您的位置：山东大学 -> 科技期刊社 -> 《山东大学学报（工学版）》

山东大学学报 (工学版) ›› 2020, Vol. 50 ›› Issue (3): 15-23.doi: 10.6040/j.issn.1672-3961.0.2019.305

• 机器学习与数据挖掘 • 上一篇    下一篇

基于Gabor特征的乳腺肿瘤MR图像分类识别模型

袁高腾1(),刘毅慧1,*(),黄伟2,胡兵3   

  1. 1. 齐鲁工业大学(山东省科学院)计算机科学与技术学院，山东 济南 250353
    2. 山东省肿瘤防治研究院，山东第一医科大学(山东省医学科学院)放疗科，山东 济南 250117
    3. 济南大学医学与生命科学学院，山东 济南 250022
  • 收稿日期:2019-06-13 出版日期:2020-06-20 发布日期:2020-06-16
  • 通讯作者: 刘毅慧 E-mail:yuangaoteng@163.com;yxl@qlu.edu.cn
  • 作者简介:袁高腾(1993—)，男，江苏盐城人，硕士研究生，主要研究方向为图像处理与模式识别. E-mail：yuangaoteng@163.com
  • 基金资助:
    国家自然科学基金(81773232);国家自然科学基金(81530060);国家自然科学基金(81402538);国家自然科学基金(61375013);山东省自然科学基金(ZR2013FM20)

MR image classification and recognition model of breast cancer based onGabor feature

Gaoteng YUAN1(),Yihui LIU1,*(),Wei HUANG2,Bing HU3   

  1. 1. School of Computer Science and Technology, Qilu University of Technology (Shandong Academy of Sciences), Jinan 205353, Shandong, China
    2. Department of Radiation Oncology, Shandong Cancer Hospitaland Institute, Shandong First Medical University and Shandong Academy of Medical Sciences, Jinan 250117, Shandong, China
    3. School of Medicine and Life Sciences, University of Jinan, Jinan 250022, Shandong, China
  • Received:2019-06-13 Online:2020-06-20 Published:2020-06-16
  • Contact: Yihui LIU E-mail:yuangaoteng@163.com;yxl@qlu.edu.cn
  • Supported by:
    国家自然科学基金(81773232);国家自然科学基金(81530060);国家自然科学基金(81402538);国家自然科学基金(61375013);山东省自然科学基金(ZR2013FM20)

RichHTML

15

PDF (PC)

860

摘要：

为研究乳腺肿瘤核磁(Magnetic Resonance，MR)图像纹理分析在鉴别乳腺纤维瘤(fibroadenoma of breast, FB)、浸润性导管癌(invasive ductal carcinoma, IDC)和浸润性小叶癌(invasive lobular carcinoma, ILC)中的临床应用价值，选择MR图像的兴趣区域并使用小波变换对MR图像进行分解，结合K-means算法完成对肿瘤区域的勾画。使用Gabor小波从8个方向、5个尺度对兴趣区域滤波，并将肿瘤部位的均值作为特征。对提取的特征进行分析、筛选，得到关键特征。比较支持向量机、贝叶斯、神经网络等不同的分类算法对关键特征进行分类预测，计算分类的准确度、灵敏度和特异性，得到最适用于分类模型的参数设置。乳腺MR图像纹理分析能够区分出常见的三类乳腺肿瘤,预测精度为77.36%。乳腺MR图像在鉴别FB、IDC和ILC方面具有重要的临床价值。

关键词: 乳腺肿瘤, 核磁图像, Gabor, 纹理分析, 机器学习

Abstract:

To investigate the clinical value of breast tumor magnetic resonance (MR) images in differentiating fibroadenoma of breast (FB), invasive ductal carcinoma (IDC) and invasive lobular carcinoma (ILC), the region of interest of MR image was selected and the MR image was decomposed by wavelet transform, and the region of tumor was segmented by K-means algorithm. Gabor wavelet was used to filter the region of interest from 8 directions and 5 scales, and the mean value of the tumor location was taken as the feature. The extracted features were analyzed and the key features were obtained. Different classification algorithms were compared in machine learning, such as support vector machine, Bayesian, and neural network, to classify and predict the key features, and calculate the accuracy, sensitivity and specificity of classification, so as to get the most suitable parameter settings for classification model. Texture analysis of breast MR images could distinguish three kinds of common breast tumors, and the prediction accuracy was 77.36%, which showed that MR image had important clinical value in differentiating FB, IDC and ILC.

