My interest in ophthalmology started at a very early age, motivated by my own amblyopia and hyperopia. These led me to study physics and optics, with my first research experience at undergraduate and master’s level at the Applied Optics Group of the Universidad de la República in Uruguay. I then pursued my PhD work and a first postdoctoral position at the Photonics Group in Imperial College London, where I worked on instrumentation to study the topography of the tear film and adaptive optics (AO). The desire to advance AO for retinal imaging took me to the University of Rochester, where the interaction with patients affected by blinding conditions provided me with the determination to go beyond the proof-of-principle experiments after which many technologies are abandoned. Therefore, since starting my research group at the University of Rochester first, and at the Medical College of Wisconsin later and now at Stanford we have focused on the development and translation of AO and microscopy techniques into tools that can be used to address real clinical problems.

Academic Appointments

Administrative Appointments

  • Honorary Lecturer of Ophthalmology, University College London (UK) (2013 - Present)
  • Associate Adjunct Professor of Ophthalmology, Medical College of Wisconsin (2016 - Present)

Honors & Awards

  • Overseas Research Student Award, Universities UK (Ex. Committee of Vice-Chancellors & Principals) (2000)
  • Career Award at the Scientific Interface, Burroughs Wellcome Fund (2008)
  • Career Development Award, Research to Prevent Blindness (2012)

Boards, Advisory Committees, Professional Organizations

  • Member, The International Society for Optics and Photonics (2011 - Present)
  • Member, Association for Research in Vision and Ophthalmology (2001 - Present)
  • Member, Optical Society of America (2008 - Present)

Professional Education

  • PhD, Imperial College London, UK, Physics (2004)
  • MSc, Universidad de la República, Uruguay, Physics (2000)
  • BSc, Universidad de la República, Uruguay, Physics (1998)


  • Alfredo Dubra. "United States Patent 8,226,236 Method and apparatus for imaging in an eye", University Of Rochester, May 18, 2006

Research & Scholarship

Current Research and Scholarly Interests

The optics of the eye can be thought of as an imperfect microscope objective through which the retina can be observed. Our lab uses adaptive optics, a technology originally developed to observe distant stars and galaxies to improve this microscope objective so that individual retinal and blood cells can be visualized. Moreover, we use and invent new microscopy imaging methods to reveal cellular and sub-cellular structures in the eye through a multidisciplinary approach that integrates optics, computer science, vision science, electrical engineering and other engineering disciplines.


2017-18 Courses

Stanford Advisees


All Publications

  • Deep learning based detection of cone photoreceptors with multimodal adaptive optics scanning light ophthalmoscope images of achromatopsia BIOMEDICAL OPTICS EXPRESS Cunefare, D., Langlo, C. S., Patterson, E. J., Blau, S., Dubra, A., Carroll, J., Farsiu, S. 2018; 9 (8)
  • Automatic Cone Photoreceptor Localisation in Healthy and Stargardt Afflicted Retinas Using Deep Learning SCIENTIFIC REPORTS Davidson, B., Kalitzeos, A., Carroll, J., Dubra, A., Ourselin, S., Michaelides, M., Bergeles, C. 2018; 8: 7911


    We present a robust deep learning framework for the automatic localisation of cone photoreceptor cells in Adaptive Optics Scanning Light Ophthalmoscope (AOSLO) split-detection images. Monitoring cone photoreceptors with AOSLO imaging grants an excellent view into retinal structure and health, provides new perspectives into well known pathologies, and allows clinicians to monitor the effectiveness of experimental treatments. The MultiDimensional Recurrent Neural Network (MDRNN) approach developed in this paper is the first method capable of reliably and automatically identifying cones in both healthy retinas and retinas afflicted with Stargardt disease. Therefore, it represents a leap forward in the computational image processing of AOSLO images, and can provide clinical support in on-going longitudinal studies of disease progression and therapy. We validate our method using images from healthy subjects and subjects with the inherited retinal pathology Stargardt disease, which significantly alters image quality and cone density. We conduct a thorough comparison of our method with current state-of-the-art methods, and demonstrate that the proposed approach is both more accurate and appreciably faster in localizing cones. As further validation to the method's robustness, we demonstrate it can be successfully applied to images of retinas with pathologies not present in the training data: achromatopsia, and retinitis pigmentosa.

    View details for DOI 10.1038/s41598-018-26350-3

    View details for Web of Science ID 000432531400018

    View details for PubMedID 29784939

    View details for PubMedCentralID PMC5962538

  • The Reliability of Cone Density Measurements in the Presence of Rods TRANSLATIONAL VISION SCIENCE & TECHNOLOGY Morgan, J. W., Vergilio, G. K., Hsu, J., Dubra, A., Cooper, R. F. 2018; 7 (3): 21


    Recent advances in adaptive optics scanning light ophthalmoscopy (AOSLO) have enabled visualization of cone inner segments through nonconfocal split-detection, in addition to rod and cone outer segments revealed by confocal reflectance. Here, we examined the interobserver reliability of cone density measurements in both AOSLO imaging modalities.Five normal subjects (nine eyes) were imaged along the horizontal and vertical meridians using a custom AOSLO with confocal and nonconfocal split-detection modalities. The resulting images were montaged using a previously described semiautomatic algorithm. Regions of interest (ROIs) were selected from the confocal montage at 190 μm, and from split-detection and confocal montages at 900 and 1800 μm from the fovea. Four observers (three experts, one naïve) manually identified cone locations in each ROI, and these locations were used to calculate bound densities. Intraclass correlation coefficients and Dice's coefficients were calculated to assess interobserver agreement.Interobserver agreement was high in cone-only images (confocal 190 μm: 0.85; split-detection 900 μm: 0.91; split-detection 1800 μm: 0.89), moderate in confocal images at 900 μm (0.68), and poor in confocal images at 1800 μm (0.24). Excluding the naïve observer data substantially increased agreement within confocal images (190 μm: 0.99; 900 μm: 0.80; 1800 μm: 0.68).Interobserver measurements of cone density are more reliable in rod-free retinal images. Moreover, when using manual cell identification, it is essential that observers are trained, particularly for confocal AOSLO images.This study underscores the need for additional reliability studies in eyes containing pathology where identifying cones can be substantially more difficult.

