2017 年 73 巻 11 号 p. 1140-1146
Background: Invasive-fractional flow reserve (FFR) is the reference standard to evaluate functional ischemia of coronary arteries, and is used to decide if percutaneous transluminal coronary angioplasty is necessary. Recently, computed tomography-derived FFR (CT-FFR) is emerged as an alternative non-invasive method. Objectives: To evaluate the effect of reconstruction methods and image parameters on the accuracy of CT-FFR calculation. Methods: A total of 26 segments in the consecutive 10 coronary CT angiography (CCTA) studies were evaluated. All studies were reconstructed using three different techniques: 1) filtered back projection (FBP), 2) adaptive iterative dose reduction 3D (AIDR 3D), and 3) forward projected model-based iterative reconstruction solution (FIRST). Vessel segmentation was performed automatically by CT-FFR software, with manual adjustment if necessary. Calculated CT-FFR was compared with the invasive FFR data. Results: Compared to FBP, AIDR 3D and FIRST resulted in more successful automatic segmentation. When using FIRST, 7 segments (27%) were completed without manual adjustment. These segments had relatively larger vessel diameter, higher CT number, and lower noise. The difference between the calculated CT-FFR and invasive-FFR was 0.02±0.01. Among the remaining, 10 segments (38%) required manual adjustments of centerline, 7 segments (27%) required manual adjustments of contour, and 2 segments (8%) did not reach to the CT-FFR calculation. Conclusion: AIDR 3D and FIRST were useful for reliable automatic segmentation and analysis of CT-FFR.