Key words: breast tumor, MR images, Gabor, texture analysis, machine learning

中图分类号: 

  • TP391

图1

不同类型的乳腺肿瘤MR图像"

图2

乳腺肿瘤MR图像ROI区域"

图3

DWT分解过程"

图4

DWT分解后的子图"

图5

肿瘤部位分割过程"

图6

不同方向下的二维Gabor小波函数"

图7

5个尺度、8个方向Gabor小波滤波图像"

图8

Gabor小波方向为0时三种肿瘤特征均值对比图"

图9

方向为0，尺度为$2\sqrt 2$时三种肿瘤的特征值"

表1

乳腺纤维瘤和浸润性小叶癌预测结果"

关键特征个数 支持向量机 BP神经网络 贝叶斯
正确率 特异性 灵敏度 正确率 特异性 灵敏度 正确率 特异性 灵敏度
3 72.22 88.89 55.56 61.11 44.44 77.78 72.22 55.56 88.88
4 72.22 77.78 66.67 72.22 66.66 77.78 61.10 44.44 77.78
5 83.33 100.00 66.67 61.11 66.67 66.66 72.22 44.44 66.66
6 61.10 33.33 88.88 72.22 88.88 55.56 72.22 77.78 66.66
7 61.10 88.88 22.22 72.22 77.78 66.67 77.77 55.56 100.00
8 44.44 44.44 44.44 66.67 77.78 44.44 44.55 22.22 66.66

表2

乳腺纤维瘤和浸润性导管癌预测结果"

关键特征个数 支持向量机 BP神经网络 贝叶斯
正确率 特异性 灵敏度 正确率 特异性 灵敏度 正确率 特异性 灵敏度
3 97.22 100.00 94.44 100.00 100.00 100.00 100.00 100.00 100.00
4 94.44 100.00 88.89 97.22 100.00 94.44 100.00 100.00 100.00
5 100.00 100.00 100.00 83.33 66.67 100.00 100.00 100.00 100.00
6 100.00 100.00 100.00 97.22 100.00 94.44 100.00 100.00 100.00
7 97.22 100.00 94.44 97.22 100.00 94.44 100.00 100.00 100.00
8 100.00 100.00 100.00 97.22 100.00 94.44 97.22 100.00 94.44

表3

三个类型乳腺肿瘤预测结果"

