〔Abstract〕 Objective To evaluate the accuracy and feasibility of coronary artery calcium score(CACS) on virtual non-contrast scan(VNC) images obtained from coronary artery CT angiography(CCTA) scan with dual-layer spectral detector CT(SDCT). Methods The data of 197 patients who underwent CCTA scan in hospital were analyzed retrospectively, and 88 patients with CACS>0 were further analyzed. Linear regression analysis of CACS and coronary artery calcium volume(CACV) of true non-contrast(TNC) images and VNC images(CACS-TNC, CACS-VNC, CACV-TNC, CACV-VNC) was performed to obtain linear regression equation and correction coefficients λ1AVGand λ2AVG. CACS-VNC and CACV-VNC were corrected by the corresponding regression equation and recorded as CCACS-VNC and CCACV-VNC, respectively. Spearman correlation coefficient was used for correlation analysis and Bland-Altman plot was used for consistency test. Mann-WhitneyUtest was used to compare the difference between the two groups. Results For the total coronary artery, there was a strong correlation between CACS-TNC and CACS-VNC(rs=0.952,P<0.001, λ1AVG=2.19), CACV-TNC and CACV-VNC(rs=0.954,P<0.001, λ2AVG=1.93). The results of Mann-WhitneyUtest showed that there was no significant difference between CACS-TNC and CCACS-VNC or between CACV-TNC and CCACV-VNC, and the Bland-Altman plot showed good consistency between CACS-TNC and CCACS-VNC, CACV-TNC and CCACV-VNC. Conclusion VNC images based on SDCT can accurately measure CACS and be used for cardiovascular risk classification, which is expected to replace TNC scan and reduce the radiation dose of patients.