    View details for DOI 10.1167/tvst.7.3.21

    View details for Web of Science ID 000436426300004

    View details for PubMedID 29946495

    View details for PubMedCentralID PMC6016505

  • Non-invasive assessment of human cone photoreceptor function (vol 8, pg 5098, 2017) BIOMEDICAL OPTICS EXPRESS Cooper, R. F., Tuten, W. S., Dubra, A., Brainard, D. H., Morgan, J. W. 2018; 9 (4): 1842


    [This corrects the article on p. 5098 in vol. 8, PMID: 29188106.].

    View details for DOI 10.1364/BOE.9.001842

    View details for Web of Science ID 000428954600033

    View details for PubMedID 29676401

    View details for PubMedCentralID PMC5905927

  • Ultrahigh resolution retinal imaging by visible light OCT with longitudinal achromatization BIOMEDICAL OPTICS EXPRESS Chong, S., Zhang, T., Kho, A., Bernucci, M. T., Dubra, A., Srinivasan, V. J. 2018; 9 (4): 1477–91


    Chromatic aberrations are an important design consideration in high resolution, high bandwidth, refractive imaging systems that use visible light. Here, we present a fiber-based spectral/Fourier domain, visible light OCT ophthalmoscope corrected for the average longitudinal chromatic aberration (LCA) of the human eye. Analysis of complex speckles from in vivo retinal images showed that achromatization resulted in a speckle autocorrelation function that was ~20% narrower in the axial direction, but unchanged in the transverse direction. In images from the improved, achromatized system, the separation between Bruch's membrane (BM), the retinal pigment epithelium (RPE), and the outer segment tips clearly emerged across the entire 6.5 mm field-of-view, enabling segmentation and morphometry of BM and the RPE in a human subject. Finally, cross-sectional images depicted distinct inner retinal layers with high resolution. Thus, with chromatic aberration compensation, visible light OCT can achieve volume resolutions and retinal image quality that matches or exceeds ultrahigh resolution near-infrared OCT systems with no monochromatic aberration compensation.

    View details for DOI 10.1364/BOE.9.001477

    View details for Web of Science ID 000428954600006

    View details for PubMedID 29675296

    View details for PubMedCentralID PMC5905900

  • Sub-Airy Confocal Adaptive Optics Scanning Ophthalmoscopy TRANSLATIONAL VISION SCIENCE & TECHNOLOGY Sredar, N., Fagbemi, O. E., Dubra, A. 2018; 7 (2): 17


    To demonstrate the viability of improving transverse image resolution in reflectance scanning adaptive optics ophthalmoscopy using sub-Airy disk confocal detection.The foveal cone mosaic was imaged in five human subjects free of known eye disease using two custom adaptive optics scanning light ophthalmoscopes (AOSLOs) in reflectance with 7.75 and 4.30 mm pupil diameters. Confocal pinholes of 0.5, 0.6, 0.8, and 1.0 Airy disk diameters (ADDs) were used in a retinal conjugate plane before the light detector. Average cone photoreceptor intensity profile width and power spectrum were calculated for the resulting images. Detected energy using a model eye was recorded for each pinhole size.The cone photoreceptor mosaic is better resolved with decreasing confocal pinhole size, with the high spatial frequency content of the images enhanced in both the large- and small-pupil AOSLOs. The average cone intensity profile width was reduced by ∼15% with the use of a 0.5 ADD pinhole when compared to a 1.0 ADD, with an accompanying reduction in signal greater than a factor of four.The use of sub-Airy disk confocal pinhole detection without increasing retinal light exposure results in a substantial improvement in image resolution at the cost of larger than predicted signal reduction.Improvement in transverse resolution using sub-Airy disk confocal detection is a practical and low-cost approach that is applicable to all point- and line-scanning ophthalmoscopes, including optical coherence tomographers.

    View details for DOI 10.1167/tvst.7.2.17

    View details for Web of Science ID 000429537100001

    View details for PubMedID 29629239

    View details for PubMedCentralID PMC5886094

  • Phenotypic diversity in autosomal-dominant cone-rod dystrophy elucidated by adaptive optics retinal imaging. The British journal of ophthalmology Song, H., Rossi, E. A., Stone, E., Latchney, L., Williams, D., Dubra, A., Chung, M. 2018; 102 (1): 136–41


    Several genes causing autosomal-dominant cone-rod dystrophy (AD-CRD) have been identified. However, the mechanisms by which genetic mutations lead to cellular loss in human disease remain poorly understood. Here we combine genotyping with high-resolution adaptive optics retinal imaging to elucidate the retinal phenotype at a cellular level in patients with AD-CRD harbouring a defect in theGUCA1Agene.Nine affected members of a four-generation AD-CRD pedigree and three unaffected first-degree relatives underwent clinical examinations including visual acuity, fundus examination, Goldmann perimetry, spectral domain optical coherence tomography and electroretinography. Genome-wide scan followed by bidirectional sequencing was performed on all affected participants. High-resolution imaging using a custom adaptive optics scanning light ophthalmoscope (AOSLO) was performed for selected participants.Clinical evaluations showed a range of disease severity from normal fundus appearance in teenaged patients to pronounced macular atrophy in older patients. Molecular genetic testing showed a mutation in inGUCA1Asegregating with disease. AOSLO imaging revealed that of the two teenage patients with mild disease, one had severe disruption of the photoreceptor mosaic while the other had a normal cone mosaic.AOSLO imaging demonstrated variability in the pattern of cone and rod cell loss between two teenage cousins with early AD-CRD, who had similar clinical features and had the identical disease-causing mutation inGUCA1A. This finding suggests that a mutation inGUCA1Adoes not lead to the same degree of AD-CRD in all patients. Modifying factors may mitigate or augment disease severity, leading to different retinal cellular phenotypes.

