摘要
目的与常规重建算法对比,探讨深度学习图像重建(DLIR)算法在提高胸腹主动脉CTA图像质量中的临床应用价值。方法回顾性纳入52例疑诊胸腹主动脉疾病的患者,重建出滤波反投影重建(FBP),自适应统计迭代重建(ASIR-V50%、ASIR-V80%),深度学习低(DLIR-L)、中(DLIR-M)、高(DLIR-H),共6种0.625 mm的薄层CT图像。通过对比图像CT值、噪声值、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。并由两位放射诊断科具有5年、11年工作经验的影像医师采用双盲法对图像质量进行主观评价(Liker5分制)。结果六组重建所得图像的信噪比、对比噪声比从高到低依次是:DLIR-H、DLIR-M、ASIR-V80%、DLIR-L、ASIR-V50%、FBP,差异具有统计学意义(P<0.05),重建图像主观评分差异有统计学意义(P<0.05),经DLIR-H处理后的图像噪声最低、整体图像质量最佳。结论与传统的FBP、ASIR-V重建算法相比,基于深度学习图像重建算法DLIR能显著降低噪声,提高胸腹主动脉CTA图像质量,在临床应用方面,新的DLIR算法具有很大的应用潜力。
Objiective Purpose To evaluate the deep learning image reconstruction(DLIR)algorithm in improving the image quality of thoracic and abdominal aorta CTA,compared with Adaptive Statistical iterative Reconstruction-V(ASIR-V)and FBP.Methods 52 patients who underwent thoracic and abdominal aorta CTA with the same scan conditions were included retrospectively.Filtered backprojection reconstruction(FBP),adaptive statistical iterative reconstruction(ASIRV50%,ASIRV80%),low deep learning(DLIRL),moderate(DLIRM)and high(DLIRH)thin CT images of 6 kinds of 0.625 mm were reconstructed.Objective evaluation of different image reconstruction algorithms was carried out by comparing CT value,noise value,SNR and CNR.Two radiologists with 5 years and 11 years of experience in the department of radiology were selected to evaluate the image quality(Liker 5-point system).Results The signal to noise ratio and contrast noise ratio of the reconstructed images of the six groups from high to low were:DLIR H,DLIR M,ASIR-V80%,DLIR L,ASIR-V50%,FBP,the differences were statistically significant(P<0.05),and the subjective scores of reconstructed images were statistically significant(P<0.05).The value of DLIR H had the lowest noise and the best overall image quality.Conclusion Compared with FBP and ASIR V,DLIR algorithm can significantly reduce noises and improve the image quality of thoracic and abdominal aorta CTA.Therefore,DLIR algorithm has great application potential in clinical application.
作者
陆晓军
罗春材
齐叶青
杨铁
LU Xiaojun;LUO Chuncai;QI Yeqing(Department of Radiology,The First Medical Center of PLA General Hospital,Beijing 100853,P.R.China)
出处
《临床放射学杂志》
北大核心
2022年第7期1359-1364,共6页
Journal of Clinical Radiology
关键词
胸腹主动脉
图像质量
深度学习图像重建
迭代重建
CT血管成像
thoracic and abdominal aorta
image quality
deep learning image reconstruction
iterative reconstruction
computed tomography angiography
