Artificial intelligence in personalized computed tomography dose optimization: a systematic review of techniques and clinical outcomes.

INTRODUCTION: Artificial intelligence (AI) for personalized computed tomography (CT) dose optimization has gained increasing attention due to its potential to reduce radiation exposure while maintaining image quality. CT dose optimization tailors according to individual patient characteristics, including age, weight, body composition, and clinical condition. The study aims to evaluate AI techniques in CT dose optimization, focusing on reducing radiation exposure while maintaining image quality. METHODS: The relevant studies published between 2010 and 2024 were screened from databases, such as PubMed, IEEE Xplore, Scopus, ScienceDirect, and Web of Science, by following "Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA)" guidelines and "Population, Intervention, Comparison, and Outcome (PICO)" model. Risk of bias analysis revealed possible biases in studies. Meta-analysis was performed to elucidate the symmetry of the evidence and data from relevant studies. RESULTS: The review underscores the significance of deep learning (DL) and machine learning (ML) algorithms, including reconstructive strategies, predictive modeling, and hybrid models, as critical in enhancing dose optimization. CONCLUSION: AI-driven approaches in CT dose optimization show promising potential to reduce radiation exposure safely and effectively without compromising image quality. Future research necessitating larger, diverse clinical trials are essential to validate these findings and ensure the equitable application of AI technologies across varied patient demographics. Studies have consistently indicated AI's potential to reduce the dose exposure while maintaining the image and diagnostic quality. They highlight the importance of DL- and ML-based algorithms, such as reconstructive strategies, predictive modeling, and hybrid models, in CT dose optimization.