Skip to main content
  • News in Review

    Exploring a New Approach to Retinal Imaging


    Download PDF

    Researchers at the National Eye Institute (NEI) have developed an artificial intelligence (AI) method that—when combined with adaptive optics optical coherence tomography (AO-OCT)—improved their ability to visualize retinal pigment epithelial (RPE) cells in images taken from par­ticipants with healthy eyes.

    Their study, published in Com­munications Medicine, describes how integration of AI with AO-OCT helped minimize “speckle noise,” a phenome­non that obscures details in images of the RPE. Their experimental AI-driven imaging technique provided results faster than current retinal imaging methods, the authors wrote.1

    Overcoming speckle noise. When ophthalmologists want to take a close look at the retina and the RPE, AO-OCT allows the visualization of RPE cells by capturing high-resolution 3D images of the retina. But due to speckle noise, which obscures cellular details in the average AO-OCT image, hundreds of 3D images are typically required to improve contrast and visualization of the RPE.

    “We wanted to contribute a more efficient way to image the RPE cells directly in the living human eye,” said author Johnny Tam, PhD, who leads NEI’s Clinical and Translational Imag­ing Section, in Bethesda, Maryland.

    Dr. Tam and colleagues developed a novel deep learning algorithm they call P-GAN, or parallel discriminator generative adversarial network. By “feeding” it close to 6,000 manually analyzed AO-OCT images of human RPE, they trained the network to iden­tify and recover features of RPE images that are not obscured by speckles—in other words, teaching it to work around the speckles to reconstruct the cellular tissue hidden behind them.

    “We leveraged natural eye motion to reconstruct a large number of images for each location, increasing the size of our training data by over 40-fold,” said Dr. Tam.

    NEI scientists combined AI with AO-OCT to explore a new method of visualizing the RPE cell mosaic in healthy human eyes.

    SEEING PAST SPECKLES. NEI scientists combined AI with AO-OCT to explore a more efficient method of visualizing the RPE cell mosaic in healthy human eyes.

    Speeding up the RPE imaging pro­cess. Application of the AI method to eight eyes in seven healthy participants showed that using the P-GAN method improved RPE cell contrast by 3.5-fold compared to the traditional AO-OCT method. And it reduced the time re­quired for RPE visualization by nearly 100-fold compared to traditional aver­aging methods, the authors reported.

    Using P-GAN, the researchers could capture the RPE mosaic across large retinal areas (63 overlapping locations) in a much shorter time frame than the six hours it takes with traditional meth­ods, Dr. Tam said.

    According to the authors, RPE cell spacing in AI-recovered images agreed with expected normative ranges from previous studies.

    More validation needed. “The results are impressive with improvement over previous optical or AI-based methods,” said retina specialist Tien Y. Wong, MMed (MD), PhD, MPH, MBBS, Founding Head of Tsinghua Medicine at Tsinghua University in Beijing, China, and Consulting Professor in the Department of Ophthalmology at Duke University, in Durham, North Carolina.

    However, fidelity remains a concern with P-GAN networks, cautioned Dr. Wong, who was not involved in the study.

    Dr. Tam said if validated, AI im­plementation could enhance routine ophthalmic imaging and allow earlier detection of retinal diseases affecting RPE cells.

    “Adaptive optics can reveal cellular details, but the imaging window for each acquisition is small, and the patient needs to remain still for a long time to stitch together a larger image,” said Dr. Tam. But a dramatic improvement in speed, like that demonstrated in their study, “would allow us to better link cellular-scale findings to tissue-scale clinical imaging by capturing a popu­lation of cells and following them over time.”

    Research involving participants with eye diseases. Future studies would also need to evaluate the performance of P-GAN in diseased eyes, where RPE cell appearance and contrast may differ from healthy eyes, said Dr. Tam. His team plans to expand the potential of AI-assisted imaging by exploring its application to other retinal cells and structures, such as ganglion cells and blood vessels.

    “This is an exciting time for oph­thalmology because of this amazing intersection between imaging and AI,” said Dr. Tam. He is hopeful that the incorporation of AI into imaging will eventually help lower the barrier to deploying adaptive optics by making it easier to obtain high-quality images.

    Reflecting on the broader implica­tions of integrating AI with imaging modalities, Dr. Wong said, “Combining advanced imaging methods with AI is the future for optical imaging devices.”


    —Christos Evangelou, PhD


    1 Das V et al. Commun Med. 2024;4(1):68.

    Relevant financial disclosures—Dr. Tam: None. Dr. Wong: None.

    For full disclosures and the disclosure key, see below.

    Full Financial Disclosures

    Dr. Baxter None

    Dr. McCourt None 

    Dr. Sabater AbbVie: C; Brill Pharma: C; Laboratorios Sophia: C; Ocubio: EO; Tissue-Cor: EO. 

    Dr. Swaminathan AbbVie: C; Heidelberg Engineering: S; Lumata Health: C, PS; National Institutes of Health: S; Topcon: C. 

    Dr. Tam None

    Dr. Wong Astellas: C; Bayer: C; Boehringer-Ingelheim: C; Genentech: C; Novartis: C; Plano: C; Roche: C; Sanofi: C; Shanghai Henlius: C; EyRIS: P; Visre: P.

    Disclosure Category



    Consultant/Advisor C Consultant fee, paid advisory boards, or fees for attending a meeting.
    Employee E Hired to work for compensation or received a W2 from a company.
    Employee, executive role EE Hired to work in an executive role for compensation or received a W2 from a company.
    Owner of company EO Ownership or controlling interest in a company, other than stock.
    Independent contractor I Contracted work, including contracted research.
    Lecture fees/Speakers bureau L Lecture fees or honoraria, travel fees or reimbursements when speaking at the invitation of a commercial company.
    Patents/Royalty P Beneficiary of patents and/or royalties for intellectual property.
    Equity/Stock/Stock options holder, private corporation PS Equity ownership, stock and/or stock options in privately owned firms, excluding mutual funds.
    Grant support S Grant support or other financial support from all sources, including research support from government agencies (e.g., NIH), foundations, device manufacturers, and\or pharmaceutical companies. Research funding should be disclosed by the principal or named investigator even if your institution receives the grant and manages the funds.
    Stock options, public or private corporation SO Stock options in a public or private company.
    Equity/Stock holder, public corporation US Equity ownership or stock in publicly traded firms, excluding mutual funds (listed on the stock exchange).


    More from this month’s News in Review