About

Theodore Leng is a Professor of Ophthalmology with a focus on Ophthalmology Research and Clinical Trials. He is also, by courtesy, associated with the Department of Anesthesiology, Perioperative and Pain Medicine at Stanford University. His work is centered on the intersection of artificial intelligence and medicine, particularly within the context of the Center for Artificial Intelligence in Medicine & Imaging (AIMI). As a faculty member at Stanford, he contributes to advancing research and clinical applications in ophthalmology, leveraging AI technologies to improve healthcare outcomes.

Research topics

• Medicine
• Computer Science
• Political Science
• Artificial Intelligence
• Engineering ethics
• Law
• Pathology
• Ophthalmology
• Engineering

Selected publications

• A Simplified Classification for Age-Related Macular Degeneration Based on Optical Coherence Tomography

  Ophthalmic surgery, lasers & imaging retina · 2026-05-21

  article

  BACKGROUND AND OBJECTIVE: As optical coherence tomography (OCT) has enabled the identification of an expanding set of age-related macular degeneration (AMD) risk biomarkers and become central to routine clinical practice, there remains a need for a simplified grading scheme that allows physicians to communicate and synchronize AMD grading directly from standard OCT imaging rather than relying on traditional color fundus imaging. This study aims to establish a standardized OCT-based AMD classification that balances diagnostic accuracy with practicality for use across clinical and research settings. PATIENTS AND METHODS: Spectral-domain OCT scans were independently graded by two retinal specialists following the newly proposed Stanford OCT-Based AMD Classification (SOAC). Discrepancies were adjudicated by a third independent retinal specialist. Inter-grader agreement was assessed using weighted kappa coefficients. RESULTS: Among the 109 eyes from 108 patients (mean age 79.61 ± 7.57 years; 41.7% men, 58.3% women), AMD staging based on SOAC was distributed as follows: normal aging in nine patients (8.3%), early AMD in 16 (14.7%), intermediate AMD in 32 (29.4%), neovascular AMD (nAMD) in 18 (16.5%), geographic atrophy (GA) in 20 (18.3%), and combined nAMD and GA in 14 (12.8%). The overall intergrader agreement demonstrated robust consistency, with a weighted kappa value of 0.95 (95% CI: 0.92-0.98), signifying excellent intergrader reliability and reinforcing the validity of SOAC. CONCLUSION: SOAC provides a standardized, OCT-based framework for AMD grading that demonstrates high intergrader agreement. By enabling consistent classification from commonly acquired OCT scans, SOAC supports reliable disease staging and facilitates integration across clinical studies and translational research. As imaging and molecular data continue to expand, SOAC can serve as a common OCT-based reference for phenotype refinement and longitudinal AMD studies.

  PublisherDOI

• One-year Real-world Outcomes With Faricimab in Neovascular Age-related Macular Degeneration

  Ophthalmic surgery, lasers & imaging retina · 2026-03-27

  article

  Background and Objective: This study evaluated real-world treatment patterns and 1-year outcomes in patients with neovascular age-related macular degeneration (nAMD) initiating faricimab. Patients and Methods: FARETINA-AMD was a retrospective study using data from the IRIS ® Registry for patients diagnosed with nAMD initiating faricimab from February 2022 to March 2023. Results: Included in the study were 2,025 treatment-naive patients (2,184 eyes) and 22,253 patients (26,851 eyes) previously treated with anti-vascular endothelial growth factor (anti-VEGF) therapy. Visual acuity improved by 2.0 ± 15.0 (mean ± SD) letters in treatment-naive eyes ( P < .001) and was maintained in previously treated eyes at injection 7. Central subfield thickness (CST) improved by −53.1 ± 64.8 μm in treatment-naive and −28.5 ± 79.7 μm in previously treated eyes (both P < .0001); 78.4% and 66.8%, respectively, had achieved/maintained C ST ≤ 280 μm at injection 7. Dosing frequency was reduced in the second 6 months (mean 2.4–3.2 injections) versus the first 6 months (4.0–4.2) of treatment. Conclusion: Outcomes among patients with nAMD receiving faricimab over 1-year follow-up support the real-world effectiveness and extended durability of treatment.

