pigment Epithelium: from Normal Aging to Age-Related Macular Degeneration.
        Int J Mol Sci. 2019; 20: 3578
        • Milam A.H.
        • Li Z.Y.
        • Fariss R.N.
        Histopathology of the human retina in retinitis pigmentosa.
        Prog Retin Eye Res. 1998; 17: 175-205
        • Shu X.
        • Beckmann L.
        • Zhang H.
        Visible-light optical coherence tomography: a review.
        J Biomed Opt. 2017; 22: 1-14
        • Calvo-Maroto A.M.
        • Cerviño A.
        Spotlight on fundus autofluorescence.
        Clin Optom (Auckl). 2018; 10: 25-32
        • Roorda A.
        • Zhang Y.
        • Duncan J.L.
        High-resolution in vivo imaging of the RPE mosaic in eyes with retinal disease.
        Invest Ophthalmol Vis Sci. 2007; 48: 2297-2303
        • Gao W.
        • Cense B.
        • Zhang Y.
        • Jonnal R.S.
        • Miller D.T.
        Measuring retinal contributions to the optical Stiles-Crawford effect with optical coherence tomography.
        Opt Express. 2008; 16: 6486-6501
        • Morgan J.I.
        • Dubra A.
        • Wolfe R.
        • Merigan W.H.
        • Williams D.R.
        In vivo autofluorescence imaging of the human and macaque retinal pigment epithelial cell mosaic.
        Invest Ophthalmol Vis Sci. 2009; 50: 1350-1359
        • Scoles D.
        • Sulai Y.N.
        • Dubra A.
        In vivo dark-field imaging of the retinal pigment epithelium cell mosaic.
        Biomed Opt Express. 2013; 4: 1710-1723
        • Liu Z.
        • Kocaoglu O.P.
        • Miller D.T.
        3D Imaging of Retinal Pigment Epithelial Cells in the Living Human Retina.
        Invest Ophthalmol Vis Sci. 2016; 57: OCT533-O543
        • Baraas R.C.
        • Pedersen H.R.
        • Knoblauch K.
        • Gilson S.J.
        Human Foveal Cone and RPE Cell Topographies and Their Correspondence With Foveal Shape.
        Invest. Ophthalmol. Vis. Sci. 2022; 63: 8
        • Liu Z.
        • Kurokawa K.
        • Hammer D.X.
        • Miller D.T.
        In vivo measurement of organelle motility in human retinal pigment epithelial cells.
        Biomed Opt Express. 2019; 10: 4142-4158
      1. Tam J, Liu J, Dubra A, Fariss R. In Vivo Imaging of the Human Retinal Pigment Epithelial Mosaic Using Adaptive Optics Enhanced Indocyanine Green Ophthalmoscopy. Invest Ophthalmol Vis Sci. 2016;57(10):4376-4384.

        • Liu T.
        • Jung H.
        • Liu J.
        • Droettboom M.
        • Tam J.
        Noninvasive near infrared autofluorescence imaging of retinal pigment epithelial cells in the human retina using adaptive optics.
        Biomed Opt Express. 2017; 8: 4348-4360
        • Granger C.E.
        • Yang Q.
        • Song H.
        • Saito K.
        • Nozato K.
        • Latchney L.R.
        • Leonard B.T.
        • Chung M.M.
        • Williams D.R.
        • Rossi E.A.
        Human Retinal Pigment Epithelium: In Vivo Cell Morphometry, Multispectral Autofluorescence, and Relationship to Cone Mosaic.
        Invest Ophthalmol Vis Sci. 2018; 59: 5705-5716
        • Vienola K.V.
        • Zhang M.
        • Snyder V.C.
        • Sahel J.A.
        • Dansingani K.K.
        • Rossi E.A.
        Microstructure of the retinal pigment epithelium near-infrared autofluorescence in healthy young eyes and in patients with AMD.
        Sci Rep. 2020; 10: 9561
        • Bower A.J.
        • Liu T.
        • Aguilera N.
        • Li J.
        • Liu J.
        • Lu R.
        • Giannini J.P.
        • Huryn L.A.
        • Dubra A.
        • Liu T.
        • Hammer D.C.
        • Tam J.
        Integrating adaptive optics-SLO and OCT for multimodal visualization of the human retinal pigment epithelial mosaic.
        Biomed. Opt. Express. 2021; 12: 1449-1466
        • Grieve K.
        • Gofas-Salas E.
        • Ferguson R.D.
        • Sahel J.A.
        • Paques M.
        • Rossi E.A.
        In vivo near-infrared autofluorescence imaging of retinal pigment epithelial cells with 757 nm excitation.
        Biomed Opt Express. 2018; 9: 5946-5961
        • Masella B.D.
        • Williams D.R.
        • Fischer W.S.
        • Rossi E.A.
        • Hunter J.J.
        Long-term reduction in infrared autofluorescence caused by infrared light below the maximum permissible exposure.
        Invest Ophthalmol Vis Sci. 2014; 55: 3929-3938
        • Wynne N.
        • Carroll J.
        • Duncan J.L.
        Promises and pitfalls of evaluating photoreceptor-based retinal disease with adaptive optics scanning light ophthalmoscopy (AOSLO).
        Prog Retin Eye Res. 2020; 100920
        • Laforest T.
        • Künzi M.
        • Kowalczuk L.
        • Carpentras D.
        • Behar-Cohen F.
