探讨70 kV联合深度学习重建算法对大体重患者CCTA图像质量的影响  

Effect of 70 kV Combined with Deep Learning Reconstruction Algorithm on the Quality of CCTA Images in Overweight Patients

作　　者：朱丽娟 禹志鹏 沈云 石骁萌 哈若水[1] 杨利莉[1] 汪芳[1] ZHU Lijuan;YU Zhipeng;SHEN Yun;SHI Xiaomeng;HA Ruoshui;YANG Lili;WANG Fang(Medical Imaging Center,People's Hospital of Ningxia Hui Autonomous Region;GE(China)CT lmaging Research Center)

机构地区：[1]宁夏回族自治区人民医院医学影像中心,银川750000 [2]GE(中国)CT影像研究中心

出　　处：《中国医学计算机成像杂志》2022年第6期664-668,共5页Chinese Computed Medical Imaging

基　　金：宁夏回族自治区重点研发(一般)项目(2019BEG03046,2021BEG03092)。

摘　　要：目的:探讨70 kV联合深度学习重建算法(DLIR)对大体重患者冠状动脉CT血管成像(CCTA)图像质量的影响。方法:纳入2021年9月—2022年1月在我院采用Revolution Apex CT行低管电压(70 kV)行CCTA检查的患者96例,根据患者体重指数(BMI)平均分为2组,即标准体重组与大体重组。标准体重组图像采用基于多模型的自适应统计迭代重建-Veo (ASiR-V40%),大体重组图像分别进行ASiR-V40%及低、中、高三档深度学习图像重建(DLIR-L、DLIR-M、DLIR-H),对两组图像的冠状动脉主要节段进行主观评分,记录两组图像主动脉根部、各冠状动脉近段[包括右冠状动脉(RCA)、左主干(LMA)、左前降支(LAD)、左回旋支(LCX)]及竖脊肌的CT值及SD值以及有效辐射剂量,分析比较两组组内及组间不同重建算法的差异。结果:低管电压(70 kV) CCTA检查条件下,大体重组中DLIR重建图像噪声均低于常规临床检查的ASiR-V40%重建图像,信噪比(SNR)及对比噪声比(CNR)均高于ASiR-V40%重建图像,随着DLIR重建算法的降噪级别的递增,客观评分逐步提高,以DLIR-H为最著,差异均有统计学意义(均P<0.05);DLIR主观图像质量评价均明显高于ASiR-V40%重建图像(均P<0.05),DLIR-H主观图像评分最高。标准体重组和大体重组间ASiR-V40%重建图像主客观评分没有差异(P>0.05);两组有效辐射剂量(ED)分别为(1.36±0.42)mSv、(1.59±0.46)mSv,大体重组高于标准体重组,差异有统计学意义(P<0.05)。结论:低管电压(70 kV) CCTA检查条件下,大体重患者进行冠状动脉CT常规ASiR-V40%重建算法与标准体重者图像质量间未见差异,均可进行诊断;相较于ASiR-V40%重建算法,经DLIR重建可以进一步降低图像噪声,提升图像质量,提高诊断信心。Purpose: To investigate the effect of 70 kV combined with deep learning image reconstruction algorithm(DLIR) on coronary CT angiography(CCTA) image quality in overweight patients.Methods: A total of 96 patients who underwent low tube voltage(70 kV) CCTA examination with Revolution Apex CT in our hospital from September 2021 to January 2022 were selected. The patients were categorized into two groups according to their body mass index(BMI): standard weight group(control group) and overweight group(experimental group). The images of the control group were reconstructed by adaptive statistical iterative reconstruction-Veo(ASiR-V40%), and the images of the experimental group were reconstructed by ASiR-V40% and 3 levels of DLIR algorithm(DLIR-low, DLIR-medium, DLIR-high). The two groups of images were subjectively scored according to main segments of the coronary arteries, and the CT values and SD of the aortic root, proximal coronary segments, including right coronary artery(RCA), left main coronary artery(LMA), left anterior descending branch(LAD) and left circumflex branch(LCX), and erector spinae in the two groups of images were recorded. The differences of reconstruction algorithms within and between groups were analyzed and compared.Results: Under the condition of low tube voltage(70 kV), in the experimental group, the noise of the DLIR reconstructed images were lower than that of the ASiR-V40% reconstructed images of the routine clinical examination, and the signal-to-noise ratio(SNR) and contrast-to-noise ratio(CNR) were higher. In ASiR-V40% reconstructed images, with the increase of the noise reduction level of the DLIR reconstruction algorithm, the objective scores gradually improved, with DLIR-H being the most significant, and the differences were with statistical significance(all P<0.05);the subjective scores of DLIR images were significantly higher than those of ASiR-V40% reconstructed images(all P<0.05), and the subjective image score of DLIR-H was the highest among all DLIR images. There was no differen

关 键 词：肥胖 CT血管成像 深度学习 图像质量 

分 类 号：R445.3[医药卫生—影像医学与核医学]