基于超声图像的生物组织损伤判定方法研究
|
摘要
针对高强度聚焦超声(HIFU)治疗的生物组织变性监测,提出一种基于超声图像灰度-梯度共生矩阵和模糊C均值聚类(FCM)的生物组织损伤判定方法。首先对新鲜离体猪肉组织进行高强度聚焦超声辐照,对采集到的超声图像进行预处理后提取减影图像的灰度-梯度共生矩阵的灰度熵、混合熵作为组织损伤判定的表征参数,再使用FCM聚类方法对特征参数进行聚类判别。实验结果表明,结合灰度熵与混合熵的多参数判定比单参数灰度熵、混合熵判定的识别率更高,相比灰度均值与小波系数结合方法,其识别率提高3. 23%。可以更准确地判定在高强度聚焦超声治疗过程中的组织损伤状况。如进一步与无损测温相结合,可进一步提高高强度聚焦超声的疗效。
In order to monitor the biological tissue lesion when using HIFU to treat,a method that biological tissue lesion identification based on ultrasound image by Graylevel-gradient co-occurrence matrix and FCM clustering is proposed. First,obtaining ultrasound images of fresh pork tissue which irradiated by high intensity focused ultrasound and extract parameters of Gray level-gradient cooccurrence matrix after preprocessing. The parameters of gray entropy and entropy of mixing are the characterization parameter of identification of biological tissue lesion. And then using FCM clustering method to analysis. The results show that the rate of identification is higher when combining Gray entropy with entropy of mixing to identify biological tissue lesion than signal parameter,which is 3. 32 percentage more than that gray average and wavelet transform coefficient method. Thus,this method can better identify tissue lesion in the course of HIFU treatment. The method is helpful to improve the efficacy of HIFU treatment,relating the noninvasive temperature measurement with tissue lesion identification further.
In order to monitor the biological tissue lesion when using HIFU to treat,a method that biological tissue lesion identification based on ultrasound image by Graylevel-gradient co-occurrence matrix and FCM clustering is proposed. First,obtaining ultrasound images of fresh pork tissue which irradiated by high intensity focused ultrasound and extract parameters of Gray level-gradient cooccurrence matrix after preprocessing. The parameters of gray entropy and entropy of mixing are the characterization parameter of identification of biological tissue lesion. And then using FCM clustering method to analysis. The results show that the rate of identification is higher when combining Gray entropy with entropy of mixing to identify biological tissue lesion than signal parameter,which is 3. 32 percentage more than that gray average and wavelet transform coefficient method. Thus,this method can better identify tissue lesion in the course of HIFU treatment. The method is helpful to improve the efficacy of HIFU treatment,relating the noninvasive temperature measurement with tissue lesion identification further.
引文
[1] FENG Y L.TAN X. Research on data mining techniqueapplication in non-invasive porcine tissue temperatureestimation[C]. International Conference on SportMedicine,Sport Science and Computer Science(SMSSCS2010),2010:217-220.
[2] LAI P,MCLAUGHLAN J R,DRAUDT A B,et al.Real-time monitoring of high intensity focused ultrasoundlesion formation using acousto optic sensing[J].Ultrasound in Medicine and Biology,2011,37(2):239-252.
[3] DING Y J,QIAN S Y,LIAO Z Y.Simulation of thermalfieldinduced by concave spherical transducer in multi-layer media[J]. Journal of Central South University ofTechnology,2013,20(11):3166-3170.
[4]陈洪波.基于超声成像的HIFU治疗效果实时评价中的关键问题研究[D].长沙:中南大学,2009.CHEN H B.Key problems in HIFU treatment effect real-time estimation based on ultrasound image[D].Changsha:Central South University,2009.
[5]邹孝,钱盛友.一种基于B超图像处理的生物组织变性监测方法[J].计算机工程与科学,2011,33(12):57-60.ZOU X,QIAN SH Y. A method to check biological tissuedenaturation based on mode B ultrasonic image processing[J].Computer Engineering&Science,2011,33(12):57-60.
[6]颜佩,丁亚军,钱盛友,等.基于小波系数Hu矩的生物组织损伤监测方法[J].电子测量与仪器学报,2016,30(7):1062-1067.YAN P,DING Y J,QIAN SH Y,et al. A method ofbiological tissue lesion monitoring based on Hu moment ofthe wavelet transform coefficient[J].Journal of ElectronicMeasurement and Instrumentation,2016,30(7):1062-1067.
