非线性超声射频信号熵对乳腺结节良恶性的定征

张玫玫; 高凡; 屠娟; 吴意赟; 章东

doi:10.7498/aps.70.20201919

摘要

本文提出了一种基于非线性超声射频(radio frequency, RF)信号熵对乳腺结节良恶性进行定征的方法. 对306例乳腺结节样本(良性158例, 恶性148例)提取了基于超声RF信号二次谐波的熵和加权熵, 以及常规超声参数(图像灰度、纵横比、不规则度、乳腺结节大小、深度); 采用t检验和线性分类器检测参数对乳腺结节良恶性的区分度; 进一步将有效参数组合输入支持向量机对乳腺结节良恶性进行分类. 结果表明: 除图像灰度外, 其余参数均在乳腺结节的良性与恶性间有显著差异. 多参数结合输入支持向量机的良恶性分类的准确率、敏感性和特异性分别为81.4%, 78.4%和84.2%. 本文工作表明非线性超声RF信号的熵可有效地定征乳腺结节的良恶性, 有望成为乳腺结节良恶性定征新参量.

关键词:

熵 /
加权熵 /
乳腺超声 /
组织定征 /
非线性

Abstract

In this paper the classification of benign and malignant breast masses is investigated by using the entropy of nonlinear ultrasound radio frequency (RF) signal. The parameters (entropy and weighted entropy) derived from the nonlinear ultrasound RF signal and the conventional ultrasound parameters (image grayscale, aspect ratio, irregularity, breast mass size, and depth) are extracted from 306 image samples (158 benign and 148 malignant); t-test and linear-discriminant classifier (LDC) are used to test the distinction between benign and malignant breast masses by each parameter; furthermore the effective parameters are combined to classify benign and malignant breast masses. The results show that except the image grayscale, the other parameters are significantly different between benign and malignant breast masses. Multi-parameter combined with support vector machine (SVM) is used to classify breast masses as benign and malignant. The accuracy is 81.4%, the sensitivity is 78.4%, and the specificity is 84.2%. The present work shows that the combination of the nonlinear entropy of ultrasound RF signal and traditional ultrasound parameters can more effectively characterize the benign and malignant breast masses. The entropy of nonlinear ultrasound RF signal can become a new parameter for characterizing the benign and malignant breast masses.

Keywords:

作者及机构信息

1.
南京大学物理学院，近代声学教育部重点实验室，南京　210093

2.
江苏省中医院超声科，南京　210029

通信作者: 章东, dzhang@nju.edu.cn

基金项目: 江苏省重点研发计划(批准号: BE2018703)和湖南省战略性新兴产业科技攻关与重大成果转化项目(批准号: 2019GK4046)资助的课题

Authors and contacts

1.
Key Laboratory of Modern Acoustics of the Ministry of Education, School of Physics, Nanjing University, Nanjing 210093, China

2.
Department of Ultrasound, Jiangsu Provincial Hospital of Chinese Medicine, Nanjing 210029, China

Corresponding author: Zhang Dong, dzhang@nju.edu.cn

Funds: Project supported by the Jiangsu Provincial Key R&D Program, China (Grant No. BE2018703) and the Strategic Emerging Industries Science and Technology and Major Achievements Transformation Project of Hunan Province, China (Grant No. 2019GK4046)