    View details for DOI 10.1136/bjophthalmol-2017-310498

    View details for PubMedID 29074494

    View details for PubMedCentralID PMC5754866

  • New Clinical Opportunities for Retinal Vascular Imaging Adaptive Optics to OCT Angiography Rosen, R., Chui, T., Weitz, R., Dubra, A., Carroll, J., Garcia, P., Pinhas, A., Scripsema, N., Mo, S., Agemy, S., Krawitz, B., Podoleanu, A., Bang, O. SPIE-INT SOC OPTICAL ENGINEERING. 2018

    View details for DOI 10.1117/12.2304296

    View details for Web of Science ID 000432466900001

  • Multimodal imaging of small hard retinal drusen in young healthy adults. The British journal of ophthalmology Pedersen, H. R., Gilson, S. J., Dubra, A., Munch, I. C., Larsen, M., Baraas, R. C. 2018; 102 (1): 146–52


    Small hard macular drusen can be observed in the retina of adults as young as 18 years of age. Here, we seek to describe the in vivo topography and geometry of these drusen.Retinal images were acquired in young, healthy adults using colour fundus photography, spectral domain optic coherence tomography (SD-OCT), reflectance flood-illuminated adaptive optic ophthalmoscopy (AO flood) and reflectance adaptive optic scanning light ophthalmoscopy (AOSLO) in both confocal and non-confocal split-detection modalities. Small bright yellow hard drusen within a 10 degree radius from the foveal centre were characterised.Small hard drusen were seen on colour photographs in 21 out of 97 participants and 26 drusen in 12 eyes in 11 participants were imaged using the full protocol. Drusen were easily identifiable in all modalities, except a few very small ones, which were not visible on SD-OCT. On AOSLO images, these drusen appeared as round, oval or lobular areas (up to three lobules) of diameter 22-61 µm where cone photoreceptor reflectivity and density was decreased (p=0.049). This was usually associated with discrete thickening of the retinal pigment epithelium (RPE) complex.High lateral resolution imaging of small lobular hard retinal drusen suggests formation through the confluence of two or more smaller round lesions. The outline and size of these smaller lesions corresponds to 1-4 RPE cells. Prospective longitudinal studies are needed to determine the ultimate fate of small hard drusen and their potential relation to age-related macular degeneration.

    View details for DOI 10.1136/bjophthalmol-2017-310719

    View details for PubMedID 29051326

    View details for PubMedCentralID PMC5754867

  • Visible light optical coherence microscopy of the brain with isotropic femtoliter resolution in vivo. Optics letters Merkle, C. W., Chong, S. P., Kho, A. M., Zhu, J., Dubra, A., Srinivasan, V. J. 2018; 43 (2): 198–201


    Most flying-spot optical coherence tomography and optical coherence microscopy (OCM) systems use a symmetric confocal geometry, where the detection path retraces the illumination path starting from and ending with the spatial mode of a single-mode optical fiber. Here we describe a visible light OCM instrument that breaks this symmetry to improve transverse resolution without sacrificing collection efficiency in scattering tissue. This was achieved by overfilling a water immersion objective on the illumination path while maintaining a conventional Gaussian mode detection path (1/e2intensity diameter ∼0.82 Airy disks), enabling ∼1.1  μm full width at half-maximum (FWHM) transverse resolution. At the same time, a ∼0.9  μm FWHM axial resolution in tissue, achieved by a broadband visible light source, enabled femtoliter volume resolution. We characterized this instrument according to paraxial coherent microscopy theory and, finally, used it to image the meningeal layers, intravascular red blood cell-free layer, and myelinated axons in the mouse neocortex in vivo through the thinned skull.

    View details for DOI 10.1364/OL.43.000198

    View details for PubMedID 29328237

  • An Automated Reference Frame Selection (ARFS) Algorithm for Cone Imaging with Adaptive Optics Scanning Light Ophthalmoscopy. Translational vision science & technology Salmon, A. E., Cooper, R. F., Langlo, C. S., Baghaie, A., Dubra, A., Carroll, J. 2017; 6 (2): 9-?


    To develop an automated reference frame selection (ARFS) algorithm to replace the subjective approach of manually selecting reference frames for processing adaptive optics scanning light ophthalmoscope (AOSLO) videos of cone photoreceptors.Relative distortion was measured within individual frames before conducting image-based motion tracking and sorting of frames into distinct spatial clusters. AOSLO images from nine healthy subjects were processed using ARFS and human-derived reference frames, then aligned to undistorted AO-flood images by nonlinear registration and the registration transformations were compared. The frequency at which humans selected reference frames that were rejected by ARFS was calculated in 35 datasets from healthy subjects, and subjects with achromatopsia, albinism, or retinitis pigmentosa. The level of distortion in this set of human-derived reference frames was assessed.The average transformation vector magnitude required for registration of AOSLO images to AO-flood images was significantly reduced from 3.33 ± 1.61 pixels when using manual reference frame selection to 2.75 ± 1.60 pixels (mean ± SD) when using ARFS (P = 0.0016). Between 5.16% and 39.22% of human-derived frames were rejected by ARFS. Only 2.71% to 7.73% of human-derived frames were ranked in the top 5% of least distorted frames.ARFS outperforms expert observers in selecting minimally distorted reference frames in AOSLO image sequences. The low success rate in human frame choice illustrates the difficulty in subjectively assessing image distortion.Manual reference frame selection represented a significant barrier to a fully automated image-processing pipeline (including montaging, cone identification, and metric extraction). The approach presented here will aid in the clinical translation of AOSLO imaging.

