Computed tomography calcium score scan for attenuation correction of N-13 ammonia cardiac positron emission tomography: effect of respiratory phase and registration method

Abstract

The use of coronary calcium scoring (CaScCT) for attenuation correction (AC) of (13)N-ammonia PET/CT studies (NH3) is still being debated. We compare standard ACCT to CaScCT using various respiratory phases and co-registration methods for AC. Forty-one patients underwent a stress/rest NH3. Standard ACCT scans and CaScCT acquired during inspiration (CaScCTinsp, 26 patients) or expiration (CaScCTexp, 15 patients) were used to correct PET data for photon attenuation. Resulting images were compared using Pearson's correlation and Bland-Altman (BA) limits of agreement (LA) on segmental relative and absolute coronary blood flow (CBF) using both manual and automatic co-registration methods (rigid-body and deformable). For relative perfusion, CaScCTexp correlates better than CaScCTinsp with ACCT when using manual co-registration (r = 0.870; P < 0.001 and r = 0.732; P < 0.001, respectively). Automatic co-registration provides the best correlation between CaScCTexp and ACCT for relative perfusion (r = 0.956; P < 0.001). Both CaScCTinsp and CaScCTexp yielded excellent correlations with ACCT for CBF when using manual co-registration (r = 0.918; P < 0.001; BA mean bias 0.05 ml/min/g; LA: -0.42 to +0.3 ml/min/g and r = 0.97; P < 0.001; BA mean bias 0.1 ml/min/g; LA: -0.65 to +0.5 ml/min/g, respectively). The use of CaScCTexp and deformable co-registration is best suited for AC to quantify relative perfusion and CBF enabling substantial radiation dose reduction.