[7]陈华,钱盛友,谭乔来,等.基于超声图像特征参数检测高强度聚焦超声引起的组织损伤阈值[J].中国医学物理学杂志,2016,33(11):1144-1148.CHEN H,QIAN SH Y,TAN Q L,et al. Detectingthreshold of tissue lesion induced by HIFU based onultrasonic image feature parameter[J].Chinese Journal ofMedical Physics,2016,33(11):1144-1148.
[8]肖达,王润民,邹孝,等.基于Gabor变换和灰度梯度共生矩阵的超声无损测温研究[J].传感技术学报,2017,30(11):1684-1688.XIAO D,WANG R M,ZOU X. Studyon ultrasonicnondestructive temperature measurement based on Gabortransform and gray gradient symbiotic matrix[J]. Journalof Sensing Technology,2017,30(11):1684-1688.
[9]梁晓霞,封筠.基于Gabor变换和灰度梯度共生矩阵的人耳识别研究[J].石家庄铁道大学学报(自然科学版),2011,24(1):78-83,90.LIANG X X,FENG Y. Human ear recognition researchbased on Gabor transformation and gray gradientsymbiosis matrix[J]. Journal of Shijiazhuang RailwayUniversity(Nature Science Edition),2011,24(1):78-83,90.
[10]盛磊,周著黄,吴水才,等.热消融组织B超图像纹理特征参数温度相关性[J].北京工业大学学报,2013,39(8):1275-1280.SHENG L,ZHOU ZH H,WU SH C,et al. Temperaturecorrelation between thermal ablation tissue B modeultrasound image texture feature parameters[J]. Journalof Beijing University of Technology, 2013,39(8):1275-1280.
[11] LAM S W. Texture feature extraction using gray levelgradient based co-occurence matrices[C]. IEEEInternational Conference on Systems,1996:267-271.
[12]徐涛,叶志前,蔡卫民.肝脏B超图像的纹理分析方法[J].国际生物医学工程杂志,2006,29(4):228-230.XU T,YE ZH Q,CAI W M. Texture analysis methodsused in B-scan liver images[J]. International Journal ofBiomedical Engineering,2006,29(4):228-230.
[13]朱宏擎.基于灰度-梯度共生矩阵的视网膜血管分割方法[J].上海交通大学学报,2004,38(9):1485-1488.ZHU H Q. Method of retinal vascular segmentation basedon gray-gradient symbiotic matrix[J]. Journal ofShanghai Jiaotong University,2004,38(9):1485-1488.
[14] BEZDEK J C,EHRLICH R,FULL W. FCM:The fuzzyC-means clustering algorithm[J]. Computers&Geosciences,1984,10(2-3):191-203.
[15] SUN H,WANG S, JIANG Q. FCM-based modelselection algorithms for determining the number ofclusters[J]. Pattern recognition, 2004, 37(10):2027-2037.
[16] GAO X B,LI J,XIE W X. Parameter optimization inFCM clustering algorithms[C] International Conferenceon Signal Processing,IEEE,2000:1457-1461.
[17]颜佩,丁亚军,钱盛友,等.基于K均值聚类的组织损伤等级判定研究[J].电子测量与仪器学报,2017,31(3):468-473.YAN P,DING Y J,QIAN SH Y,et al. Research onbiological tissue lesion level judgment based on K-meansclustering[J]. Journal of Electronic Measurement andInstrumentation,2017,31(3):468-473.
[2] LAI P,MCLAUGHLAN J R,DRAUDT A B,et al.Real-time monitoring of high intensity focused ultrasoundlesion formation using acousto optic sensing[J].Ultrasound in Medicine and Biology,2011,37(2):239-252.
[3] DING Y J,QIAN S Y,LIAO Z Y.Simulation of thermalfieldinduced by concave spherical transducer in multi-layer media[J]. Journal of Central South University ofTechnology,2013,20(11):3166-3170.
[4]陈洪波.基于超声成像的HIFU治疗效果实时评价中的关键问题研究[D].长沙:中南大学,2009.CHEN H B.Key problems in HIFU treatment effect real-time estimation based on ultrasound image[D].Changsha:Central South University,2009.
