Temporal Averaging Method for Higher-Quality Cerebrovascular 4D-CTA

The purpose of this study was to propose a temporal averaging (TA) method for acquiring higher-quality cerebrovascular 4D-CTA. This paper investigated whether the TA method could improve image quality for 4D-CTA. CT images were acquired at 80 kV and 80 mA. CTA source images (CTA-SI) were obtained at 1-s intervals in the arterial phase. Three 4D-CTA data were obtained using a double dose (pseudo 160 mA) with filtered back projection (FBP) and standard dose (actual 80 mA) with FBP or hybrid iterative reconstruction (HIR). Pseudo 160 mA data were generated from actual 80 mA data (CTA-SI) using the TA method. We defined pseudo 160 mA data as TA-FBP, actual 80 mA data with FBP as LD-FBP, and actual 80 mA data with HIR as LD-HIR. CTA-SI of 14 patients were collected for this study. Objective image quality was assessed based on image noise and the time-enhancement curve (TEC). Subjective image quality was also investigated by observer performance testing. In the objective assessment, TA-FBP reduced image noise by 26.3% compared with LD-FBP, while TECs of TA-FBP and LD-FBP were quite similar. According to the observer performance test, TA-FBP significantly improved the visual conspicuity of the cortical vessel as compared with the other two methods. These results indicated that the TA method was able to improve image quality from 4D-CTA. The TA method proved useful for the acquisition of more appropriate 4D-CTA for diagnosis.