Resolving the open questions in individualized and cumulative patient dose in the era of AI.

Authors

Abstract

Radiation dose in medical imaging has been quantified through measurable metrics such as computed tomography dose index volume (CTDIvol), dose length product (DLP), kerma area product, kerma at reference point etc. They are benchmarked through diagnostic reference levels (DRLs); and communicated in terms of effective dose and population-based risk estimates. None of these tools was designed to describe the dose, the reference standard, or the risk applicable to the individual patient. In every other domain of medicine where a potentially harmful agent accompanies a beneficial clinical intervention, from contrast nephrotoxicity in CT imaging to anesthetic risk in surgery, the field has moved beyond population metrics to individualized measurement and individual clinical decision-making. Radiation protection in medical imaging has not yet made this transition. This paper resolves several open questions on cumulative dose that have long been used to justify inaction. It argues that the transition is now technically achievable, that the barriers to it are primarily architectural, regulatory, and cultural rather than physical or computational, and that artificial intelligence and emerging imaging technologies provide the enabling infrastructure to complete it. We describe the three-tier structure of barriers to individual dose tracking: technical, institutional and regulatory, and cultural; and propose the elements of a system that would place the individual patient's cumulative radiation dose record where it belongs: in their medical record, alongside their medication list, visible at the point of every subsequent clinical imaging decision.

Tags

Topics