A recent paper (Jia et al. European Radiology (2026) 36:2186–2197, https://doi.org/10.1007/s00330-025-12023-z) looks at the emissions caused by medical imaging data storage. Although data storage is currently a relatively small part of the carbon footprint of most cancer treatment, the volume of data generated per patient is rising with time (at my centre, almost exponentially for radiotherapy data in recent years) so this can be expected to be an increasingly significant issue.
The paper finds that a significant fraction of data stored on PACS is “derived” data, so things like MIPs and post-processed images, which do not need to be stored because they can be re-created, if required, from the raw image data. The authors modelled cancer incidence projections, and found that avoiding storing these derived data has the potential to decrease emissions from data storage by almost 70%, over a 20-year period, for their example of chest-abdo-pelvic imaging. The figures will vary depending on the imaging protocols and the modelling assumptions used, but it’s clear that a significant decrease can be achieved just by not storing data that don’t need to be kept.
Minimising emissions due to data storage is something that’s immediately available to us all. Avoid sending to PACS data that can be derived from the raw axial CT images. In addition, don’t store test reconstructions indefinitely; at my centre, we were able to turn off extra reconstructions that had been added to protocols experimentally, but were never removed once the project for which they were intended was completed, and were therefore being added unnecessarily to PACS.
Please log in or sign up to comment.