A total of 122 studies were selected for evidence mapping and 9 for the meta-analysis. Most studies were conducted in Asia (50.8%), followed by North America (24.6%) and Europe (23.8%), with more studies undertaken in China (n=49) and the USA (n=31) than in other countries. Of the studies using CT imaging modalities (n=111), AI-based analysis was used in 46 computerized tomography fractional flow reserve, 36 plaque or stenosis, and 29 calcium scoring studies. , the pooled sensitivity and specificity for the diagnostic performance of AI applications for detecting ≥50% stenosis was 0.94 (95% CI 0.84–0.98) and 0.69 (95% CI 0.60–0.76), respectively; for vessel-level data the sensitivity and specificity was 0.81 (95% CI 0.69–0.89) and 0.88 (95% CI 0.84–0.91). The areas under the summary receiver operating characteristic curves for patient- and vessel-data were 0.83 (95% CI 0.79–0.86) and 0.92 (95% CI 0.89–0.94), respectively. Additionally, there was considerable heterogeneity between studies included in the meta-analysis, and QUASDAS-2 showed that the risk of bias was generally low for the index test, reference standard, and flow and timing domains, while the risk for the patient selection domain was unclear. Plaque imaging studies were not analyzed with a meta-analysis in our study due to variability between included studies in terms of how AI was used to quantify plaques. This highlights the need for additional studies to assess the diagnostic performance of AI in plaque imaging.