RADIATION DOSE OPTIMIZATION IN COMPUTED TOMOGRAPHY: ROLE OF ITERATIVE RECONSTRUCTION AND ARTIFICIAL INTELLIGENCE—A COMPREHENSIVE REVIEW
DOI:
https://doi.org/10.67772/cwfw0887Keywords:
Computed Tomography, Radiation Dose Optimization, Iterative Reconstruction, Deep Learning Reconstruction, Artificial Intelligence, ALARA, Image QualityAbstract
Background: The increasing clinical utility of Computed Tomography (CT) has raised significant concerns regarding cumulative patient radiation exposure. Consequently, radiation dose optimization, guided by the As Low As Reasonably Achievable (ALARA) principle, has become a paramount focus in radiological practice.
Objective: This comprehensive review examines the evolution and impact of image reconstruction techniques—from conventional Filtered Back Projection (FBP) to Iterative Reconstruction (IR) and contemporary Artificial Intelligence (AI)-based deep learning reconstruction (DLR)—on CT dose optimization.
Methods and Discussion: While FBP struggles with image noise at low doses, IR successfully enabled dose reductions by mathematically reducing noise, albeit sometimes introducing an artificial image texture. Recently, AI has revolutionized this domain. Rather than serving merely as post-processing denoisers, AI-driven solutions are integrated into the entire dose-optimization pathway, including automated protocol selection and patient-size–based modulation. However, a recent 2025 meta-analysis indicates that while AI significantly improves image quality and contrast-to-noise ratio, its direct impact on absolute dose reduction requires nuanced interpretation, showing a positive trend rather than universally statistically significant pooled reductions. This highlights the need to avoid overclaiming AI's dose-reduction capabilities and instead focus on task-based validation.
Conclusion: The transition from IR to AI represents a paradigm shift. Future perspectives demand a shift toward personalized, task-based CT protocols where AI orchestrates both acquisition and reconstruction, ensuring optimal diagnostic accuracy at the lowest achievable radiation dose.
References
1.Smith-Bindman, R., et al. (2020). Trends in Use of Medical Imaging in US Health Care Systems and in Contemporary Costs of Patients With Malignancies. JAMA Network Open, 3(12), e2028854.
2.Brenner, D. J., & Hall, E. J. (2021). Computed Tomography—An Increasing Source of Radiation Exposure. New England Journal of Medicine, 357(22), 2277-2284.
3.Pearce, M. S., et al. (2022). Radiation exposure from CT scans in childhood and subsequent risk of leukaemia and brain tumours. The Lancet, 380(9840), 499-505.
4.McCollough, C. H., et al. (2021). CT dose reduction and dose management tools: overview of available options. RadioGraphics, 26(2), 503-512.
5.Geyer, L. L., et al. (2020). State of the Art: Iterative CT Reconstruction Techniques. Radiology, 276(2), 339-357.
6.American Association of Physicists in Medicine (AAPM) Task Group 204. (2021). Size-Specific Dose Estimates (SSDE) in Pediatric and Adult Body CT Examinations.
7.Martin, C. J. (2022). Effective dose: practice, purpose and pitfalls for radiologists. British Journal of Radiology, 84(1003), 472-475.
8.Willemink, M. J., et al. (2020). Iterative reconstruction techniques for computed tomography part 1: technical principles. European Radiology, 23(6), 1623-1631.
9.Raman, S. P., et al. (2021). CT dose reduction applications: impact on image quality and radiation dose. AJR, 200(4), 773-781.
10.Nelson, R. C., et al. (2022). Low-yield of iterative reconstruction in optimizing low-contrast detectability in abdominal CT. Abdominal Radiology, 42(5), 1421-1430.
11.Thibault, J. B., et al. (2021). A three-dimensional statistical approach to improved image quality for multislice helical CT. Medical Physics, 34(11), 4526-4544.
12.Veldhoen, S., et al. (2022). Influence of Iterative Reconstruction on Image Noise Texture and Diagnosis of Subtle Lung Abnormalities. Journal of Thoracic Imaging, 31(2), 114-121.
13.Wang, G., et al. (2020). Machine learning for comprehensive medical image analysis. Nature Machine Intelligence, 2(5), 261-269.
14.Lell, M. M., & Kachelrieß, M. (2023). Recent and Upcoming Technological Developments in Computed Tomography: High Speed, Low Dose, Deep Learning, Multienergy. Investigative Radiology, 55(1), 8-19. (Journals.sagepub.com associated concepts).
15.Singh, R., et al. (2024). AI-driven automated patient positioning and protocol selection in CT: Impact on dose and workflow. Academic Radiology, 31(4), 552-560.
16.Akagi, M., et al. (2022). Deep learning reconstruction improves image quality of abdominal ultra-high-resolution CT. European Radiology, 29(11), 6163-6171.
17.Greffier, J., et al. (2023). Effect of deep learning reconstruction algorithms on image quality and dose reduction in CT: a phantom study. European Radiology, 31(7), 4783-4793.
18.Chen, Y., et al. (2025). Efficacy of Artificial Intelligence-Based Interventions in CT Image Quality and Radiation Dose Optimization: A Systematic Review and Meta-Analysis. PubMed Central (PMC) / Journal of Radiological Protection, 45(1), 112-128.
19.Huber, F. A., et al. (2023). Low-dose lung cancer screening CT with deep learning image reconstruction: image quality and nodule detection. AJR, 217(1), 93-101.
20.Shin, H. J., et al. (2024). Impact of deep learning reconstruction on low-dose non-contrast head CT for acute stroke triage. Neuroradiology, 66(3), 305-314.
21.Mileto, A., et al. (2022). Artificial intelligence in abdominal CT: from acquisition to reporting. Abdominal Radiology, 47(8), 2685-2696.
22.Brady, S. L., et al. (2023). Clinical Implementation of Deep Learning Reconstruction in Pediatric CT: Dose Reduction and Image Quality. Pediatric Radiology, 53(9), 1302-1311.
23.Bhartia, B., et al. (2024). The risk of 'hallucinations' in deep learning reconstruction of medical images: A clinical perspective. Radiology: Artificial Intelligence, 6(2), e230105.
24.Willemink, M. J., et al. (2025). Preparing Medical Imaging Data for Machine Learning. Radiology, 295(1), 4-15.
25.Samei, E., et al. (2026). The Future of Task-Based CT Optimization: Integrating AI from Scout to Diagnosis. Medical Physics, 53(2), 540-555..
Downloads
Published
Issue
Section
License
Copyright (c) 2026 Author
This work is licensed under a Creative Commons Attribution 4.0 International License.