    View details for DOI 10.1167/tvst.6.2.9

    View details for PubMedID 28392976

  • PHOTORECEPTOR INNER SEGMENT MORPHOLOGY IN BEST VITELLIFORM MACULAR DYSTROPHY. Retina (Philadelphia, Pa.) Scoles, D., Sulai, Y. N., Cooper, R. F., Higgins, B. P., Johnson, R. D., Carroll, J., Dubra, A., Stepien, K. E. 2017; 37 (4): 741-748


    To characterize outer retina structure in best vitelliform macular dystrophy (BVMD) and to determine the effect of macular lesions on overlying and adjacent photoreceptors.Five individuals with BVMD were followed prospectively with spectral domain optical coherence tomography and confocal and nonconfocal split-detector adaptive optics scanning light ophthalmoscopy (AOSLO). The AOSLO cone photoreceptor mosaic images were obtained within and around retinal lesions. Cone density was measured inside and outside lesions. In 2 subjects, densities were compared with published measurements acquired ∼2.5 years before. One subject was imaged 3 times over a 5-month period.The AOSLO imaging demonstrated that photoreceptor morphology within BVMD retinal lesions was highly variable depending on the disease stage, with photoreceptor structure present even in advanced disease. The AOSLO imaging was repeatable even in severe disease over short-time and long-time intervals. Photoreceptor density was normal in retinal areas immediately adjacent to lesions and stable over ∼2.5 years. Mobile disk-like structures possibly representing subretinal macrophages were also observed.Combined confocal and nonconfocal split-detector AOSLO imaging reveals substantial variability within clinical lesions in all stages of BVMD. Longitudinal cellular photoreceptor imaging could prove a powerful tool for understanding disease progression and monitoring emerging therapeutic treatment response in inherited degenerations such as BVMD.

    View details for DOI 10.1097/IAE.0000000000001203

    View details for PubMedID 27467379

    View details for PubMedCentralID PMC5362286

  • Assessing the spatial relationship between fixation and foveal specializations. Vision research Wilk, M. A., Dubis, A. M., Cooper, R. F., Summerfelt, P., Dubra, A., Carroll, J. 2017; 132: 53-61


    Increased cone photoreceptor density, an avascular zone (FAZ), and the displacement of inner retinal neurons to form a pit are distinct features of the human fovea. As the fovea provides the majority of our vision, appreciating how these anatomical specializations are related is important for understanding foveal development, normal visual function, and retinal disease. Here we evaluated the relationship between these specializations and their location relative to the preferred retinal locus of fixation (PRL). We measured foveal pit volume, FAZ area, peak cone density, and location of the PRL in 22 subjects with normal vision using optical coherence tomography and adaptive optics scanning light ophthalmoscopy. Foveal pit volume was positively correlated with FAZ area; however, peak cone density was not correlated with pit volume. In addition, there was no systematic offset of the location of any of these specializations relative to PRL, and there was no correlation between the magnitude of the offset from PRL and the corresponding foveal specialization measurements (pit volume, FAZ area, peak cone density). The standard deviation of our PRL measurements was consistent with previous measurements of fixational stability. These data provide insight into the sequence of events during foveal development and may have implications for visual function and retinal disease.

    View details for DOI 10.1016/j.visres.2016.05.001

    View details for PubMedID 27286921

    View details for PubMedCentralID PMC5164985

  • Vision science and adaptive optics, the state of the field. Vision research Marcos, S., Werner, J. S., Burns, S. A., Merigan, W. H., Artal, P., Atchison, D. A., Hampson, K. M., Legras, R., Lundstrom, L., Yoon, G., Carroll, J., Choi, S. S., Doble, N., Dubis, A. M., Dubra, A., Elsner, A., Jonnal, R., Miller, D. T., Paques, M., Smithson, H. E., Young, L. K., Zhang, Y., Campbell, M., Hunter, J., Metha, A., Palczewska, G., Schallek, J., Sincich, L. C. 2017


    Adaptive optics is a relatively new field, yet it is spreading rapidly and allows new questions to be asked about how the visual system is organized. The editors of this feature issue have posed a series of question to scientists involved in using adaptive optics in vision science. The questions are focused on three main areas. In the first we investigate the use of adaptive optics for psychophysical measurements of visual system function and for improving the optics of the eye. In the second, we look at the applications and impact of adaptive optics on retinal imaging and its promise for basic and applied research. In the third, we explore how adaptive optics is being used to improve our understanding of the neurophysiology of the visual system.

    View details for DOI 10.1016/j.visres.2017.01.006

    View details for PubMedID 28212982

  • REPEATABILITY AND LONGITUDINAL ASSESSMENT OF FOVEAL CONE STRUCTURE IN CNGB3-ASSOCIATED ACHROMATOPSIA. Retina (Philadelphia, Pa.) Langlo, C. S., Erker, L. R., Parker, M., Patterson, E. J., Higgins, B. P., Summerfelt, P., Razeen, M. M., Collison, F. T., Fishman, G. A., Kay, C. N., Zhang, J., Weleber, R. G., Yang, P., Pennesi, M. E., Lam, B. L., Chulay, J. D., Dubra, A., Hauswirth, W. W., Wilson, D. J., Carroll, J. 2017


    Congenital achromatopsia is an autosomal recessive disease causing substantial reduction or complete absence of cone function. Although believed to be a relatively stationary disorder, questions remain regarding the stability of cone structure over time. In this study, the authors sought to assess the repeatability of and examine longitudinal changes in measurements of central cone structure in patients with achromatopsia.Forty-one subjects with CNGB3-associated achromatopsia were imaged over a period of between 6 and 26 months using optical coherence tomography and adaptive optics scanning light ophthalmoscopy. Outer nuclear layer (ONL) thickness, ellipsoid zone (EZ) disruption, and peak foveal cone density were assessed.ONL thickness increased slightly compared with baseline (0.184 μm/month, P = 0.02). The EZ grade remained unchanged for 34/41 subjects. Peak foveal cone density did not significantly change over time (mean change 1% per 6 months, P = 0.126).Foveal cone structure showed little or no change in this group of subjects with CNGB3-associated achromatopsia. Over the time scales investigated (6-26 months), achromatopsia seems to be a structurally stable condition, although longer-term follow-up is needed. These data will be useful in assessing foveal cone structure after therapeutic intervention.

