Greenhouse gas emissions due to long-term data storage of CT with reformats and strategies for mitigation

Objectives

Medical image data storage and associated greenhouse gas (GHG) emissions are increasing. We aimed to measure non-essential storage and model mitigation strategies.

Materials and methods

The proportion of stored post-processed series (reformats and reconstructions) was retrospectively recorded in 183 baseline staging CT chest–abdomen–pelvis studies (CT-CAP) for endometrial cancer in a UK referral centre between 2013 and 2016 (Cohort A). File size (megabytes, MB) of each series was recorded for 30 studies (Cohort B) and compared with 100 Canadian studies (Cohort C), contextualised by a survey of protocols across 17 global centres (including Cohort C). Storage-associated GHG emissions were modelled over 20 years for various mitigation strategies.

Results

Post-processed series were stored in 179/183 (97%) of cohort A, 29/30 (97%) of cohort B and 16/17 (94%) of global centres. Median file size was 787 MB (IQR 460, 1257) for the entire CT study (all stored series) and 290 MB (224, 355) for the acquired axial series alone. On-premises storage of all series for new UK endometrial cancer baseline studies 2020–2040 is estimated to generate 381 metric tons CO2 equivalent (MTCO2e). Over this period, modelled mitigation strategies achieved emission reductions of 69% by storing only acquired axial series (117MTCO2e), 82% combining axial-only with cloud storage (70MTCO2e), 81% combining axial-only with an 8-year data retention policy (72MTCO2e), and 89% combining all three strategies (43MTCO2e).

Conclusion

CT data storage has a large environmental cost, necessitating global action. Various mitigation strategies are achievable in reducing storage-related emissions by up to 89%.