基于变分模态分解与多尺度排列熵的生物组织变性识别
|
摘要
根据高强度聚焦超声(HIFU)治疗中超声散射回波信号的特点,本文利用变分模态分解(VMD)与多尺度排列熵(MPE)对生物组织变性识别进行了研究.首先对生物组织中的超声散射回波信号进行变分模态分解,根据各阶模态的功率谱信息熵值分离出噪声分量和有用分量;对分离出的有用信号进行重构并提取其多尺度排列熵;然后通过Gustafson-Kessel (GK)模糊聚类确定聚类中心,采用欧氏贴近度与择近原则对生物组织进行变性识别.将所提方法应用于HIFU治疗中超声散射回波信号实验数据,用遗传算法对多尺度排列熵的参数优化后,对293例未变性组织和变性组织的超声散射回波信号数据进行了多尺度排列熵分析,发现变性组织的超声散射回波信号的多尺度排列熵值要高于未变性组织;多尺度排列熵可以较好地识别生物组织是否变性.相对于EMD-MPE-GK模糊聚类以及VMD-小波熵(WE)-GK模糊聚类变性识别方法,本文所提方法中变性与未变性组织特征交叠区域数据点更少,聚类效果和分类性能更好;本实验环境下生物组织变性识别结果表明,该方法的识别率更高,高达93.81%.
It is an important practical problem to accurately recognize whether biological tissue is denatured during high intensity focused ultrasound(HIFU) treatment. Ultrasonic scattering echo signals are related to some physical properties of biological tissues. According to the characteristics of ultrasonic scattering echo signals, the recognition of denatured biological tissues is studied based on the variational mode decomposition(VMD) and multi-scale permutation entropy(MPE) in this paper. The ultrasonic echo signals are decomposed into various modal components by the VMD. The noise components and the useful components are separated according to the power spectrum information entropy of various modal components. The separated useful signals are reconstructed and the MPE are extracted. Furthermore, Gustafson-Kessel(GK) fuzzy clustering analysis is employed to obtain the standard clustering center, and the recognition of denatured biological tissues is carried out by Euclid approach degree and principle of proximity. The proposed method is applied to ultrasonic scattering echo signal during HIFU treatment. In order to determine the parameters of MPE algorithm for ultrasonic scattering echo signals, the embedding dimension of the MPE is discussed, and the scale factor of the MPE algorithm is optimized by genetic algorithm. When the delay time and the embedding dimension are 2and 7 respectively, the MPE values decrease with scale factor increasing. Assuming that the scale factor is 12 from optimization results, the 293 ultrasonic scattering echo signals from normal tissues and denatured tissues are analyzed by the MPE. It is found that the MPE values of the denatured tissues are higher than those of the normal tissues. The MPE can be used to distinguish normal tissues and denatured tissues. Comparing with the recognition methods of the EMD-MPE-GK fuzzy clustering method and the VMD-WE-GK fuzzy clustering, the proposed method has good clustering performance and separability. Its partition coefficient(PC) is close to 1and the Xie-Beni(XB) index is smaller. There are fewer feature points in the overlap region between MPE features of denatured tissues and normal tissues. The recognition results of denatured biological tissues in this experimental environment show that the recognition rate based on this method is higher, reaching up to93.81%.
It is an important practical problem to accurately recognize whether biological tissue is denatured during high intensity focused ultrasound(HIFU) treatment. Ultrasonic scattering echo signals are related to some physical properties of biological tissues. According to the characteristics of ultrasonic scattering echo signals, the recognition of denatured biological tissues is studied based on the variational mode decomposition(VMD) and multi-scale permutation entropy(MPE) in this paper. The ultrasonic echo signals are decomposed into various modal components by the VMD. The noise components and the useful components are separated according to the power spectrum information entropy of various modal components. The separated useful signals are reconstructed and the MPE are extracted. Furthermore, Gustafson-Kessel(GK) fuzzy clustering analysis is employed to obtain the standard clustering center, and the recognition of denatured biological tissues is carried out by Euclid approach degree and principle of proximity. The proposed method is applied to ultrasonic scattering echo signal during HIFU treatment. In order to determine the parameters of MPE algorithm for ultrasonic scattering echo signals, the embedding dimension of the MPE is discussed, and the scale factor of the MPE algorithm is optimized by genetic algorithm. When the delay time and the embedding dimension are 2and 7 respectively, the MPE values decrease with scale factor increasing. Assuming that the scale factor is 12 from optimization results, the 293 ultrasonic scattering echo signals from normal tissues and denatured tissues are analyzed by the MPE. It is found that the MPE values of the denatured tissues are higher than those of the normal tissues. The MPE can be used to distinguish normal tissues and denatured tissues. Comparing with the recognition methods of the EMD-MPE-GK fuzzy clustering method and the VMD-WE-GK fuzzy clustering, the proposed method has good clustering performance and separability. Its partition coefficient(PC) is close to 1and the Xie-Beni(XB) index is smaller. There are fewer feature points in the overlap region between MPE features of denatured tissues and normal tissues. The recognition results of denatured biological tissues in this experimental environment show that the recognition rate based on this method is higher, reaching up to93.81%.