  PublisherDOI

• A Framework for Healthcare from the Eye

  Ophthalmology · 2026-01-21

  article
  PublisherDOI

• OCT to OCTA translation using Brownian bridge diffusion model

  2026-01-16

  article
  PublisherDOI

• A Simplified Classification for Age-Related Macular Degeneration Based on Optical Coherence Tomography

  medRxiv · 2026-03-31

  articleOpen access

  Abstract Background and Objective As optical coherence tomography (OCT) has enabled the identification of an expanding set of age-related macular degeneration (AMD) risk biomarkers and become central to routine clinical practice, there remains a need for a simplified grading scheme that allows physicians to communicate and synchronize AMD grading directly from standard OCT imaging rather than relying on traditional color fundus imaging. This study aims to establish a standardized OCT-based AMD classification that balances diagnostic accuracy with practicality for use across clinical and research settings. Patients and Methods Spectral-domain optical coherence tomography scans were independently graded by two retinal specialists following the newly proposed Stanford OCT-Based AMD Classification (SOAC). Discrepancies were adjudicated by a third independent retinal specialist. Intergrader agreement was assessed using weighted kappa coefficients. Results Among the 109 eyes from 108 patients (mean age 79.61□±□7.57 years; 41.7% male, 58.3% female), AMD staging based on SOAC was distributed as follows: normal aging in 9 patients (8.3%), early AMD in 16 (14.7%), intermediate AMD in 32 (29.4%), neovascular AMD (nAMD) in 18 (16.5%), geographic atrophy (GA) in 20 (18.3%), and combined nAMD and GA in 14 (12.8%). The overall intergrader agreement demonstrated robust consistency, with a weighted kappa value of 0.95 (95% CI: 0.92–0.98), signifying excellent intergrader reliability and reinforcing the validity of SOAC. Conclusion SOAC provides a standardized, OCT-based framework for AMD grading that demonstrates high intergrader agreement. By enabling consistent classification from commonly acquired OCT scans, SOAC supports reliable disease staging and facilitates integration across clinical studies and translational research. As imaging and molecular data continue to expand, SOAC can serve as a common OCT-based reference for phenotype refinement and longitudinal AMD studies. Key questions What is already known on this topic? Existing classifications of age-related macular degeneration (AMD) rely heavily on color fundus photography and inconsistently incorporate optical coherence tomography (OCT) biomarkers, despite OCT’s superior resolution for detecting subclinical structural changes. What this study adds? This study introduces an OCT-based classification framework that consolidates key biomarkers with proven prognostic relevance into a simplified, tiered staging protocol. By prioritizing a curated set of evidence-supported, high-risk OCT biomarkers, the Stanford OCT-Based AMD Classification (SOAC) provides a common framework and shared language for OCT-based AMD staging, facilitates longitudinal monitoring, and reduces intergrader variability through standardized reporting guidelines. The framework deliberately balances clinical practicality with scientific rigor, excluding less significant or potentially confusing features to ensure scalability across diverse clinical and research settings. How this study might affect research, practice or policy? SOAC provides an OCT-based framework for AMD staging with high inter-physician agreement. This is particularly important in real-world clinical and translational research settings, where color fundus photography is not always available. By offering a common OCT-based reference, SOAC reduces diagnostic variability and supports more consistent AMD staging across clinicians and studies. The classification incorporates up-to-date, evidence-based, high-risk OCT features associated with disease progression and is designed to be scalable, allowing integration with emerging multimodal data, including AI-driven imaging analysis and molecular profiling. By bridging advances in retinal imaging with practical clinical application, SOAC lays the groundwork for improved risk stratification, longitudinal outcome studies, and more standardized reporting in AMD.

