Quantitative AI-based OCT analysis demonstrated that the ELM and EZ have distinct yet complementary relationships with fluid biomarkers and visual acuity, with EZ alterations more prevalent and associated with inflammatory foci, while ELM disruption reflects deeper structural damage and stronger functional impairment.
EZ interruption was nearly twice as prevalent as ELM interruption in eyes with diabetic macular edema.
Eyes with interrupted ELM and/or EZ showed significantly greater intraretinal fluid and subretinal fluid volumes.
Inflammatory hyper-reflective retinal foci counts were significantly higher in eyes with predominant EZ damage.
ELM disruption showed the strongest correlation with visual acuity and subretinal fluid burden among the outer retinal layer biomarkers examined.
Four distinct patterns of ELM/EZ integrity were identified and used to classify eyes in the study population.
This was a large cross-sectional population study using an AI platform to quantify multiple OCT biomarkers simultaneously in DME.
This research suggests that two thin layers at the back of the eye — called the external limiting membrane (ELM) and ellipsoid zone (EZ) — behave differently in people with diabetic macular edema (DME), a condition where fluid accumulates in the central part of the retina and can impair vision. Using an artificial intelligence tool to analyze detailed eye scans (OCT) from over 2,300 eyes across nearly 1,700 patients in Italy, researchers found that the EZ layer was disrupted more frequently (in about 38% of eyes) than the ELM layer (about 22% of eyes), and this was true whether or not patients had been previously treated for DME. The two layers were also linked to different aspects of disease severity. Disruption of the EZ was more strongly associated with higher counts of inflammatory spots within the retina, suggesting a connection between EZ damage and retinal inflammation. By contrast, disruption of the ELM was more strongly linked to worse visual acuity and greater amounts of fluid beneath the retina, suggesting it reflects more serious structural and functional damage to the eye. Eyes with disruption of either layer tended to have larger overall fluid volumes and fluid concentrated more toward the center of the retina. This research suggests that mapping the integrity of these two retinal layers — rather than treating them as a single indicator — could help doctors better understand how advanced a patient's DME is and tailor treatment decisions more precisely. The use of AI to automatically measure these subtle features from routine eye scans could make this kind of detailed assessment more practical in clinical settings, potentially supporting more personalized approaches to managing diabetic eye disease.
Midena E, Lupidi M, Frizziero L, Midena G, Pilotto E, Toto L, et al.. (2026). The Behavior of the External Limiting Membrane and Ellipsoid Zone in Diabetic Macular Edema: A Population AI-Assisted Optical Coherence Tomography Study.. Investigative ophthalmology & visual science. https://doi.org/10.1167/iovs.67.11.40