像素闪耀算法在提高单源双能CT低单能量图像对
肝内小转移瘤显示的价值

doi:10.12117/jccmi.2018.11.005

摘要
图/表
参考文献(15)
相关文章 (15)

全文: PDF (0 KB) HTML (1 KB)
输出: BibTeX | EndNote (RIS)

摘要目的：探讨深度学习像素闪烁（Pixel shine，PS）（AlgoMedica，Inc，Sunnyvale，CA）算法提升单源双能CT平扫单能量图像显示肝内小转移瘤图像质量的价值，尤其提高低能量图像对病灶显示的价值。方法：回顾性搜集肝转移瘤的患者14例，共38个病灶。使用单源双能CT（GE Discovery HD 750，USA），行GSI扫描，对AW4.6工作站所得的平扫低单能量图像进行PS算法（A7模式）处理，得到PS前后平扫40～70 KeV单能量图像。由1名影像诊断医师在40～70 KeV单能量图像的同一层面（显示病灶的最大层面）、同一位置分别在肝转移瘤、正常肝实质勾画感兴趣区（ROI），测量并记录各ROI的CT值及SD值，计算PS前后各单能量图像肝转移瘤的SNR、肝转移瘤相对于正常肝实质的CNR以及PS后SNR和CNR较PS前增大的比率。采用Shapiro-Wilk正态分布检验检查数据正态性。比较PS前后40～70 KeV四组单能量图像中肝转移瘤和正常肝实质的CT值、肝转移瘤SNR及CNR，并且，将PS后40 KeV、50 KeV、60 KeV图像中的肝转移瘤SNR、CNR与PS前70 KeV图像相比较。其中符合正态分布的数据使用配对样本t检验比较其差异，余不符合正态分布则使用Wilcoxon符号秩和检验。采用非参数Friedman检验分别比较PS后40～70 KeV各单能量图像SNR和CNR增大比率的差异，若存在差异则采用Wilcoxon符号秩和检验进行两两比较。结果：PS前后40～70 KeV四组单能量图像中肝转移瘤、肝实质的CT值无统计学差异（P＞0.05），而PS后40～70 KeV四组单能量图像中肝转移瘤的SNR、CNR均显著大于PS前（P均为0.000）。与PS前70 KeV图像中肝转移瘤SNR、CNR相比，PS后40 KeV图像的肝转移瘤SNR小于PS前70 KeV图像，而PS后50 KeV、60 KeV图像的肝转移瘤SNR以及PS后40 KeV、50 KeV、60 KeV图像的肝转移瘤CNR均明显大于PS前70 KeV图像。Friedman检验显示PS后40～70 KeV单能量图像肝转移瘤SNR和CNR较相应PS前图像SNR和CNR增大的比率之间具有统计学差异（P=0.000，P=0.000），其中40 KeV与50 KeV之间的SNR和CNR的增大比率无统计学差异（P=0.255，P=0.429），40 KeV和50 KeV单能量图像PS后SNR、CNR增大比率均显著大于60 KeV、70 KeV图像。结论：PS算法可以明显提高单源双能CT平扫单能量图像对显示肝内小转移瘤灶的图像质量，并且PS算法在低单能量图像中对病灶显示方面具有更好的效果。

	服务

	把本文推荐给朋友
	加入我的书架
	加入引用管理器
	E-mail Alert
	RSS
	作者相关文章
	徐明哲1
	刘爱连1
	刘义军1
	刘静红1
	潘聚东2

关键词 ：肝肿瘤, 肿瘤转移, 体层摄影术, 螺旋计算机

Abstract：Objective: To explore the value of pixel shine(PS) algorithm to improve the quality of single-energy image, especially the low single-energy image by single-source dual-energy CT, and to improve the detection of small liver metastases. Methods: Fourteen patients, who underwent spectral CT imaging using spectrum imaging modality(GSI) were retrospectively enrolled with 38 lesions in this study. The PS algorithm(A7 mode) was applied to process the low single-energy images obtained from the AW4.6 workstation to obtain four groups of 40~70 KeV single energy images before and after calculation by PS. The ROIs were draw in the liver metastases and normal liver parenchyma respectively at the same location and on the same slice, showing the maximum area of the lesion, of 40~70 KeV single-energy images by a senior radiologist, then CT and SD values were measured and recorded. The SNRs of liver metastases, the CNRs of liver metastases relative to normal liver parenchyma and the ratios of SNRs and CNRs in images after calculation by PS relative to before were calculated. The Shapiro-Wilk normal distribution test was used to check the normality of data. The CT values of liver metastases and normal liver parenchyma, SNRs and CNRs of hepatic metastases were compared between four groups of 40~70 KeV single energy images before and after calculation by PS, then the SNRs and CNRs of hepatic metastases in 40 KeV, 50 KeV and 60 KeV images processed by PS algorithm were compared with those in 70 KeV images. Paired samples t-test was used to compare the data that fit the normal distribution, and the Wilcoxon signed-rank test was used to compare the data that didn’t conform to normal distribution. The nonparametric Friedman test was used to compare the differences of SNR and CNR increase ratios of 40~70 KeV single energy images processed by PS algorithm. If there was a difference, the Wilcoxon signed rank test was used for comparison. Results: There were no differences between the CT values of liver metastases and liver parenchyma in four groups of 40~70 KeV single energy images before and after calculation by PS(P＞0.05). The SNR and CNR of liver metastases in 40~70 KeV single energy images processed by PS algorithm were greater than those in images without PS calculation(all P=0.000). Compared with SNRs and CNRs of hepatic metastases in 70 KeV PS images, the SNRs of hepatic metastases in 40 KeV images processed by PS was less than that in 70 KeV images without PS calculation, while the SNRs of hepatic metastases in 50 KeV and 60 KeV images processed by PS and the CNRs of liver metastases in 40 KeV, 50 KeV and 60 KeV images processed by PS were significantly greater than that in 70 KeV images without PS calculation. Conclusion: PS algorithm can significantly improve quality of single-energy images to show the intrahepatic metastasis lesion, and the PS algorithm used in the low single energy image has better effect on the display of lesions.