文章全文 : translate this paragraph

参考文献

[1]	Jia M, Zheng R, Zhang S, Zeng H, Zou X, Chen W 2015 J. Thorac. Dis. 7 1221
[2]	Masafumi K, Hiroyuki T 2007 Breast Cancer-Tokyo 14 342 Google Scholar
[3]	Tahoces G P, Correa J, Souto M, Gonzalez C, Gomez L 1991 IEEE Trans. Med. Imaging 30 330
[4]	Boone J M, Nelson T R, Lindfors K K, Seibert J A 2001 Radiology 221 657 Google Scholar
[5]	Ji D J, Qu G R, Hu C H, Liu B D, Jian J B, Guo X K 2017 Chin. Phys. B 26 0607018
[6]	Kuhl C K 2000 Eur. Radiol. 10 46 Google Scholar
[7]	方晟, 吴文川, 应葵, 郭华 2013 62 048702 Google Scholar Fang S, Wu W C, Ying K, Guo H 2013 Acta Phys. Sin. 62 048702 Google Scholar
[8]	Mendelson B E, Marcela B V, Berg A W 2013 ACR BI-RADS® Atlas-Breast Ultrasound (Reston: American College of Radiology) pp35−100
[9]	Li J W, Tong Y Y, Zhou J, Shi Z T, Sun P X, Chang C 2020 J. Ultrasound Med. 39 1589 Google Scholar
[10]	Wojcinski S, Stefanidou N, Hillemanns P, Degenhardt F 2013 Bmc Womens Health 13 47 Google Scholar
[11]	Chang Y W, Chen Y R, Ko C C, Lin W Y, Lin K P 2020 Appl. Sci. 10 1830 Google Scholar
[12]	Koundal D, Gupta S, Singh S 2018 Biomed. Signal Proces. Control 40 117 Google Scholar
[13]	Burckhardt C B 1978 IEEE Trans. Sonics Ultrason. 25 1 Google Scholar
[14]	类成新, 吴振森 2010 59 5692 Google Scholar Lei C X, Wu Z S 2010 Acta Phys. Sin. 59 5692 Google Scholar
[15]	Wagner R F, Insana M F, Brown D G 1987 J. Opt. Soc. Am. A: 4 910
[16]	Weng L, Reid J M, Shankar P M, Soetanto K 1991 J. Acoust. Soc. Am. 89 2992 Google Scholar
[17]	Karmeshu, Agrawal R 2006 Ultrasound Med. Biol. 32 371 Google Scholar
[18]	Tsui P H 2015 Entropy 17 6598 Google Scholar
[19]	Tsui P H, Wan Y L 2016 Entropy 18 341 Google Scholar
[20]	Liu C, Xie L, Kong W, Lu X, Zhang D, Wu M, Zhang L, Yang B 2019 Ultrasonics 99 105951 Google Scholar
[21]	Zhang D, Gong X F 1999 Ultrasound Med. Biol. 25 593 Google Scholar
[22]	Gong X F, Zhang D, Liu J H, Wang H L, Yan Y S, Xu X C 2004 J. Acoust. Soc. Am. 116 1819 Google Scholar
[23]	Cortes C, Vapnik V 1995 Machine Learning 20 273
[24]	Chang C C, Lin C J 2011 Acm T. Intel. Syst. Tec. 2 1
[25]	行鸿彦, 金天力 2010 59 140 Google Scholar Xing H Y, Jin T L 2010 Acta Phys. Sin. 59 140 Google Scholar
[26]	Shannon C E 1948 Bell Syst. Tech. J. 27 379 Google Scholar
[27]	Guiasu S 1986 J. Stat. Plan. Infer. 15 63 Google Scholar
[28]	Tranquart F, Grenier N, Eder V, Pourcelot L 1999 Ultrasound Med. Biol. 25 889 Google Scholar
[29]	Ward B, Baker A C, Humphrey V F 1997 J. Acoust. Soc. Am. 101 143 Google Scholar
[30]	Rosen E L, Soo M S 2001 Clin. Imag. 25 379 Google Scholar
[31]	周志华 2016 机器学习 (北京: 清华大学出版社) pp121−140 Zhou Z H 2016 Machine Learning (Beijing: Tsinghua University Press) pp121−140 (in Chinese)
[32]	Box J F 1987 Stat. Sci. 2 45
[33]	Shan J, Alam S K, Garra B, Zhang Y, Ahmed T 2016 Ultrasound Med. Biol. 42 980 Google Scholar
[34]	Yap M H, Pons G, Marti J, Ganau S, Sentis M, Zwiggelaar R, Davison A K, Marti R, Moi Hoon Y, Pons G, Marti J, Ganau S, Sentis M, Zwiggelaar R, Davison A K, Marti R 2018 IEEE J. Biomed. Health Inform. 22 1218 Google Scholar

施引文献

图 1 (a) 超声RF灰度图像和ROI选取示意图; (b) RF信号线的波形图

Fig. 1. (a) Illustration of ultrasound RF gray imaging and selection of ROI; (b) waveforms of RF signals.