    View details for DOI 10.1097/IAE.0000000000001434

    View details for PubMedID 28145975

    View details for PubMedCentralID PMC5537050

  • Reliability and Repeatability of Cone Density Measurements in Patients With Stargardt Disease and RPGR-Associated Retinopathy. Investigative ophthalmology & visual science Tanna, P., Kasilian, M., Strauss, R., Tee, J., Kalitzeos, A., Tarima, S., Visotcky, A., Dubra, A., Carroll, J., Michaelides, M. 2017; 58 (9): 3608–15


    To assess reliability and repeatability of cone density measurements by using confocal and (nonconfocal) split-detector adaptive optics scanning light ophthalmoscopy (AOSLO) imaging. It will be determined whether cone density values are significantly different between modalities in Stargardt disease (STGD) and retinitis pigmentosa GTPase regulator (RPGR)-associated retinopathy.Twelve patients with STGD (aged 9-52 years) and eight with RPGR-associated retinopathy (aged 11-31 years) were imaged using both confocal and split-detector AOSLO simultaneously. Four graders manually identified cone locations in each image that were used to calculate local densities. Each imaging modality was evaluated independently. The data set consisted of 1584 assessments of 99 STGD images (each image in two modalities and four graders who graded each image twice) and 928 RPGR assessments of 58 images (each image in two modalities and four graders who graded each image twice).For STGD assessments the reliability for confocal and split-detector AOSLO was 67.9% and 95.9%, respectively, and the repeatability was 71.2% and 97.3%, respectively. The differences in the measured cone density values between modalities were statistically significant for one grader. For RPGR assessments the reliability for confocal and split-detector AOSLO was 22.1% and 88.5%, respectively, and repeatability was 63.2% and 94.5%, respectively. The differences in cone density between modalities were statistically significant for all graders.Split-detector AOSLO greatly improved the reliability and repeatability of cone density measurements in both disorders and will be valuable for natural history studies and clinical trials using AOSLO. However, it appears that these indices may be disease dependent, implying the need for similar investigations in other conditions.

    View details for DOI 10.1167/iovs.17-21904

    View details for PubMedID 28738413

    View details for PubMedCentralID PMC5525557

  • Open source software for automatic detection of cone photoreceptors in adaptive optics ophthalmoscopy using convolutional neural networks. Scientific reports Cunefare, D., Fang, L., Cooper, R. F., Dubra, A., Carroll, J., Farsiu, S. 2017; 7 (1): 6620


    Imaging with an adaptive optics scanning light ophthalmoscope (AOSLO) enables direct visualization of the cone photoreceptor mosaic in the living human retina. Quantitative analysis of AOSLO images typically requires manual grading, which is time consuming, and subjective; thus, automated algorithms are highly desirable. Previously developed automated methods are often reliant on ad hoc rules that may not be transferable between different imaging modalities or retinal locations. In this work, we present a convolutional neural network (CNN) based method for cone detection that learns features of interest directly from training data. This cone-identifying algorithm was trained and validated on separate data sets of confocal and split detector AOSLO images with results showing performance that closely mimics the gold standard manual process. Further, without any need for algorithmic modifications for a specific AOSLO imaging system, our fully-automated multi-modality CNN-based cone detection method resulted in comparable results to previous automatic cone segmentation methods which utilized ad hoc rules for different applications. We have made free open-source software for the proposed method and the corresponding training and testing datasets available online.

    View details for DOI 10.1038/s41598-017-07103-0

    View details for PubMedID 28747737

    View details for PubMedCentralID PMC5529414

  • Non-invasive assessment of human cone photoreceptor function. Biomedical optics express Cooper, R. F., Tuten, W. S., Dubra, A., Brainard, D. H., Morgan, J. I. 2017; 8 (11): 5098–5112


    Vision begins when light isomerizes the photopigments within photoreceptors. Noninvasive cellular-scale observation of the structure of the human photoreceptor mosaic is made possible through the use of adaptive optics (AO) enhanced ophthalmoscopes, but establishing noninvasive objective measures of photoreceptor function on a similar scale has been more difficult. AO ophthalmoscope images acquired with near-infrared light show that individual cone photoreceptor reflectance can change in response to a visible stimulus. Here we show that the intrinsic response depends on stimulus wavelength and intensity, and that its action spectrum is well-matched to the spectral sensitivity of cone-mediated vision. Our results demonstrate that the cone reflectance response is mediated by photoisomerization, thus making it a direct measure of photoreceptor function.

    View details for DOI 10.1364/BOE.8.005098

    View details for PubMedID 29188106

    View details for PubMedCentralID PMC5695956

  • CELLULAR IMAGING OF THE TAPETAL-LIKE REFLEX IN CARRIERS OF RPGR-ASSOCIATED RETINOPATHY. Retina (Philadelphia, Pa.) Kalitzeos, A., Samra, R., Kasilian, M., Tee, J. J., Strampe, M., Langlo, C., Webster, A. R., Dubra, A., Carroll, J., Michaelides, M. 2017


    To examine the features of the tapetal-like reflex (TLR) in female carriers of RPGR-associated retinopathy by means of adaptive optics scanning light ophthalmoscopy (AOSLO) and spectral domain optical coherence tomography.Nine molecularly confirmed RPGR carriers and three healthy controls underwent ocular examination and the following retinal imaging modalities: color photography, near-infrared reflectance, fundus autofluorescence, spectral domain optical coherence tomography, and AOSLO. After identifying TLR areas across all imaging modalities, normalized local contrast of outer retinal bands on spectral domain optical coherence tomography was calculated and AOSLO-acquired photoreceptor mosaic analysis was performed.Seven carriers had TLR areas, which colocalized with increased rod photoreceptor reflectivity on confocal AOSLO and reduced cone photoreceptor densities. Parafoveal TLR areas also exhibited reduced local contrast (i.e., increased reflectivity) of the outer retinal bands on spectral domain optical coherence tomography (inner segment ellipsoid zone and outer segment interdigitation zone). Healthy controls did not show TLR.The cellular resolution provided by AOSLO affords the characterization of the photoreceptor mosaic in RPGR carriers with a TLR. Features revealed include reduced cone densities, increased cone inner segment diameters, and increased rod outer segment reflectivity.This is an open access article distributed under the Creative Commons Attribution License 4.0 (CCBY), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