  PublisherOA PDFDOI

• FedSim: foundational federated multi-task learning for ophthalmic diagnostics

  2026-01-16

  article
  PublisherDOI

• A Matching-Adjusted Comparison of Faricimab and Aflibercept 8 mg in Neovascular Age-Related Macular Degeneration and Diabetic Macular Edema

  Ophthalmic Research · 2026-05-15

  articleOpen access

  INTRODUCTION: Matching adjusted network meta-analyses (NMA) provide an established method to indirectly compare treatments across trials that share a common comparator. In this exploratory analysis, we conducted a matching-adjusted NMA to compare the efficacy of faricimab with aflibercept 8 mg for treatment of diabetic macular edema (DME) and neovascular age-related macular degeneration (nAMD) via the common comparator of aflibercept 2 mg during the first 12 weeks of treatment when all agents are dosed equally. METHODS: Weighting was applied to match patient populations from YOSEMITE/RHINE (NCT03622580/NCT03622593) and TENAYA/LUCERNE (NCT03823287/NCT03823300) based on their baseline characteristics to published aggregated baseline characteristics for PHOTON (NCT04429503) and PULSAR (NCT04423718), respectively. The matched populations were used to recalculate outcomes from faricimab trials and an NMA anchored to aflibercept 2 mg was conducted. The analysis focused on change in best-corrected visual acuity (BCVA) in Early Treatment Diabetic Retinopathy Study letters and central subfield thickness (CST) in µm during the first 12 weeks of treatment when dosing was every 4 weeks for all three agents. Results are expressed as mean difference in change in BCVA or CST for each aflibercept dose. RESULTS: For nAMD, BCVA change was similar between faricimab and aflibercept 2 mg and 8 mg, whereas CST reduction was greater for faricimab versus aflibercept 8 mg (-17.0 µm; 95% credible interval [CrI], -25.0, -8.4) and 2 mg (-19.0 µm; 95% CrI, -24.0, -13.0). For DME, BCVA change from baseline at 12 weeks was similar between faricimab and aflibercept 2 mg and 8 mg, while CST reduction was greater for faricimab versus aflibercept 8 mg (-19.0 µm; 95% CrI, -35.0, -3.2) and 2 mg (-19.0 µm; 95% CrI, -27.0, -12.0). CONCLUSION: This matching-adjusted NMA indicated that dual angiopoietin-2/vascular endothelial growth factor (VEGF)-A inhibition with faricimab was associated with greater retinal drying during the matched-dosing phase in patients with nAMD and DME. Early dual pathway inhibition may therefore improve outcomes beyond anti-VEGF monotherapy.

  PublisherDOI

• GlaucomaVLM : a domain-specific vision-language model for glaucoma

  2026-01-16

  article
  PublisherDOI

• Anatomically guided vision–language model for efficient OCT disease classification and reporting

  2026-01-16

  article
  PublisherDOI

• Correction: Aflibercept 8 mg versus Faricimab Treat-and-Extend for Diabetic Macular Edema or Neovascular Age-Related Macular Degeneration: A Bayesian Fixed-Effect Network Meta-analysis of Clinical Trials

  Ophthalmology and Therapy · 2026-04-04

  articleOpen access

  and the arrow pointing to the right is labelled as "favors group [1]", as shown in the corrected figure.This is because a positive change in BCVA reflects a clinical benefit, compared with CST/ CRT, where an improvement is shown by a reduction, hence the directional labels in Fig. 2A should be opposite to those in Fig. 2 and2C.For completeness and transparency, the old incorrect versions are displayed below.

  PublisherOA PDFDOI

Frequent coauthors

• Daniel L. Rubin

  Stanford University

  50 shared

• Luís de Sisternes

  Carl Zeiss (United States)

  43 shared

• Darius M. Moshfeghi

  32 shared

• Qiang Chen

  Nanjing University of Science and Technology

  21 shared

• Peter A. Karth

  Minnesota Eye Consultants

  19 shared

• Sijie Niu

  University of Jinan

  19 shared

• Minhaj Nur Alam

  17 shared

• Yannis M. Paulus

  University of Michigan–Ann Arbor

  16 shared

Education

• MD

  Stanford University School of Medicine

  2005

• MS, Biological Sciences

  Stanford University

  2000

• AB/BS, Philosophy and Biological Sciences

  Stanford University

  1999