下载: 全尺寸图片幻灯片

图 2 熵及加权熵成像过程示意图

Fig. 2. Illustration of entropy and weighted entropy imaging.

下载: 全尺寸图片幻灯片

图 3 某恶性结节表现　(a) 病理; (b) RF灰度图; (c) 熵图; (d) 加权熵图

Fig. 3. Presentations of a typical malignant mass: (a) Micrograph; (b) RF grayscale image; (c) entropy image; (d) weighted entropy image.

下载: 全尺寸图片幻灯片

图 4 某良性结节表现　(a) 病理; (b) RF灰度图; (c) 熵图; (d)加权熵图

Fig. 4. Presentations of a typical benign mass: (a) Micrograph; (b) RF grayscale image; (c) entropy image; (d) weighted entropy image

下载: 全尺寸图片幻灯片

图 5 不同参数组合的ROC曲线

Fig. 5. ROC curves with various input parameter combinations.

下载: 全尺寸图片幻灯片

图 6 不同滑动窗口大小重构的熵图　(a) 0.2倍脉冲长度; (b) 0.5倍脉冲长度; (c) 1.0倍脉冲长度; (d) 2.0倍脉冲长度

Fig. 6. The reconstructed entropy images with various window sizes: (a) 0.2 times pulse length; (b) 0.5 times pulse length; (c) 1.0 pulse length; (d) 2.0 times pulse length.

下载: 全尺寸图片幻灯片

图 7 306例乳腺RF信号的平均熵值随窗口大小的变化

Fig. 7. Dependence of averaged entropy on window size for 306 samples.

下载: 全尺寸图片幻灯片

表 1 特征参数的分布

Table 1. Distribution of various parameters.

参数	平均值 ± 标准差		双样本t检验(p < 0.05)		线性分类器(LDC)
参数	良性	恶性	差异性	p值	AUC	准确率/%
图像灰度	5.01 ± 0.97	4.98 ± 0.93	否	0.79
纵横比	0.78 ± 0.37	1.06 ± 0.36	是	6.2 × 10^–12	0.75	70.6
不规则度	2.82 ± 1.07	3.45 ± 1.11	是	1.4 × 10^–7	0.64	63.7
深度/mm	16.72 ± 5.29	20.91 ± 5.49	是	4.1 × 10^–13	0.72	66.7
大小/mm²	114 ± 142	171 ± 172	是	4.5 × 10^–4	0.59	58.2
熵	4.64 ± 0.40	4.87 ± 0.15	是	6.7 × 10^–12	0.75	72.5
加权熵	1.69 ± 0.13	1.76 ± 0.05	是	1.7 × 10^–11	0.74	69.0

