Dark Data In Retinal Imaging: A Cross-Sectional Audit Of Redundancy And Storage-Related Carbon Impact
Published 2026 - 44th Congress of the ESCRS
Reference: POSUS015 | Type: Free paper | DOI: 10.82333/ym2p-hc32
Authors: Dr Carelys Suescum Coelho* 1 , Ms Carlysmar Suescum Coelho 2 , Dr Car-Emyr Suescum Coelho 3 , Dr Carluys Suescum Coelho 2
1CENTRO OFTALMOLÓGICO,CARACAS,Venezuela, Bolivarian Republic Of, 2CEGA,CARACAS,Venezuela, Bolivarian Republic Of, 3UNIVERSIDAD METROPOLITANA,CARACAS,Venezuela, Bolivarian Republic Of
Purpose
To quantify “dark data” in outpatient age-related macular degeneration (AMD) imaging by measuring digital storage volume and clinically redundant imaging files retained in primary storage, and to model the associated long-term storage-related energy and carbon impact.
Setting
Retrospective cross-sectional audit of a digital ophthalmic imaging archive from a general ophthalmology service; the audited cohort comprised patients followed for AMD
Methods
Electronic health records (EHR) and imaging metadata from 200 patients with age-related macular degeneration (AMD) (≥12 months follow-up) were reviewed. To avoid conflating routine follow-up (6-monthly AMD; monthly anti-vascular endothelial growth factor [anti-VEGF] protocols) with waste, “dark data” was defined per session (same eye/modality/date) as: excess captures (>3 optical coherence tomography [OCT] or fundus photos/eye/visit; >2 visual field tests/eye/visit), duplicate files, and/or sub-threshold image quality retained in primary storage. Storage energy was modelled at 4.2 kWh/TB/year and carbon using the European Attribute Mix (2024; 536.94 gCO₂/kWh).
Results
Mean stored volume was 1.14 GB/patient/year (SD 0.3), totalling 228 GB/year for the cohort. Dark data was identified in 26.5% (53/200) of patient records, predominantly driven by retained low-quality OCT repeats and redundant capture sets beyond the minimum standard per session. Under the storage-intensity model, cohort storage corresponded to 0.96 kWh/year and 0.51 kg CO₂/year. Extrapolated to a high-volume clinic (10,000 patients/year), baseline storage would be 11.4 TB/year (47.9 kWh/year; 25.7 kg CO₂/year), with any retained dark data increasing this proportionally.
Conclusions
Even after accounting for clinically appropriate monitoring intervals, a meaningful proportion of AMD imaging records contained redundant or sub-optimal files retained in primary storage. While emissions per patient-year are small, the cumulative effect of retaining redundant imaging across a high-volume service and over long retention periods can become non-negligible-making digital hygiene a practical sustainability lever. Quality-based retention governance (automated flagging, quarantine/tiered storage, and medicolegal-aligned retention rules) can reduce avoidable long-term digital footprint without compromising patient care.