重组卷积核联合高级迭代重组在优化CT冠状动脉支架成像中的价值

Reconstruction Convolution Kernel Combined with Advanced Iterative Reconstruction in Optimizing CT Coronary Stent Imaging

目的探讨重组卷积核联合高级迭代重组对改善冠状动脉CT血管造影(CCTA)支架图像质量的应用价值。方法搜集在第三代双源CT行CCTA扫描的冠状动脉支架术后患者37例,共52只支架,以3种重组卷积核(软核,Bv40;中核,Bv44;锐核,Bv49)联合5种高级迭代重组(ADMIRE 1~5)得到15种组合的冠状动脉支架图像。由两名诊断医师以双盲法对图像质量进行主观评价,按1~5分评分。以支架内腔的CT值、噪声、对比噪声比(CNR)作为图像质量的客观评价指标。分别对比分析15种组合间、具有可判读性(平均分≥3)的图像间以及优秀图像间(平均分≥4)的差异。结果主观评价中,15种组合间、具有可判读性的图像间差异有统计学意义(均P<0.001),但优秀图像间的差异无统计学意义(P=0.636)。客观评价中,随着迭代强度的增加图像噪声逐级降低,CNR逐级增高;随着卷积核值的增加,图像噪声逐级增加,CNR逐级降低。各组图像的CT值差异无统计学意义(P>0.05)。15种组合间、具有可判读性的图像间、优秀图像间噪声及CNR差异均有统计学意义(P均<0.001)。结论重组卷积核与高级迭代重组影响支架的显示情况。锐核可以明显提高图像的空间分辨率,高迭代强度能够降低图像噪声并提高CNR;Bv49+ADMIRE 5的组合获得的冠状动脉支架图像最优。

Objective To investigate the value of reconstruction kernel combined with advanced iterative reconstruction(ADMIRE)to improve the image quality of coronary CT angiography stents.Methods 37 patients with a total of 52 stents underwent CCTA examination after coronary stents implantation.All examinations were performed on a third generation dual source CT,with 3 reconstruction convolution kernel(soft,Bv40;medium,Bv44;sharp,Bv49)combined with 5 advanced iterative reconstructions(ADMIRE1~5)to get 15 types combined coronary stent images.Attenuation,image noise and contrast to noise ratios(CNR)were calculated.Image quality was assessed subjectively by two observers using five point Likert scales.The differences between 15 combinations,interpretable images,and excellent images were compared and analyzed.Results In subjective evaluation,the difference between 15 combinations and interpretable images(average score≥3)was statistically significant(P<0.001),but the difference between excellent images(average score≥4)was not statistically significant(P=0.636).In the objective evaluation,as the iteration intensity increases,the image noise decreases gradually,and the CNR increases gradually.As the convolution kernel value increases,the image noise increases gradually,and the CNR decreases gradually.There was no significant difference in CT values of images in each group(P>0.05).The differences in noise and contrast-to-noise ratios among 15 combinations,interpretable images,and excellent images were statistically significant(P<0.001).Conclusion The reconstruction convolution kernel and advanced iterative reconstruction could affect the image quality of coronary stent.The sharp core can improve the spatial resolution significantly,and the high iteration intensity can reduce the image noise and increase the CNR.The combination of Bv49+ADMIR5 can obtain the best coronary stent image.