Key words： Liver neoplasms Neoplasm metastasis Tomography, spiral computed

收稿日期: 2017-12-09

PACS:	Ｒ735.7
	Ｒ814.42

通讯作者: 刘爱连，大连医科大学附属第一医院放射科，116011。E-mail：cjr.liuailian@vip.163.com

作者简介: 徐明哲（1992-），女，河南平顶山人，在读硕士研究生。E-mail：501687323@qq.com

引用本文:

徐明哲1，刘爱连1，刘义军1，刘静红1，潘聚东2. 像素闪耀算法在提高单源双能CT低单能量图像对肝内小转移瘤显示的价值[J]. 中国临床医学影像杂志, 2018, 29(11): 782-787.
XU Ming-zhe1, LIU Ai-lian1, LIU Yi-jun1, LIU Jing-hong1, PAN Ju-dong2. The value of the pixel shine algorithm in improving the display of small hepatocellular metastases in low single energy image with single-source dual-energy CT. JOURNAL OF CHINA MEDICAL IMAGING, 2018, 29(11): 782-787.

链接本文:

http://www.jccmi.com.cn/CN/10.12117/jccmi.2018.11.005 或 http://www.jccmi.com.cn/CN/Y2018/V29/I11/782

[1]林晓珠，沈云，陈克敏． CT能谱成像的基本原理与临床应用研究进展[J]. 中华放射学杂志，2011，45（8）：798-800. [2]张龙敏，刘爱连，刘义军，等. 低浓度对比剂能谱CT单能量成像对提高门静脉图像质量的研究[J]. 放射学实践，2015，30（4）：360-363. [3]Ohana M, Jeung MY, Labani A, et al. Thoracic dual energy CT: acquisition protocols, current application and future developments[J]. Diagn Interv Imaging, 2014, 95(11): 1017-1026. [4]王贵生，高建华，赵帅，等. 肝脏增强扫描门静脉期能谱CT与传统多层螺旋CT辐射剂量和图像质量的比较[J]. 中华放射学杂志，2013，47（4）：340-343. [5]武明辉，孙明华，史大鹏，等. 能谱CT最佳CNR技术在提高布加综合征患者成像质量的应用价值[J]. 实用放射学杂志，2015，31（4）：659-662. [6]雷军强，马文婷，王寅中，等. 特异性肝胆对比剂钆塞酸二钠对肝转移瘤的诊断价值的Meta分析[J]. 中国循证医学杂志，2015，9（12）：1378-1386. [7]Mainenti PP, Mancini M, Mainolfi C, et al. Detection of colorectal liver metastases: Prospective comparison of contrast enhanced US, multidetector CT, PET/CT, and 1.5Tesla MR with extracellular and reticulo-endothelial cell specific contrast agents[J]. Abdom Imaging, 2010, 35(5): 511-521. [8]潘元威，周志刚，高剑波，等. 能谱CT智能匹配技术在不同体质量指数患者腹部CT扫描中的应用[J]. 临床放射学杂志，2017，36（7）：1034-1037. [9]刘卓，张诚. ASiR与ASiR-V重建算法对CT图像质量影响的体模研究[J]. 医疗卫生装备，2016，37（9）：103-105. [10]Lim K, Kwon H, Cho J, et al. Initial phantom study comparing image quality in computed tomography using adaptive statistical iterative reconstruction and new adaptive statistical iterative reconstruction V[J]. J Comput Assist Tomogr, 2015, 39(3): 443-448. [11]刘爱连，沈云．能谱CT临床应用图谱[M]．北京：人民军医出版社，2012：8-20. [12]Matsumoto K, Jinzaki M, Tanami Y, et al. Virtual monochromatic spectral imaging with fast kilovoltage switching: improved image quality as compared with that obtained with conventional 120-kVp CT[J]. Radiology, 2011, 259(1): 257. [13]Geoffrey H, Simon O, Max W, et al. Unsupervised discovery of nonlinear structure using contrastive back propagation[J]. Science, 2006, 30(4): 725-732. [14]张琪. 人工智能的发展及其在医学领域中的应用[J]. 电子技术与软件工程，2016，8（20）：259. [15]王弈，李传富. 人工智能方法在医学图像处理中的研究新进展[J]. 中国医学物理学杂志，2013，30（3）：4138-4143.