关键特征个数 支持向量机 BP神经网络 贝叶斯
Q1 QFB QIDC QILC Q1 QFB QIDC QILC Q1 QFB QIDC QILC
3 71.70 66.67 82.14 62.50 69.80 52.94 78.57 75.00 67.92 58.82 71.43 75.00
4 73.58 66.67 85.71 62.50 67.92 47.06 85.71 50.00 71.70 70.59 71.43 75.00
5 77.36 66.57 92.86 62.50 75.47 66.67 85.71 75.00 77.36 70.59 82.14 75.00
6 69.81 64.70 78.57 50.00 67.92 52.94 78.57 62.50 67.92 58.82 75.00 62.50
7 73.58 64.70 85.71 50.00 73.58 58.82 85.71 62.50 71.69 64.71 78.57 62.50
8 71.69 66.67 85.71 50.00 67.92 52.94 78.57 62.50 71.69 58.82 78.57 75.00
1 MCGRATH S E , RING A . Chemotherapy for breast cancer in pregnancy: evidence and guidance for oncologists[J]. Therapeutic Advances in Medical Oncology, 2011, 3 (2): 73- 83.
2 郑莹, 吴春晓, 张敏璐. 乳腺癌在中国的流行状况和疾病特征[J]. 中国癌症杂志, 2013, 23 (8): 561- 569.
ZHENG Ying , WU Chunxiao , ZHANG Minlu . The epidemic and characteristics of female breast cancer in China[J]. China Oncology, 2013, 23 (8): 561- 569.
3 李明慧, 柳莉莎. 超声弹性成像评分标准对乳腺良恶性肿块的诊断价值[J]. 肿瘤, 2011, 31 (5): 453- 456.
LI Minghui , LIU Lisha . The diagnostic value of ultrasonic elastography in identifying malignancies of breast diseases[J]. Tumor, 2011, 31 (5): 453- 456.
4 唐玮, 刘剑仑, 杨华伟, 等. 整形保乳术与常规保乳术在早期乳腺癌治疗中的比较分析[J]. 中国肿瘤临床, 2016, 43 (6): 235- 239.
TANG Wei , LIU Jianlun , YANG Huawei , et al. Clinical comparative study of oncoplastic and standard breast-conserving surgery in the treatment of early breast cancer[J]. Chinese Journal of Clinical Oncology, 2016, 43 (06): 235- 239.
5 O'SULLIVAN T D , LEPROUX A , CHEN J H , et al. Optical imaging correlates with magnetic resonance imaging breast density and reveals composition changes during neoadjuvant chemotherapy[J]. Breast Cancer Research, 2013, 15 (1): 14.
6 SUTTON E J , DASHEVSKY B Z , OH J H , et al. Breast cancer molecular subtype classifier that incorporates MRI features[J]. Journal of Magnetic Resonance Imaging, 2016, 44 (1): 122- 129.
7 刘丽, 赵凌君, 郭承玉, 等. 图像纹理分类方法研究进展和展望[J]. 自动化学报, 2018, 44 (4): 584- 607.
LIU Li , ZHAO Lingjun , GUO Chengyu , et al. Texture classification: state-of-the-art methods and prospects[J]. Acta Automatica Sinica, 2018, 44 (4): 584- 607.
8 ZACHARAKI EI , WANG S , CHAWLA S , et al. Classification of brain tumor type and grade using MRI texture and shape in a machine learning scheme[J]. Magnetic Resonance in Medicine, 2010, 62 (6): 1609- 1618.
9 LIU Yihui , MUFTAH M , DAS T , et al. Classification of MR tumor images based on Gabor wavelet analysis[J]. Journal of Medical & Biological Engineering, 2012, 32 (1): 22- 28.
10 PAN Y, HUANG W, LIN Z, et al. Brain tumor grading based onneural networks and convolutional neural networks[C]// Conf Proc IEEE Eng Med Biol Soc, (2015).[S.l.]: [s.n.], 2015: 699-702.
11 KOOI T , LITJENS G , GINNEKEN B V , et al. Large scale deep learning for computer aided detection of mammographic lesions[J]. Medical Image Analysis, 2017, 35, 303- 312.
12 MERCKEL L G , VERKOOIJEN H M , PETERS N H G M , et al. Theadded diagnostic value of dynamic contrast-enhanced MRI at 3.0 T in nonpalpable breast lesions[J]. Plos One, 2014, 9 (4): e94233.
13 MICHAUT M , CHIN S F , MAJEWSKI I , et al. Integration of genomic, transcriptomic and proteomic data identifies two biologically distinct subtypes ofinvasive lobular breast cancer[J]. Sci Rep, 2016, 6, 18517.