    View details for DOI 10.1097/IAE.0000000000001965

    View details for PubMedID 29190250

  • Automated Photoreceptor Cell Identification on Nonconfocal Adaptive Optics Images Using Multiscale Circular Voting. Investigative ophthalmology & visual science Liu, J., Jung, H., Dubra, A., Tam, J. 2017; 58 (11): 4477–89


    Adaptive optics scanning light ophthalmoscopy (AOSLO) has enabled quantification of the photoreceptor mosaic in the living human eye using metrics such as cell density and average spacing. These rely on the identification of individual cells. Here, we demonstrate a novel approach for computer-aided identification of cone photoreceptors on nonconfocal split detection AOSLO images.Algorithms for identification of cone photoreceptors were developed, based on multiscale circular voting (MSCV) in combination with a priori knowledge that split detection images resemble Nomarski differential interference contrast images, in which dark and bright regions are present on the two sides of each cell. The proposed algorithm locates dark and bright region pairs, iteratively refining the identification across multiple scales. Identification accuracy was assessed in data from 10 subjects by comparing automated identifications with manual labeling, followed by computation of density and spacing metrics for comparison to histology and published data.There was good agreement between manual and automated cone identifications with overall recall, precision, and F1 score of 92.9%, 90.8%, and 91.8%, respectively. On average, computed density and spacing values using automated identification were within 10.7% and 11.2% of the expected histology values across eccentricities ranging from 0.5 to 6.2 mm. There was no statistically significant difference between MSCV-based and histology-based density measurements (P = 0.96, Kolmogorov-Smirnov 2-sample test).MSCV can accurately detect cone photoreceptors on split detection images across a range of eccentricities, enabling quick, objective estimation of photoreceptor mosaic metrics, which will be important for future clinical trials utilizing adaptive optics.

    View details for DOI 10.1167/iovs.16-21003

    View details for PubMedID 28873173

    View details for PubMedCentralID PMC5586244

  • Unsupervised identification of cone photoreceptors in non-confocal adaptive optics scanning light ophthalmoscope images. Biomedical optics express Bergeles, C., Dubis, A. M., Davidson, B., Kasilian, M., Kalitzeos, A., Carroll, J., Dubra, A., Michaelides, M., Ourselin, S. 2017; 8 (6): 3081–94


    Precise measurements of photoreceptor numerosity and spatial arrangement are promising biomarkers for the early detection of retinal pathologies and may be valuable in the evaluation of retinal therapies. Adaptive optics scanning light ophthalmoscopy (AOSLO) is a method of imaging that corrects for aberrations of the eye to acquire high-resolution images that reveal the photoreceptor mosaic. These images are typically graded manually by experienced observers, obviating the robust, large-scale use of the technology. This paper addresses unsupervised automated detection of cones in non-confocal, split-detection AOSLO images. Our algorithm leverages the appearance of split-detection images to create a cone model that is used for classification. Results show that it compares favorably to the state-of-the-art, both for images of healthy retinas and for images from patients affected by Stargardt disease. The algorithm presented also compares well to manual annotation while excelling in speed.

    View details for DOI 10.1364/BOE.8.003081

    View details for PubMedID 28663928

    View details for PubMedCentralID PMC5480451

  • Progression of Local Glaucomatous Damage Near Fixation as Seen with Adaptive Optics Imaging. Translational vision science & technology Hood, D. C., Lee, D., Jarukasetphon, R., Nunez, J., Mavrommatis, M. A., Rosen, R. B., Ritch, R., Dubra, A., Chui, T. Y. 2017; 6 (4): 6


    Deep glaucomatous defects near fixation were followed over time with an adaptive optics-scanning light ophthalmoscope (AO-SLO) to better understand the progression of these defects and to explore the use of AO-SLO in detecting them.Six eyes of 5 patients were imaged with an AO-SLO from 2 to 4 times for a range of 14.6 to 33.6 months. All eyes had open-angle glaucoma with deep defects in the superior visual field (VF) near fixation as defined by 10-2 VFs with 5 or more points less than -15 dB; two of the eyes had deep defects in the inferior VF as well. AO-SLO images were obtained around the temporal edge of the disc.In 4 of the 6 eyes, the edge of the inferior-temporal disc region of the retinal nerve fiber (RNF) defect seen on AO-SLO moved closer to fixation within 10.6 to 14.7 months. In 4 eyes, RNF bundles in the affected region appeared to lose contrast and/or disappear.Progressive changes in RNF bundles associated with deep defects on 10-2 VFs can be seen within about 1 year with AO-SLO imaging. These changes are well below the spatial resolution of the 10-2 VF. On the other hand, subtle thinning of regions with RNF bundles is not easy to see with current AO-SLO technology, and may be better followed with OCT.AO-SLO imaging may be useful in clinical trials designed to see very small changes in deep defects.

