Introduction: CT imaging is essential in oncology, yet its widespread use raises concerns about radiation exposure. Therefore, dose optimization strategies are needed without compromising image quality. Recent advances in iterative reconstruction and AI-based algorithms offer promising solutions. Methods: This retrospective study was conducted between October 2023 and June 2024 at IRCCS CROB Rionero in Vulture. Examinations were performed on an older-generation Toshiba scanner and two new-generation Canon Aquilion ONE Genesis 320-slice scanners. The study included 14 brain CTs, 28 chest CTs, and 15 triphasic chest-abdomen-pelvis CTs, totaling 174 scans. Dose indicators (CTDIvol and DLP) were recorded and compared between the two CT systems and against diagnostic reference levels (DRLs) available in the literature. Results: The implementation of AI-based reconstruction techniques led to significant dose reactions. In chest and chest-abdomen-pelvis examinations, CTDIvol decreased by up to 82% and DLP by 70% compared to the older system. For brain imaging, iterative reconstruction was preferred to avoid excessive noise suppression associated with AI approaches, resulting in optimized protocols that preserved diagnostic image quality while enhancing anatomical detail. Conclusion: Integrating AI algorithms and iterative reconstruction techniques in CT imaging can substantially reduce radiation exposure while maintaining excellent diagnostic reliability.
Russo et al. (2025) studied this question.