With the emergence of advanced technology to date in the artificial intelligence (AI), computer aided diagnosis has gradually gained its popularity in the field of healthcare. Particularly, in the clinical practice of coronary artery disease diagnosis, the application of AI could be of great implication in alleviating the shortage of medical sources. To evaluate the effectiveness and safety of the AI-based coronary CT angiographic analysis software (RuiXin-CoronaryAI) for diagnosis of coronary artery stenosis, a retrospective, multi-center, cross-over designed, blinded, sensitivity superiority and specificity non-inferiority clinical trial will be conducted.

Patients ≥18 years old with suspected or known coronary artery disease who underwent CCTA will be included. CCTA images of subjects should be of good quality up to the DICOM 3.0 standards, obtained by CT scan with ≥64-slices. The subjects with unqualified CTA will be excluded. CCTA images will be analyzed in three methods (3 groups). Control group: CCTA images will be visually evaluated by physicians. Experiment group: CCTA images will be evaluated by physicians using RuiXin-CoronaryAI. Reference group: CCTA images will be visually evaluated by cardiologists with at least 10 years experiences, and the conclusions they offer will be used as golden standard. Primary outcomes are diagnostic sensitivity and specificity of RuiXin-CoronaryAI and coronary CTA for diagnosis of ischemia on a per-vessel basis. The effectiveness of RuiXin-CoronaryAI for diagnosis of coronary artery stenosis will be conducted by testing superiority of diagnostic sensitivity and non-inferiority of specificity.

Inclusion Criteria:

1. layer thickness of CCTA images should be less than 1mm, image quality should be up to DICOM 3.0 standards;
2. vessels should be clearly developed, contrast medium ought to be well filled, the average of CT value of aortic root cavity should be between 325-600HU in CCTA image;
3. remodeling of vessels should be intact, including coronary artery and branches, without missed or inaccurate slices;
4. CCTA image should be obtained from single- or dual-source computed tomography (CT) scanners with a minimum of 64 detector rows.

Exclusion Criteria:

1. CCTA image is of poor quality due to motion artifact, severe calcification, metal coverage, noise, poor contrast medium injection and other variables influencing the diagnosis of stenosis;
2. previous percutaneous coronary intervention (PCI) or coronary artery bypass grafting (CABG);
3. anomalous origin of coronary artery;
4. other non-atherosclerosis-related coronary diseases like coronary artery fistula, aneurysm, coronary artery ectasia, arteritis coronaria, etc.;
5. repeated enrollment;
6. other conditions not suitable for enrollment.

Patients ≥18 years old with suspected or known coronary artery disease who underwent CCTA examination.