引文
[1] Cranston D 2015 Ultrason. Sonochem. 27 654
[2] Ellens N, Hynynen K 2015 Med. Phys. 42 4896
[3] Bailey M, Khokhlova V, Sapozhnikov O, Kargl S, Crum L2003 Acoust.Phys. 49 369
[4] Worthington A E, Tranchtenberg J, Sherar M D 2002Ultrasound Med. Biol. 10 1311
[5] Damianou C A, Sanghvi N T, Fry F J, Maass-Moreno R 1997J. Acoust. Soc. Am. 102 628
[6] Sheng L, Zhou Z H, Wu S C, Ding Q Y, Zeng Y 2014 Journal of Beijing Polytechnic University 40 139(in Chinese)[盛磊,周著黄,吴水才,丁琪瑛,曾毅2014北京工业大学学报40 139]
[7] Bamber J C, Hill C R 1979 Ultrasound Med. Biol. 5 149
[8] Ming W, Tan Q L, Qian S Y, Zou X, Liao Z Y 2015 Journal of Test and Measurement Technology 5 404(in Chinese)[明文,谭乔来,钱盛友,邹孝,廖志远2015测试技术学报5 404]
[9] Arthur R M, Straube W L, Starman J D, Moros E G 2003Med. Phys. 30 1021
[10] Seip R, Ebbini E S 1995 IEEE Trans. Biomed. Eng. 42 828
[11] Parker K J 1983 Ultrasound Med. Biol. 9 363
[12] Bloch S, Bailey M R, Crum L A, Kaczkowski P J, Keilman G W, Mourad P D 1998 J. Acoust. Soc. Am. 103 2868
[13] Wu Z H, Huang N E, Chen X Y 2009 Adv. Adapt. Data Anal.11
[14] Zhou Y T, Wang Y Y 2010 Acta Acustica 5 495(in Chinese)[周彦婷,汪源源2010声学学报5 495]
[15] Dragomiretskiy K, Zosso D 2014 IEEE Trans. Signal Process.62 531
[16] Li J, Ning X 2006 Phys. Rev. E 73 88
[17] Yang X J, Yang Y, Li H Z, Zhong N 2016 Acta Phys. Sin. 65218701(in Chinese)[杨孝敬,杨阳,李淮周,钟宁2016物理学报65 218701]
[18] Yan B G, Zhao T T 2011 Acta Phys. Sin. 60 078701(in Chinese)[严碧歌,赵婷婷2011物理学报60 078701]
[19] Bandt C, Pompe B 2002 Phys. Rev. Lett. 88 174102
[20] Gao Y, Villecco F, Li M, Song W 2017 Entropy 19 176
[21] Yao W P, Liu T B, Dai J F, Wang J 2014 Acta Phys. Sin. 63078704(in Chinese)[姚文坡,刘铁兵,戴加飞,王俊2014物理学报63 078704]
[22] Rahimian S,Tavakkoli J 2013 J. Ther. Ultrasound 1 1
[23] Davari A, Marhaban M H, Noor S, Karimadini M,Karimoddini A 2010 Fuzzy Sets Syst. 163 45
[2] Ellens N, Hynynen K 2015 Med. Phys. 42 4896
[3] Bailey M, Khokhlova V, Sapozhnikov O, Kargl S, Crum L2003 Acoust.Phys. 49 369
[4] Worthington A E, Tranchtenberg J, Sherar M D 2002Ultrasound Med. Biol. 10 1311
[5] Damianou C A, Sanghvi N T, Fry F J, Maass-Moreno R 1997J. Acoust. Soc. Am. 102 628
[6] Sheng L, Zhou Z H, Wu S C, Ding Q Y, Zeng Y 2014 Journal of Beijing Polytechnic University 40 139(in Chinese)[盛磊,周著黄,吴水才,丁琪瑛,曾毅2014北京工业大学学报40 139]
[7] Bamber J C, Hill C R 1979 Ultrasound Med. Biol. 5 149
[8] Ming W, Tan Q L, Qian S Y, Zou X, Liao Z Y 2015 Journal of Test and Measurement Technology 5 404(in Chinese)[明文,谭乔来,钱盛友,邹孝,廖志远2015测试技术学报5 404]
[9] Arthur R M, Straube W L, Starman J D, Moros E G 2003Med. Phys. 30 1021
[10] Seip R, Ebbini E S 1995 IEEE Trans. Biomed. Eng. 42 828
[11] Parker K J 1983 Ultrasound Med. Biol. 9 363
[12] Bloch S, Bailey M R, Crum L A, Kaczkowski P J, Keilman G W, Mourad P D 1998 J. Acoust. Soc. Am. 103 2868
[13] Wu Z H, Huang N E, Chen X Y 2009 Adv. Adapt. Data Anal.11
[14] Zhou Y T, Wang Y Y 2010 Acta Acustica 5 495(in Chinese)[周彦婷,汪源源2010声学学报5 495]
[15] Dragomiretskiy K, Zosso D 2014 IEEE Trans. Signal Process.62 531
[16] Li J, Ning X 2006 Phys. Rev. E 73 88
[17] Yang X J, Yang Y, Li H Z, Zhong N 2016 Acta Phys. Sin. 65218701(in Chinese)[杨孝敬,杨阳,李淮周,钟宁2016物理学报65 218701]
[18] Yan B G, Zhao T T 2011 Acta Phys. Sin. 60 078701(in Chinese)[严碧歌,赵婷婷2011物理学报60 078701]
[19] Bandt C, Pompe B 2002 Phys. Rev. Lett. 88 174102
[20] Gao Y, Villecco F, Li M, Song W 2017 Entropy 19 176
[21] Yao W P, Liu T B, Dai J F, Wang J 2014 Acta Phys. Sin. 63078704(in Chinese)[姚文坡,刘铁兵,戴加飞,王俊2014物理学报63 078704]
[22] Rahimian S,Tavakkoli J 2013 J. Ther. Ultrasound 1 1
[23] Davari A, Marhaban M H, Noor S, Karimadini M,Karimoddini A 2010 Fuzzy Sets Syst. 163 45