Yuki Hideaki, Utsunomiya Daisuke, Funama Yoshinori, Tokuyasu Shinichi, Namimoto Tomohiro, Hirai Toshinori, Itatani Ryo, Katahira Kazuhiro, Oshima Shuichi, Yamashita Yasuyuki
Diagnostic Radiology, Faculty of Life Sciences, Kumamoto University, 1-1-1, Honjo, Chuo-ku, Kumamoto-shi, Kumamoto, 860-8556, Japan.
Diagnostic Radiology, Faculty of Life Sciences, Kumamoto University, 1-1-1, Honjo, Chuo-ku, Kumamoto-shi, Kumamoto, 860-8556, Japan.
J Cardiovasc Comput Tomogr. 2014 Mar-Apr;8(2):115-23. doi: 10.1016/j.jcct.2013.12.010. Epub 2014 Jan 12.
Most current iterative reconstruction algorithms for CT imaging are a mixture of iterative reconstruction and filtered back projection. The value of "fully" iterative reconstruction in coronary CT angiography remains poorly understood.
We aimed to assess the value of the knowledge-based iterative model reconstruction (IMR) algorithm on the qualitative and quantitative image quality at 256-slice cardiac CT.
We enrolled 21 patients (mean age: 69 ± 11 years) who underwent retrospectively ECG gated coronary CT anhgiography at 100 kVp tube voltage. Images were reconstructed with the filtered back projection (FBP), hybrid iterative reconstruction (IR), and IMR algorithms. CT attenuation and the contrast-to-noise ratio (CNR) of the coronary arteries were calculated. With the use of a 4-point scale, 2 reviewers visually evaluated the coronary arteries and cardiac structures.
The mean CT attenuation of the proximal coronary arteries was 369.3 ± 73.6 HU, 363.9 ± 75.3 HU, and 363.3 ± 74.5 HU, respectively, for FBP, hybrid IR, and IMR and was not significantly different. The image noise of the proximal coronary arteries was significantly lower with IMR (11.3 ± 2.8 HU) than FBP (51.9 ± 12.9 HU) and hybrid IR (23.2 ± 5.2 HU). The mean CNR of the proximal coronary arteries was 9.4 ± 2.4, 20.2 ± 4.7, and 41.8 ± 9.5 with FBP, hybrid IR and IMR, respectively; it was significantly higher with IMR. The best subjective image quality for coronary vessels was obtained with IMR (proximal vessels: FBP, 2.6 ± 0.5; hybrid IR, 3.4 ± 0.5; IMR, 3.8 ± 0.4; distal vessels: FBP, 2.3 ± 0.5; hybrid IR. 3.1 ± 0.5; IMR, 3.7 ± 0.5). IMR also yielded the best visualization for cardiac systems, that is myocardium and heart valves.
The novel knowledge-based IMR algorithm yields significantly improved CNR and better subjective image quality of coronary vessels and cardiac systems with reliable CT number measurements for cardiac CT imaging.
目前大多数用于CT成像的迭代重建算法都是迭代重建与滤波反投影的混合算法。在冠状动脉CT血管造影中,“完全”迭代重建的价值仍知之甚少。
我们旨在评估基于知识的迭代模型重建(IMR)算法在256层心脏CT中对图像质量定性和定量方面的价值。
我们纳入了21例患者(平均年龄:69±11岁),这些患者在100 kVp管电压下接受了回顾性心电门控冠状动脉CT血管造影。图像分别采用滤波反投影(FBP)、混合迭代重建(IR)和IMR算法进行重建。计算冠状动脉的CT衰减值和对比噪声比(CNR)。两名阅片者使用4分制对冠状动脉和心脏结构进行视觉评估。
FBP、混合IR和IMR重建的冠状动脉近端平均CT衰减值分别为369.3±73.6 HU、363.9±75.3 HU和363.3±74.5 HU,差异无统计学意义。IMR重建的冠状动脉近端图像噪声(11.3±2.8 HU)显著低于FBP(51.9±12.9 HU)和混合IR(23.2±5.2 HU)。FBP、混合IR和IMR重建的冠状动脉近端平均CNR分别为9.4±2.4、20.2±4.7和41.8±9.5;IMR的CNR显著更高。IMR获得的冠状动脉主观图像质量最佳(近端血管:FBP,2.6±0.5;混合IR,3.4±0.5;IMR,3.8±0.4;远端血管:FBP,2.3±0.5;混合IR,3.1±0.5;IMR,3.7±0.5)。IMR对心脏系统(即心肌和心脏瓣膜)的显示也最佳。
基于知识的新型IMR算法在心脏CT成像中能显著提高CNR,改善冠状动脉和心脏系统的主观图像质量,并能获得可靠的CT数值测量结果。