[5]邹孝,钱盛友.一种基于B超图像处理的生物组织变性监测方法[J].计算机工程与科学,2011,33(12):57-60.ZOU X,QIAN SH Y. A method to check biological tissuedenaturation based on mode B ultrasonic image processing[J].Computer Engineering&Science,2011,33(12):57-60.
[6]颜佩,丁亚军,钱盛友,等.基于小波系数Hu矩的生物组织损伤监测方法[J].电子测量与仪器学报,2016,30(7):1062-1067.YAN P,DING Y J,QIAN SH Y,et al. A method ofbiological tissue lesion monitoring based on Hu moment ofthe wavelet transform coefficient[J].Journal of ElectronicMeasurement and Instrumentation,2016,30(7):1062-1067.
[7]陈华,钱盛友,谭乔来,等.基于超声图像特征参数检测高强度聚焦超声引起的组织损伤阈值[J].中国医学物理学杂志,2016,33(11):1144-1148.CHEN H,QIAN SH Y,TAN Q L,et al. Detectingthreshold of tissue lesion induced by HIFU based onultrasonic image feature parameter[J].Chinese Journal ofMedical Physics,2016,33(11):1144-1148.
[8]肖达,王润民,邹孝,等.基于Gabor变换和灰度梯度共生矩阵的超声无损测温研究[J].传感技术学报,2017,30(11):1684-1688.XIAO D,WANG R M,ZOU X. Studyon ultrasonicnondestructive temperature measurement based on Gabortransform and gray gradient symbiotic matrix[J]. Journalof Sensing Technology,2017,30(11):1684-1688.
[9]梁晓霞,封筠.基于Gabor变换和灰度梯度共生矩阵的人耳识别研究[J].石家庄铁道大学学报(自然科学版),2011,24(1):78-83,90.LIANG X X,FENG Y. Human ear recognition researchbased on Gabor transformation and gray gradientsymbiosis matrix[J]. Journal of Shijiazhuang RailwayUniversity(Nature Science Edition),2011,24(1):78-83,90.
[10]盛磊,周著黄,吴水才,等.热消融组织B超图像纹理特征参数温度相关性[J].北京工业大学学报,2013,39(8):1275-1280.SHENG L,ZHOU ZH H,WU SH C,et al. Temperaturecorrelation between thermal ablation tissue B modeultrasound image texture feature parameters[J]. Journalof Beijing University of Technology, 2013,39(8):1275-1280.
[11] LAM S W. Texture feature extraction using gray levelgradient based co-occurence matrices[C]. IEEEInternational Conference on Systems,1996:267-271.
[12]徐涛,叶志前,蔡卫民.肝脏B超图像的纹理分析方法[J].国际生物医学工程杂志,2006,29(4):228-230.XU T,YE ZH Q,CAI W M. Texture analysis methodsused in B-scan liver images[J]. International Journal ofBiomedical Engineering,2006,29(4):228-230.
[13]朱宏擎.基于灰度-梯度共生矩阵的视网膜血管分割方法[J].上海交通大学学报,2004,38(9):1485-1488.ZHU H Q. Method of retinal vascular segmentation basedon gray-gradient symbiotic matrix[J]. Journal ofShanghai Jiaotong University,2004,38(9):1485-1488.
[14] BEZDEK J C,EHRLICH R,FULL W. FCM:The fuzzyC-means clustering algorithm[J]. Computers&Geosciences,1984,10(2-3):191-203.
[15] SUN H,WANG S, JIANG Q. FCM-based modelselection algorithms for determining the number ofclusters[J]. Pattern recognition, 2004, 37(10):2027-2037.
[16] GAO X B,LI J,XIE W X. Parameter optimization inFCM clustering algorithms[C] International Conferenceon Signal Processing,IEEE,2000:1457-1461.
[17]颜佩,丁亚军,钱盛友,等.基于K均值聚类的组织损伤等级判定研究[J].电子测量与仪器学报,2017,31(3):468-473.YAN P,DING Y J,QIAN SH Y,et al. Research onbiological tissue lesion level judgment based on K-meansclustering[J]. Journal of Electronic Measurement andInstrumentation,2017,31(3):468-473.