Comparison of low‐pitch and respiratory‐averaged CT protocols for attenuation correction of cardiac PET studies

Abstract

PET/CT perfusion studies suffer from artifacts caused by misalignment of transmission and emission data due to contractile cardiac and respiratory motion. This study investigates whether substantial differences exist between two respiration-averaging approaches for attenuation correction (AC): low-pitch helical (HCT) and time-averaged CT (ACT). Fifty-four consecutive patients received paired HCT (0.45 pitch, 120 kVp, 76 mA, 24 x 1.2 mm collimated slice width, 1 s gantry rotation time, 4.93 mGy CTDI) and ACT (sequence mode: 6.1 s acq/bed, 80 kVp, 13 mA, 24 x 1.2 mm collimated slice width, 5.53 mGy CTDI) AC scans under free-breathing prior to Rb-82 rest/adenosine stress. Mismatch between the emission and paired transmission data was compared by calculating the volume of myocardial uptake overlying the left CT lung field. Data were then reconstructed with the CT AC scans and normalized to injected dose and bodyweight. Paired rest and paired stress PET images were reoriented identically along the short axis and sampled into a 17-segment polar map for comparison. The ratio of HCT-PET and ACT-PET polar maps at rest and stress was calculated and grouped by segment for all patients. 95% confidence intervals were calculated to compare changes in the polar map ratios between the two AC methods. No significant difference was observed between the HCT and ACT overlying volume in the rest or stress emission data. 68% of the patients presented visual respiratory artifacts in the HCT images compared to 32% in the ACT. That 23% of the ACT images presented with photon starvation artifacts and increasing BMI was a significant indicator for the occurrence of photon starvation in the ACT AC scans (p < 0.001). The ratio of the reconstructed PET polar segment data showed good agreement between AC methods with 95% confidence intervals ranging from 0.92 to 1.07 in the rest data and 0.93 to 1.07 in the stress data segments. Bias, calculated by averaging the polar segment ratios, showed 1% higher values in the ACT-PET rest reconstructions compared to the HCT-PET rest reconstructions and no measurable bias in the stress reconstructions. This study shows good agreement and negligible bias between low-pitch HCT and ACT protocols for attenuation correction of cardiac PET data.