        • Moser C.
        Transscleral Optical Phase Imaging of the Human Retina.
        Nat Photonics. 2020; 14: 439-445
        • Hammer M.
        • Roggan A.
        • Schweitzer D.
        • Müller G.
        Optical properties of ocular fundus tissues--an in vitro study using the double-integrating-sphere technique and inverse Monte Carlo simulation.
        Phys Med Biol. 1995; 40: 963-978
        • Latychevskaia T.
        Lateral and axial resolution criteria in incoherent and coherent optics and holography, near- and far-field regimes.
        Appl Opt. 2019; 58: 3597-3603
        • Caetano Dos Santos F.L.
        • Laforest T.
        • Künzi M.
        • Kowalczuk L.
        • Behar-Cohen F.
        • Moser C.
        Fully automated detection, segmentation, and analysis of in vivo RPE single cells.
        Eye (Lond). 2021; 35: 1473-1481
        • Tseng Q.
        • Duchemin-Pelletier E.
        • Deshiere A.
        • Balland M.
        • Guillou H.
        • Filhol O.
        • Théry M.
        Spatial organization of the extracellular matrix regulates cell-cell junction positioning.
        Proc Natl Acad Sci U S A. 2012; 109: 1506-1511
        • Thévenaz P.
        • Ruttimann U.E.
        • Unser M.
        A pyramid approach to subpixel registration based on intensity.
        IEEE Trans Image Process. 1998; 7: 27-41
        • Pollreisz A.
        • Messinger J.D.
        • Sloan K.R.
        • Mittermueller T.
        • Weinhandl A.S.
        • Benson E.
        • Kidd G.J.
        • Schmidt-Erfurth U.
        • Curcio C.A.
        Visualizing melanosomes, lipofuscin, and melanolipofuscin in human retinal pigment epithelium using serial block face scanning electron microscopy.
        Exp Eye Res. 2018; 166: 131-139
        • Pollreisz A.
        • Neschi M.
        • Sloan K.R.
        • Pircher M.
        • Mittermueller T.J.
        • Dacey D.M.
        • Schmidt-Erfurth U.
        • Curcio C.A.
        An atlas of human retinal pigment epithelium organelles significant for clinical imaging.
        Invest Ophthalmol Vis Sci. 2020; 61: 13
        • Bartol-Puyal F.A.
        • Isanta C.
        • Calvo P.
        • Ruiz-Moreno Ó.
        • Pablo L.
        Relationship between vascular densities of choriocapillaris and the whole choroid using OCTA.
        Int Ophthalmol. 2020; 40: 3135-3143
        • Weiter J.J.
        • Delori F.C.
        • Wing G.L.
        • Fitch K.A.
        Retinal pigment epithelial lipofuscin and melanin and choroidal melanin in human eyes.
        Invest Ophthalmol Vis Sci. 1986; 27: 145-152
        • Ortolan D.
        • Sharma R.
        • Volkov A.
        • Maminishkis A.
        • Hotaling N.A.
        • Huryn L.A.
        • Cukras C.
        • Di Marco S.
        • Bisti S.
        • Bharti K.
        Single-cell-resolution map of human retinal pigment epithelium helps discover subpopulations with differential disease sensitivity.
        Proc Natl Acad Sci U S A. 2022; 119e2117553119
        • Bartlett J.W.
        • Frost C.
        Reliability, repeatability and reproducibility: analysis of measurement errors in continuous variables.
        Ultrasound Obstet Gynecol. 2008; 31: 466-475
        • Jonas J.B.
        • Ohno-Matsui K.
        • Jiang W.J.
        • Panda-Jonas S.
        BRUCH MEMBRANE AND THE MECHANISM OF MYOPIZATION: A New Theory.
        Retina. 2017; 37: 1428-1440
        • Nickla D.L.
        • Wallman J.
        The multifunctional choroid.
        Prog Retin Eye Res. 2010; 29: 144-168
        • Flügel-Koch C.
        • May C.A.
        • Lütjen-Drecoll E.
        Presence of a contractile cell network in the human choroid.
        Ophthalmologica. 1996; 210: 296-302
        • Del Priore L.V.
        • Kuo Y.H.
        • Tezel T.H.
        Age-Related Changes in Human RPE Cell Density and Apoptosis Proportion In Situ.
        Invest. Ophthalmol. Vis. Sci. 2002; 43: 3312-3318
        • Ts'o M.O.M.
        • Friedman E.
        The retinal pigment epithelium: III. Growth and development.
        Archives of Ophthalmology. 1968; 80 (PMID 5661888): 214-216
        • Putnam N.M.
        • Hofer H.J.
        • Doble N.
        • Chen L.
        • Carroll J.
        • Williams D.R.
        The locus of fixation and the foveal cone mosaic.
        J Vis. 2005; 5: 632-639
      2. Roshandel D, Sampson DM, Mackey DA, Chen FK. Impact of Reference Center Choice on Adaptive Optics Imaging Cone Mosaic Analysis. Invest Ophthalmol Vis Sci. 2022;63(4):12.

      3. Kolb H. Facts and Figures Concerning the Human Retina. 2005 May 1 [updated 2007 Jul 5]. In: Kolb H, Fernandez E, Nelson R, editors. Webvision: The Organization of the Retina and Visual System [Internet]. Salt Lake City (UT): University of Utah Health Sciences Center; 1995–. PMID: 21413409.