Evaluation Of The Diagnosis And Application Value Of CCTA Assisted Diagnosis System Based On Deep Learning For Coronary Heart Disease

Objective To investigate the diagnostic efficacy of artificial intelligence-aided diagnostic system in coronary computed tomography angiography(CCTA)for atherosclerotic stenosis of coronary artery disease and to evaluate its application in the post-processing work of CCTA images.Methods A total of 112 patients with clinically suspected coronary heart disease received CCTA scan in the People’s Hospital of Inner Mongolia Autonomous Region were collected.All patients completed DSA examination of coronary artery within 3months after CCTA scan.The obtained CCTA images were post-processed and diagnosed by AI system and manual,and the post-processed time,plaque location,plaque properties and vascular stenosis degree were recorded.To DSA results as the gold standard,calculate the AI system with artificial diagnosis of moderate coronary artery stenosis and severe stenosis of sensitivity,specificity,positive predictive value,negative predictive value,accuracy and Youden index,and making the receiveroperating characteristic(ROC)curve,calculate the area under the curve(AUC),evaluate the AI system effectiveness to the diagnosis of coronary heart disease,AI system and artificial diagnosis plaque is analyzed and the consistency of the nature of plaque and to explore the AI system compared to manual operation whether has the advantage in image post-processing work efficiency.Results Through the statistical analysis,on the diagnosis of coronary artery stenosis degree,AI and artificial diagnosis moderately narrow area under the ROC curve(AUC)was 0.876 and 0.878,respectively,the difference between the AUC is 0.00137,diagnosis of severe stenosis of the area under the ROC curve(AUC)was 0.906 and0.924,respectively,the difference between the AUC is 0.0179,the results were no statistical difference(P = 0.9639,P = 0.3066);In the diagnosis of plaque site,the AI system and manual had good consistency(Kappa = 0.917,95%CI:0.8919-0.9413).In the diagnosis of plaque properties,the AI system was consistent with the artificial system(Kappa = 0.879,95%CI:0.8489-0.9087).In terms of post-processing time,the AI system and manual system took(337.14±33.98)s and(544.78±29.17)s,t=79.92,and the difference was statistically significant(P < 0.01).Conclusions(1)In diagnosing the degree of coronary artery stenosis in coronary artery disease,AI system has higher diagnostic value;(2)In diagnosing coronary artery plaque site and plaque nature,AI system has better agreement with manual diagnosis;(3)In CCTA image post-processing work,AI system has obvious advantage in processing time compared with manual operation.