下载: 导出CSV

map

[1]	Jia M, Zheng R, Zhang S, Zeng H, Zou X, Chen W 2015 J. Thorac. Dis. 7 1221
[2]	Masafumi K, Hiroyuki T 2007 Breast Cancer-Tokyo 14 342 Google Scholar
[3]	Tahoces G P, Correa J, Souto M, Gonzalez C, Gomez L 1991 IEEE Trans. Med. Imaging 30 330
[4]	Boone J M, Nelson T R, Lindfors K K, Seibert J A 2001 Radiology 221 657 Google Scholar
[5]	Ji D J, Qu G R, Hu C H, Liu B D, Jian J B, Guo X K 2017 Chin. Phys. B 26 0607018
[6]	Kuhl C K 2000 Eur. Radiol. 10 46 Google Scholar
[7]	方晟, 吴文川, 应葵, 郭华 2013 62 048702 Google Scholar Fang S, Wu W C, Ying K, Guo H 2013 Acta Phys. Sin. 62 048702 Google Scholar
[8]	Mendelson B E, Marcela B V, Berg A W 2013 ACR BI-RADS® Atlas-Breast Ultrasound (Reston: American College of Radiology) pp35−100
[9]	Li J W, Tong Y Y, Zhou J, Shi Z T, Sun P X, Chang C 2020 J. Ultrasound Med. 39 1589 Google Scholar
[10]	Wojcinski S, Stefanidou N, Hillemanns P, Degenhardt F 2013 Bmc Womens Health 13 47 Google Scholar
[11]	Chang Y W, Chen Y R, Ko C C, Lin W Y, Lin K P 2020 Appl. Sci. 10 1830 Google Scholar
[12]	Koundal D, Gupta S, Singh S 2018 Biomed. Signal Proces. Control 40 117 Google Scholar
[13]	Burckhardt C B 1978 IEEE Trans. Sonics Ultrason. 25 1 Google Scholar
[14]	类成新, 吴振森 2010 59 5692 Google Scholar Lei C X, Wu Z S 2010 Acta Phys. Sin. 59 5692 Google Scholar
[15]	Wagner R F, Insana M F, Brown D G 1987 J. Opt. Soc. Am. A: 4 910
[16]	Weng L, Reid J M, Shankar P M, Soetanto K 1991 J. Acoust. Soc. Am. 89 2992 Google Scholar
[17]	Karmeshu, Agrawal R 2006 Ultrasound Med. Biol. 32 371 Google Scholar
[18]	Tsui P H 2015 Entropy 17 6598 Google Scholar
[19]	Tsui P H, Wan Y L 2016 Entropy 18 341 Google Scholar
[20]	Liu C, Xie L, Kong W, Lu X, Zhang D, Wu M, Zhang L, Yang B 2019 Ultrasonics 99 105951 Google Scholar
[21]	Zhang D, Gong X F 1999 Ultrasound Med. Biol. 25 593 Google Scholar
[22]	Gong X F, Zhang D, Liu J H, Wang H L, Yan Y S, Xu X C 2004 J. Acoust. Soc. Am. 116 1819 Google Scholar
[23]	Cortes C, Vapnik V 1995 Machine Learning 20 273
[24]	Chang C C, Lin C J 2011 Acm T. Intel. Syst. Tec. 2 1
[25]	行鸿彦, 金天力 2010 59 140 Google Scholar Xing H Y, Jin T L 2010 Acta Phys. Sin. 59 140 Google Scholar
[26]	Shannon C E 1948 Bell Syst. Tech. J. 27 379 Google Scholar
[27]	Guiasu S 1986 J. Stat. Plan. Infer. 15 63 Google Scholar
[28]	Tranquart F, Grenier N, Eder V, Pourcelot L 1999 Ultrasound Med. Biol. 25 889 Google Scholar
[29]	Ward B, Baker A C, Humphrey V F 1997 J. Acoust. Soc. Am. 101 143 Google Scholar
[30]	Rosen E L, Soo M S 2001 Clin. Imag. 25 379 Google Scholar
[31]	周志华 2016 机器学习 (北京: 清华大学出版社) pp121−140 Zhou Z H 2016 Machine Learning (Beijing: Tsinghua University Press) pp121−140 (in Chinese)
[32]	Box J F 1987 Stat. Sci. 2 45
[33]	Shan J, Alam S K, Garra B, Zhang Y, Ahmed T 2016 Ultrasound Med. Biol. 42 980 Google Scholar
[34]	Yap M H, Pons G, Marti J, Ganau S, Sentis M, Zwiggelaar R, Davison A K, Marti R, Moi Hoon Y, Pons G, Marti J, Ganau S, Sentis M, Zwiggelaar R, Davison A K, Marti R 2018 IEEE J. Biomed. Health Inform. 22 1218 Google Scholar