    View details for DOI 10.1167/tvst.6.4.6

    View details for PubMedID 28713646

    View details for PubMedCentralID PMC5508541

  • Multimodal Imaging of Photoreceptor Structure in Choroideremia PLOS ONE Sun, L. W., Johnson, R. D., Williams, V., Summerfelt, P., Dubra, A., Weinberg, D. V., Stepien, K. E., Fishman, G. A., Carroll, J. 2016; 11 (12)


    Choroideremia is a progressive X-linked recessive dystrophy, characterized by degeneration of the retinal pigment epithelium (RPE), choroid, choriocapillaris, and photoreceptors. We examined photoreceptor structure in a series of subjects with choroideremia with particular attention to areas bordering atrophic lesions.Twelve males with clinically-diagnosed choroideremia and confirmed hemizygous mutations in the CHM gene were examined. High-resolution images of the retina were obtained using spectral domain optical coherence tomography (SD-OCT) and both confocal and non-confocal split-detector adaptive optics scanning light ophthalmoscope (AOSLO) techniques.Eleven CHM gene mutations (3 novel) were identified; three subjects had the same mutation and one subject had two mutations. SD-OCT findings included interdigitation zone (IZ) attenuation or loss in 10/12 subjects, often in areas with intact ellipsoid zones; RPE thinning in all subjects; interlaminar bridges in the imaged areas of 10/12 subjects; and outer retinal tubulations (ORTs) in 10/12 subjects. Only split-detector AOSLO could reliably resolve cones near lesion borders, and such cones were abnormally heterogeneous in morphology, diameter and density. On split-detector imaging, the cone mosaic terminated sharply at lesion borders in 5/5 cases examined. Split-detector imaging detected remnant cone inner segments within ORTs, which were generally contiguous with a central patch of preserved retina.Early IZ dropout and RPE thinning on SD-OCT are consistent with previously published results. Evidence of remnant cone inner segments within ORTs and the continuity of the ORTs with preserved retina suggests that these may represent an intermediate state of retinal degeneration prior to complete atrophy. Taken together, these results supports a model of choroideremia in which the RPE degenerates before photoreceptors.

    View details for DOI 10.1371/journal.pone.0167526

    View details for Web of Science ID 000389587100115

    View details for PubMedID 27936069

    View details for PubMedCentralID PMC5147929

  • Noninvasive imaging of the photoreceptor mosaic response to light stimulation. Proceedings of the National Academy of Sciences of the United States of America Srinivasan, V. J., Dubra, A. 2016

    View details for PubMedID 27810954

  • Residual Foveal Cone Structure in CNGB3-Associated Achromatopsia INVESTIGATIVE OPHTHALMOLOGY & VISUAL SCIENCE Langlo, C. S., Patterson, E. J., Higgins, B. P., Summerfelt, P., Razeen, M. M., Erker, L. R., Parker, M., Collison, F. T., Fishman, G. A., Kay, C. N., Zhang, J., Weleber, R. G., Yang, P., Wilson, D. J., Pennesi, M. E., Lam, B. L., Chiang, J., Chulay, J. D., Dubra, A., Hauswirth, W. W., Carroll, J. 2016; 57 (10): 3984-3995


    Congenital achromatopsia (ACHM) is an autosomal recessive disorder in which cone function is absent or severely reduced. Gene therapy in animal models of ACHM have shown restoration of cone function, though translation of these results to humans relies, in part, on the presence of viable cone photoreceptors at the time of treatment. Here, we characterized residual cone structure in subjects with CNGB3-associated ACHM.High-resolution imaging (optical coherence tomography [OCT] and adaptive optics scanning light ophthalmoscopy [AOSLO]) was performed in 51 subjects with CNGB3-associated ACHM. Peak cone density and inter-cone spacing at the fovea was measured using split-detection AOSLO. Foveal outer nuclear layer thickness was measured in OCT images, and the integrity of the photoreceptor layer was assessed using a previously published OCT grading scheme.Analyzable images of the foveal cones were obtained in 26 of 51 subjects, with nystagmus representing the major obstacle to obtaining high-quality images. Peak foveal cone density ranged from 7,273 to 53,554 cones/mm2, significantly lower than normal (range, 84,733-234,391 cones/mm2), with the remnant cones being either contiguously or sparsely arranged. Peak cone density was correlated with OCT integrity grade; however, there was overlap of the density ranges between OCT grades.The degree of residual foveal cone structure varies greatly among subjects with CNGB3-associated ACHM. Such measurements may be useful in estimating the therapeutic potential of a given retina, providing affected individuals and physicians with valuable information to more accurately assess the risk-benefit ratio as they consider enrolling in experimental gene therapy trials. (, NCT01846052.).

    View details for DOI 10.1167/iovs.16-19313

    View details for Web of Science ID 000383981600005

    View details for PubMedID 27479814

    View details for PubMedCentralID PMC4978151

  • In Vivo Imaging of the Human Retinal Pigment Epithelial Mosaic Using Adaptive Optics Enhanced Indocyanine Green Ophthalmoscopy INVESTIGATIVE OPHTHALMOLOGY & VISUAL SCIENCE Tam, J., Liu, J., Dubra, A., Fariss, R. 2016; 57 (10): 4376-4384


    The purpose of this study was to establish that retinal pigment epithelial (RPE) cells take up indocyanine green (ICG) dye following systemic injection and that adaptive optics enhanced indocyanine green ophthalmoscopy (AO-ICG) enables direct visualization of the RPE mosaic in the living human eye.A customized adaptive optics scanning light ophthalmoscope (AOSLO) was used to acquire high-resolution retinal fluorescence images of residual ICG dye in human subjects after intravenous injection at the standard clinical dose. Simultaneously, multimodal AOSLO images were also acquired, which included confocal reflectance, nonconfocal split detection, and darkfield. Imaging was performed in 6 eyes of three healthy subjects with no history of ocular or systemic diseases. In addition, histologic studies in mice were carried out.The AO-ICG channel successfully resolved individual RPE cells in human subjects at various time points, including 20 minutes and 2 hours after dye administration. Adaptive optics-ICG images of RPE revealed detail which could be correlated with AO dark-field images of the same cells. Interestingly, there was a marked heterogeneity in the fluorescence of individual RPE cells. Confirmatory histologic studies in mice corroborated the specific uptake of ICG by the RPE layer at a late time point after systemic ICG injection.Adaptive optics-enhanced imaging of ICG dye provides a novel way to visualize and assess the RPE mosaic in the living human eye alongside images of the overlying photoreceptors and other cells.

