ai-driven computed tomography angiography for coronary artery disease: a systematic review of recent advancements in diagnosis

Introduction: coronary artery disease (cad) accounts for a significant share of deaths worldwide. computed tomography angiography (cta) is a growing non-invasive diagnostic method competing with invasive alternatives. however, it faces challenges such as being individual-based, variable interpretation and time-consuming. artificial intelligence (ai) can improve cta by reducing interpretation time, improving diagnostic accuracy, and optimizing workflow. the aim of this systematic review is to evaluate the effectiveness of ai in cta for cad by changing diagnostic accuracy, workflow efficiency, and statistical prediction, by reviewing all recent relevant studies. search strategy: in this review study, which was conducted based on the prisma guideline, a comprehensive search was conducted on english-language articles published between january 2019 and november 2024. the search utilized following mesh terms: coronary artery disease, computed tomography angiography, artificial intelligence, and machine learning. scientific databases such as pubmed, web of science, science direct, cochrane library, scopus and apa psycnet were used to gather the articles, resulting in a total of 1232 retrieved studies. the inclusion criteria for this study were: valid studies evaluating applications of ai in cta for cad diagnosis, with a preference for randomized controlled trials (rcts) and observational studies, publication within the last five years (from january 2019 to november 2024), and availability of free full text. after removing duplicates, the 75 remaining articles were screened, and after applying the inclusion and exclusion criteria, 35 studies were selected and included in the analysis. results: this systematic review included 35 studies, including 12 rcts and 23 observational studies. ai-enhanced cta demonstrated a sensitivity of 94% and specificity of 92% for the diagnosis of cad, outperforming traditional cta interpretation. significant improvements in stenosis detection, plaque characterization, and automated coronary artery segmentation were reported in the ai models. workflow improvements were inferred due to a reduction in reporting time of up to 40%. furthermore, ai-based predictive models provided greater accuracy and success in identifying patients at risk than simple scoring systems. several studies have shown that ai algorithms hold promise for significantly reducing radiation exposure by optimizing imaging protocols without compromising diagnostic quality. conclusion and discussion: ai-based cta offers significant advantages, including improved diagnostic accuracy, rapid workflow, and predictive insights for patient management. however, challenges such as data heterogeneity, population adaptation issues, and ethical concerns remain. more multicenter rcts are needed to validate the clinical utility of ai and address potential biases. the combination of ai and cta promises to revolutionize the diagnosis and management of cad. ai has proven to reduce the workload of cta, increase its efficiency, and improve the use of new and advanced tools. future studies on the ethical competence of ai in the patient clinic and the establishment of protocols could be useful