深度学习图像重建算法在提高CCTA图像质量中的临床价值研究

Clinical Application Value of Deep Learning Image Reconstruction Algorithm in Improving CCTA Image Quality

目的对比常规迭代重建算法,评价深度学习图像重建(DLIR)算法(TrueFidelity^(TM))在提高冠状动脉CT血管造影(CCTA)图像质量中的临床应用价值。方法回顾性纳入40例接受相同扫描条件CCTA检查的患者。用ASiR-V 0%、ASiR-V 50%、ASiR-V 80%和DLIR两个水平(中[M]和高[H])重建CCTA数据。通过对比图像噪声(SD)、信噪比(SNR)、对比度噪声比(CNR)进行不同图像重建算法间的客观评价。由两位高年资影像医师双盲对主干血管图像质量进行主观评价(Liker 4分制)。结果ASiR-V 0%、ASiR-V 50%、ASiR-V 80%、DLIR-M、DLIR-H重建图像的整体对比图像噪声逐渐降低,差异具有统计学意义(P<0.05),经DLIR-H处理后的图像噪声明显低于其他处理方式(P<0.05),SNR和CNR值逐渐升高,差异具有统计学意义(P<0.01),DLIR-H重建图像的SNR和CNR均优于其他重建图像(P<0.05)。ASiR-V 0%、ASiR-V 50%、ASiR-V 80%、DLIR-M、DLIR-H算法在冠状动脉主要分支的主观评分均无显著性差异(P>0.05)。结论与ASiR-V算法相比,DLIR算法能显著降低噪声,提高CCTA的SNR和CNR。因此在提高CCTA图像质量的临床应用方面,DLIR算法具有较大的应用潜力。

Objective To evaluate the deep learning image reconstruction(DLIR)algorithm(Truefidelity^(TM))to improve the image quality of coronary CT angiography(CCTA),compared Adaptive Statistical iterative Reconstruction-V(ASiR-V).Methods Forty patients who underwent CCTA with the same scan conditions were included retrospectively.The CCTA data were reconstructed with ASiR-V 0%、ASiR-V 50%,ASiR-V 80%and DLIR at two levels(medium[M]and high[M]).The objective evaluation of image is carried out by imaging noise(IN),signal to noise ratio(SNR)and contrast to noise ratio(CNR).Two senior radiologists were selected to evaluate the image quality(Liker 4-point system).Results Noise of images reconstructed with ASiR-V 0%,ASiR-V 50%,ASiR-V 80%,DLIR-M,DLIR-H were gradually decreased with significant difference(P<0.05).The value of DLIR-H was significantly lower than the other image noise(P<0.05).SNR and CNR values were gradually increasing.The difference was statistically significant(P<0.01).SNR and CNR of image reconstructed with DLIR-H were better than other reconstructed images(P<0.05).There was no significant difference in the subjective scores of ASiR-V 0%,ASiR-V 50%,ASiR-V 80%,DLIR-M and DLIR-H algorithms in the main coronary artery branches(P>0.05).Conclusion Compared with ASiR-V,DLIR algorithm can significantly reduce the noise and improve the SNR and CNR of CCTA.Therefore,DLIR algorithm has great potential in clinical application of improving CCTA image quality.