Application value of artificial intelligence in dual-source CT coronary angiography / 中国基层医药

ABSTRACT

Objective:To investigate the application value of artificial intelligence in dual-source CT coronary angiography.Methods:The imaging data of 50 patients with coronary artery diseases who received treatment in Benxi Central Hospital from January to December 2021 were retrospectively analyzed. All patients underwent coronary computed tomography angiography examination and coronary arteriography. Coronary computed tomography angiography images were uploaded to the post-processing workstation and post-processed and analyzed by two radiologists. At the same time, the images were also post-processed by AI software and diagnosis reports were generated. In terms of coronary artery stenosis, the diagnostic results provided by AI software were compared with coronary angiography results. In terms of myocardial bridging, the diagnostic results provided by AI software were compared with radiologist's diagnosis. The accuracy of AI software was judged.Results:In the diagnosis of coronary artery stenosis, the sensitivity, specificity, positive predictive value and the negative predictive value of AI software were 93.22%, 81.32%, 76.39% and 94.87% respectively. These were well consistent with coronary arteriography results (Kappa = 0.71, P < 0.001). In the diagnosis of myocardial bridging, the sensitivity, specificity, positive predictive value and the negative predictive value of AI software were 30.77%, 81.45%, 25.81% and 84.87% respectively. These were poorly consistent with radiologists' diagnostic results (Kappa = 0.11, P = 0.162). Conclusion:AI-based dual-source CT coronary angiography is of high value in the diagnosis of coronary artery stenosis, but it is of low value in the diagnosis of myocardial bridging.