    View details for DOI 10.1167/iovs.16-19503

    View details for Web of Science ID 000383981600056

    View details for PubMedID 27564519

    View details for PubMedCentralID PMC5015921

  • Cone Photoreceptor Structure in Patients With X-Linked Cone Dysfunction and Red-Green Color Vision Deficiency INVESTIGATIVE OPHTHALMOLOGY & VISUAL SCIENCE Patterson, E. J., Wilk, M., Langlo, C. S., Kasilian, M., Ring, M., Hufnagel, R. B., Dubis, A. M., Tee, J. J., Kalitzeos, A., Gardner, J. C., Ahmed, Z. M., Sisk, R. A., Larsen, M., Sjoberg, S., Connor, T. B., Dubra, A., Neitz, J., Hardcastle, A. J., Neitz, M., Michaelides, M., Carroll, J. 2016; 57 (8): 3853-3863


    Mutations in the coding sequence of the L and M opsin genes are often associated with X-linked cone dysfunction (such as Bornholm Eye Disease, BED), though the exact color vision phenotype associated with these disorders is variable. We examined individuals with L/M opsin gene mutations to clarify the link between color vision deficiency and cone dysfunction.We recruited 17 males for imaging. The thickness and integrity of the photoreceptor layers were evaluated using spectral-domain optical coherence tomography. Cone density was measured using high-resolution images of the cone mosaic obtained with adaptive optics scanning light ophthalmoscopy. The L/M opsin gene array was characterized in 16 subjects, including at least one subject from each family.There were six subjects with the LVAVA haplotype encoded by exon 3, seven with LIAVA, two with the Cys203Arg mutation encoded by exon 4, and two with a novel insertion in exon 2. Foveal cone structure and retinal thickness was disrupted to a variable degree, even among related individuals with the same L/M array.Our findings provide a direct link between disruption of the cone mosaic and L/M opsin variants. We hypothesize that, in addition to large phenotypic differences between different L/M opsin variants, the ratio of expression of first versus downstream genes in the L/M array contributes to phenotypic diversity. While the L/M opsin mutations underlie the cone dysfunction in all of the subjects tested, the color vision defect can be caused either by the same mutation or a gene rearrangement at the same locus.

    View details for DOI 10.1167/iovs.16-19608

    View details for Web of Science ID 000381729000039

    View details for PubMedID 27447086

    View details for PubMedCentralID PMC4968428

  • Assessing Photoreceptor Structure in Retinitis Pigmentosa and Usher Syndrome INVESTIGATIVE OPHTHALMOLOGY & VISUAL SCIENCE Sun, L. W., Johnson, R. D., Langlo, C. S., Cooper, R. F., Razeen, M. M., Russillo, M. C., Dubra, A., Connor, T. B., Han, D. P., Pennesi, M. E., Kay, C. N., Weinberg, D. V., Stepien, K. E., Carroll, J. 2016; 57 (6): 2428-2442


    The purpose of this study was to examine cone photoreceptor structure in retinitis pigmentosa (RP) and Usher syndrome using confocal and nonconfocal split-detector adaptive optics scanning light ophthalmoscopy (AOSLO).Nineteen subjects (11 RP, 8 Usher syndrome) underwent ophthalmic and genetic testing, spectral-domain optical coherence tomography (SD-OCT), and AOSLO imaging. Split-detector images obtained in 11 subjects (7 RP, 4 Usher syndrome) were used to assess remnant cone structure in areas of altered cone reflectivity on confocal AOSLO.Despite normal interdigitation zone and ellipsoid zone appearance on OCT, foveal and parafoveal cone densities derived from confocal AOSLO images were significantly lower in Usher syndrome compared with RP. This was due in large part to an increased prevalence of non-waveguiding cones in the Usher syndrome retina. Although significantly correlated to best-corrected visual acuity and foveal sensitivity, cone density can decrease by nearly 38% before visual acuity becomes abnormal. Aberrantly waveguiding cones were noted within the transition zone of all eyes and corresponded to intact inner segment structures. These remnant cones decreased in density and increased in diameter across the transition zone and disappeared with external limiting membrane collapse.Foveal cone density can be decreased in RP and Usher syndrome before visible changes on OCT or a decline in visual function. Thus, AOSLO imaging may allow more sensitive monitoring of disease than current methods. However, confocal AOSLO is limited by dependence on cone waveguiding, whereas split-detector AOSLO offers unambiguous and quantifiable visualization of remnant cone inner segment structure. Confocal and split-detector thus offer complementary insights into retinal pathology.

    View details for DOI 10.1167/iovs.15-18246

    View details for Web of Science ID 000378041700010

    View details for PubMedID 27145477

    View details for PubMedCentralID PMC5089122

  • Automatic detection of cone photoreceptors in split detector adaptive optics scanning light ophthalmoscope images. Biomedical optics express Cunefare, D., Cooper, R. F., Higgins, B., Katz, D. F., Dubra, A., Carroll, J., Farsiu, S. 2016; 7 (5): 2036-2050


    Quantitative analysis of the cone photoreceptor mosaic in the living retina is potentially useful for early diagnosis and prognosis of many ocular diseases. Non-confocal split detector based adaptive optics scanning light ophthalmoscope (AOSLO) imaging reveals the cone photoreceptor inner segment mosaics often not visualized on confocal AOSLO imaging. Despite recent advances in automated cone segmentation algorithms for confocal AOSLO imagery, quantitative analysis of split detector AOSLO images is currently a time-consuming manual process. In this paper, we present the fully automatic adaptive filtering and local detection (AFLD) method for detecting cones in split detector AOSLO images. We validated our algorithm on 80 images from 10 subjects, showing an overall mean Dice's coefficient of 0.95 (standard deviation 0.03), when comparing our AFLD algorithm to an expert grader. This is comparable to the inter-observer Dice's coefficient of 0.94 (standard deviation 0.04). To the best of our knowledge, this is the first validated, fully-automated segmentation method which has been applied to split detector AOSLO images.

    View details for DOI 10.1364/BOE.7.002036

    View details for PubMedID 27231641

    View details for PubMedCentralID PMC4871101