Potential of artificial intelligence for radiation dose reduction in computed tomography -A scoping review.
Authors
Affiliations (3)
Affiliations (3)
- The Discipline of Medical Imaging and Radiation Therapy, University College Cork, College Road, T12 K8AF Cork, Ireland; Faculty of Applied Medical Sciences, Department of Medical Imaging, The Hashemite University, Zarqa, Jordan. Electronic address: [email protected].
- The Discipline of Medical Imaging and Radiation Therapy, University College Cork, College Road, T12 K8AF Cork, Ireland.
- The Discipline of Medical Imaging and Radiation Therapy, University College Cork, College Road, T12 K8AF Cork, Ireland; Syddansk Universitet, University of Southern Denmark Faculty of Health Sciences, Denmark; University of Sydney, Faculty of Medicine, Australia.
Abstract
Artificial intelligence (AI) is now transforming medical imaging, with extensive ramifications for nearly every aspect of diagnostic imaging, including computed tomography (CT). This current work aims to review, evaluate, and summarise the role of AI in radiation dose optimisation across three fundamental domains in CT: patient positioning, scan range determination, and image reconstruction. A comprehensive scoping review of the literature was performed. Electronic databases including Scopus, Ovid, EBSCOhost and PubMed were searched between January 2018 and December 2024. Relevant articles were identified from their titles had their abstracts evaluated, and those deemed relevant had their full text reviewed. Extracted data from selected studies included the application of AI, radiation dose, anatomical part, and any relevant evaluation metrics based on the CT parameter in which AI is applied. 90 articles met the selection criteria. Included studies evaluated the performance of AI for dose optimisation through patient positioning, scan range determination, and reconstruction across various CT scans, including the abdomen, chest, head, neck, and pelvis, as well as CT angiography. A concise overview of the present state of AI in these three domains, emphasising benefits, limitations, and impact on the transformation of dose reduction in CT scanning, is provided. AI methods can help minimise positioning offsets and over-scanning caused by manual errors and helped to overcome the limitation associated with low-dose CT settings through deep learning image reconstruction algorithms. Further clinical integration of AI will continue to allow for improvements in optimising CT scan protocols and radiation dose. This review underscores the significance of AI in optimizing radiation doses in CT imaging, focusing on three key areas: patient positioning, scan range determination, and image reconstruction.