14 ULANER G A , GOLDMAN D , GONEN M , et al. Initial results of a prospective clinical trial of 18F-Fluciclovine PET/CT in newly diagnosed invasive ductal and invasive lobular breast cancers[J]. Journal of Nuclear Medicine, 2016, 57 (9): 1350.
15 陈波, 张磊. 2017年乳腺癌新辅助治疗进展[J]. 山东大学学报(医学版), 2018, 56 (1): 12- 16.
CHEN Bo , ZHANG Lei . Current perspectives of neoadjuvant therapy for breast cancer in 2017[J]. Journal of Shandong University (Health Sciences), 2018, 56 (1): 12- 16.
16 CHEN Yunmei , YE Xiaojing , HUANG Feng , et al. A novel method and fast algorithm for MR image reconstruction with significantly under-sampled data[J]. Inverse Problems & Imaging, 2017, 4 (2): 223- 240.
17 ACHARYA U R , MOOKIAH M R , KOH J E , et al. Novel risk index for the identification of age-related macular degeneration using radon transform and DWT features[J]. Computers in Biology & Medicine, 2016, 73 (C): 131- 140.
18 刘慧, 王小宜, 龙学颖. 基于CT图像纹理分析肿瘤异质性的研究进展及应用[J]. 国际医学放射学杂志, 2016, 39 (5): 543- 548.
LIU Hui , WANG Xiaoyi , LONG Xueying . Research progress and clinical application of tumor heterogeneity based on CT texture analysis[J]. International Journal of Medical Radiology, 2016, 39 (5): 543- 548.
19 HUANG Xiuchang, SU Wei. An improved K-means clustering algorithm[C]// World Automation Congress. Hawaii, USA: Journal of Networks, 2014, 9(1): 161-167.
20 范春年, 张福炎. Gabor相位特征的人脸光照不变量提取[J]. 中国图象图形学报, 2012, 17 (5): 676- 681.
FAN Chunnian , ZHANG Fuyan . Illumination invariant extraction on Gabor phase[J]. Journal of Image and Graphics, 2012, 17 (5): 676- 681.
21 KAY S , QUAN D , BO T , et al. Probability density function estimation using the EEF With application to subset/feature selection[J]. IEEE Transactions on Signal Processing, 2016, 64 (3): 641- 651.
22 侯霄雄, 许新征, 朱炯, 等. 基于AlexNet和集成分类器的乳腺癌计算机辅助诊断方法[J]. 山东大学学报(工学版), 2019, 49 (2): 74- 79.
HOU Xiaoxiong , XU Xinzheng , ZHU Jiong , et al. Computer aided diagnosis method for breast cancer based on AlexNet and ensemble classifiers[J]. Journal of Shandong University(Engineering Science), 2019, 49 (2): 74- 79.
[1] 祝明,石承龙,吕潘,刘现荣,孙驰,陈建城,范宏运. 基于优化长短时记忆网络的深基坑变形预测方法及其工程应用[J]. 山东大学学报 (工学版), 2025, 55(3): 141-148.
[2] 常新功,苏敏惠,周志刚. 基于进化集成的图神经网络解释方法[J]. 山东大学学报 (工学版), 2024, 54(4): 1-12.
[3] 乔慧妍,段学龙,解驰皓,赵冬慧,马玉玲. 基于异常点检测的心理健康辅助诊断方法[J]. 山东大学学报 (工学版), 2024, 54(4): 76-85.
[4] 刘新,刘冬兰,付婷,王勇,常英贤,姚洪磊,罗昕,王睿,张昊. 基于联邦学习的时间序列预测算法[J]. 山东大学学报 (工学版), 2024, 54(3): 55-63.
[5] 岳仁峰,张嘉琦,刘勇,范学忠,李琮琮,孔令鑫. 基于颜色和纹理特征的立体车库锈蚀检测技术[J]. 山东大学学报 (工学版), 2024, 54(3): 64-69.
[6] 陈成,董永权,贾瑞,刘源. 基于交互序列特征相关性的可解释知识追踪[J]. 山东大学学报 (工学版), 2024, 54(1): 100-108.
[7] 卞小曼,王小琴,蓝如师,刘振丙,罗笑南. 基于相似性保持和判别性分析的快速视频哈希算法[J]. 山东大学学报 (工学版), 2023, 53(6): 63-69.
[8] 李鸿钊,张庆松,刘人太,陈新,辛勤,石乐乐. 浅埋地铁车站施工期地表变形风险预警[J]. 山东大学学报 (工学版), 2023, 53(6): 82-91.
[9] 袁高腾,周晓峰,郭宏乐. 基于特征选择算法的ECG信号分类[J]. 山东大学学报 (工学版), 2022, 52(4): 38-44.
[10] 杨霄,袭肖明,李维翠,杨璐. 基于层次化双重注意力网络的乳腺多模态图像分类[J]. 山东大学学报 (工学版), 2022, 52(3): 34-41.
[11] 聂秀山,马玉玲,乔慧妍,郭杰,崔超然,于志云,刘兴波,尹义龙. 任务粒度视角下的学生成绩预测研究综述[J]. 山东大学学报 (工学版), 2022, 52(2): 1-14.
[12] 孙鸿昌,周风余,单明珠,翟文文,牛兰强. 基于模式划分的空调能耗混合填补方法[J]. 山东大学学报 (工学版), 2022, 52(1): 9-18.
[13] 傅桂霞,邹国锋,毛帅,潘金凤,尹丽菊. 融合Gabor特征与卷积特征的小样本行人重识别[J]. 山东大学学报 (工学版), 2021, 51(3): 22-29.
[14] 张大鹏,刘雅军,张伟,沈芬,杨建盛. 基于异质集成学习的虚假评论检测[J]. 山东大学学报 (工学版), 2020, 50(2): 1-9.
[15] 高铭壑,张莹,张蓉蓉,黄子豪,黄琳焱,李繁菀,张昕,王彦浩. 基于预测数据特征的空气质量预测方法[J]. 山东大学学报 (工学版), 2020, 50(2): 91-99.
Viewed
Full text