[1]	李蕊轩, 张勇. 熵在非晶材料合成中的作用. , 2017, 66(17): 177101. doi: 10.7498/aps.66.177101
[2]	徐红梅, 金永镐, 郭树旭. 电压控制不连续导电模式DC-DC变换器的熵特性研究. , 2013, 62(24): 248401. doi: 10.7498/aps.62.248401
[3]	潘欣裕, 赵鹤鸣. Logistic混沌系统的熵特性研究. , 2012, 61(20): 200504. doi: 10.7498/aps.61.200504
[4]	冯维, 丁辉, 林昊, 罗辽复. λ噬菌体溶源/裂解转换调控与定态熵. , 2012, 61(16): 168701. doi: 10.7498/aps.61.168701
[5]	杨波. 一般加速带电带磁的动态黑洞中标量场的熵. , 2008, 57(4): 2614-2620. doi: 10.7498/aps.57.2614
[6]	韩亦文, 洪云, 杨树政. 广义不确定关系与整体单极黑洞Dirac场的熵. , 2007, 56(1): 10-14. doi: 10.7498/aps.56.10
[7]	杨波. 变加速直线运动黑洞的温度和Dirac场的熵. , 2007, 56(11): 6772-6776. doi: 10.7498/aps.56.6772
[8]	郑元强. 球对称动态黑洞Dirac场的熵的再讨论. , 2007, 56(3): 1266-1270. doi: 10.7498/aps.56.1266
[9]	郑元强. 动态广义球对称含荷黑洞Dirac场的熵. , 2006, 55(7): 3272-3276. doi: 10.7498/aps.55.3272
[10]	牛振风, 刘文彪. 新Tortoise坐标变换与任意加速带电动态黑洞熵. , 2005, 54(1): 475-480. doi: 10.7498/aps.54.475
[11]	韩亦文, 洪　云. Schwarzschild-de-Sitter黑洞宇宙视界量子态的熵. , 2004, 53(10): 3270-3273. doi: 10.7498/aps.53.3270
[12]	孙学锋, 景　玲, 刘文彪. 黑洞熵无截断薄层模型的改进与推广. , 2004, 53(11): 4002-4006. doi: 10.7498/aps.53.4002
[13]	强丽娥, 高新芹, 赵　峥. 动态黑洞温度和熵的再讨论. , 2004, 53(10): 3619-3626. doi: 10.7498/aps.53.3619
[14]	王波波. 环面黑洞背景下量子场的熵. , 2004, 53(7): 2401-2406. doi: 10.7498/aps.53.2401
[15]	张靖仪. 一般球对称带电蒸发黑洞Dirac场的熵. , 2003, 52(9): 2354-2358. doi: 10.7498/aps.52.2354
[16]	孙鸣超. 起源于引力场的Vaidya-Bonner-de Sitter黑洞的量子熵. , 2003, 52(6): 1350-1353. doi: 10.7498/aps.52.1350
[17]	宋太平, 侯晨霞, 黄金书. 一般球对称带电蒸发黑洞的熵. , 2002, 51(8): 1901-1906. doi: 10.7498/aps.51.1901
[18]	张靖仪, 赵峥. 直线加速动态黑洞Dirac场的熵. , 2002, 51(10): 2399-2406. doi: 10.7498/aps.51.2399
[19]	宋太平, 侯晨霞, 史旺林. Vaidya-Bonner黑洞的熵. , 2002, 51(6): 1398-1402. doi: 10.7498/aps.51.1398
[20]	魏志勇, 段利敏, 吴和宇, 靳根明, 李祖玉, 诸永泰, 郗洪飞, 沈文庆, 肖志刚, 王宏伟, 张保国, 柳永英, 王素芳, 胡荣江. 35MeV/u40Ar+197Au中的熵产生. , 2001, 50(4): 649-654. doi: 10.7498/aps.50.649

计量

文章访问数: 5691
PDF下载量: 109
被引次数: 0

姓名
邮箱
手机号码
标题
留言内容
验证码

搜索

留言板

非线性超声射频信号熵对乳腺结节良恶性的定征