Abstract

Cited
  Shared   
  Discussed   
[1] 李 侃 . 嵌入式相贯线焊接控制系统开发与实现[J]. 山东大学学报(工学版), 2008, 38(4): 37 -41 .
[2] 来翔 . 用胞映射方法讨论一类MKdV方程[J]. 山东大学学报(工学版), 2006, 36(1): 87 -92 .
[3] 余嘉元1 , 田金亭1 , 朱强忠2 . 计算智能在心理学中的应用[J]. 山东大学学报(工学版), 2009, 39(1): 1 -5 .
[4] 陈瑞,李红伟,田靖. 磁极数对径向磁轴承承载力的影响[J]. 山东大学学报(工学版), 2018, 48(2): 81 -85 .
[5] 王波,王宁生 . 机电装配体拆卸序列的自动生成及组合优化[J]. 山东大学学报(工学版), 2006, 36(2): 52 -57 .
[6] 张英,郎咏梅,赵玉晓,张鉴达,乔鹏,李善评 . 由EGSB厌氧颗粒污泥培养好氧颗粒污泥的工艺探讨[J]. 山东大学学报(工学版), 2006, 36(4): 56 -59 .
[7] 王丽君,黄奇成,王兆旭 . 敏感性问题中的均方误差与模型比较[J]. 山东大学学报(工学版), 2006, 36(6): 51 -56 .
[8] Yue Khing Toh1 ， XIAO Wendong2 ， XIE Lihua1 . 基于无线传感器网络的分散目标跟踪：实际测试平台的开发应用（英文）[J]. 山东大学学报(工学版), 2009, 39(1): 50 -56 .
[9] 孙炜伟,王玉振. 考虑饱和的发电机单机无穷大系统有限增益镇定[J]. 山东大学学报(工学版), 2009, 39(1): 69 -76 .
[10] 孙玉利,李法德,左敦稳,戚美 . 直立分室式流体连续通电加热系统的升温特性[J]. 山东大学学报(工学版), 2006, 36(6): 19 -23 .
版权所有 © 2018 《山东大学学报(工学版)》编辑部　
地址: 济南市山大南路27号（250100）　 电话: 0531-88366735　 E-mail: xbgxb@sdu.edu.cn
本系统由北京玛格泰克科技发展有限公司设计开发