高级搜索

基于深度卷积神经网络的多元医学信号多级上下文自编码器

袁野 贾克斌 刘鹏宇

引用本文: 袁野, 贾克斌, 刘鹏宇. 基于深度卷积神经网络的多元医学信号多级上下文自编码器[J]. 电子与信息学报, 2020, 42(2): 371-378. doi: 10.11999/JEIT190135 shu
Citation:  Ye YUAN, Kebin JIA, Pengyu LIU. Multi-context Autoencoders for Multivariate Medical Signals Based on Deep Convolutional Neural Networks[J]. Journal of Electronics and Information Technology, 2020, 42(2): 371-378. doi: 10.11999/JEIT190135 shu

基于深度卷积神经网络的多元医学信号多级上下文自编码器

    作者简介: 袁野: 男,1991年生,博士生,研究方向为深度学习、健康信息学;
    贾克斌: 男,1962年生,教授,研究方向为多媒体信息系统、模式识别;
    刘鹏宇: 女,1979年生,副教授,研究方向为多媒体信息系统
    通讯作者: 贾克斌,kebinj@bjut.edu.cn
  • 基金项目: 国家自然科学基金(81871394),先进信息网络北京实验室(040000546618017)

摘要: 多元医学信号的典型代表有多模态睡眠图和多通道脑电图等,采用无监督深度学习表征多元医学信号是目前健康信息学领域中的一个研究热点。为了解决现有模型没有充分结合医学信号多元时序结构特点的问题,该文提出了一种无监督的多级上下文深度卷积自编码器(mCtx-CAE)。首先改进传统卷积神经网络结构,提出一种多元卷积自编码模块,以提取信号片段内的多元上下文特征;其次,提出采用语义学习技术对信号片段间的时序信息进行自编码,进一步提取时序上下文特征;最后通过共享特征表示设计目标函数,训练端到端的多级上下文自编码器。实验结果表明,该文所提模型在两种应用于不同医疗场景下的多模态和多通道数据集(UCD和CHB-MIT)上表现均优于其它无监督特征学习方法,能有效提高多元医学信号的融合特征表达能力,对提高临床时序数据的分析效率有着重要意义。

English

    1. [1]

      JOHNSON A E W, GHASSEMI M M, NEMATI S, et al. Machine learning and decision support in critical care[J]. Proceedings of the IEEE, 2016, 104(2): 444–466. doi: 10.1109/JPROC.2015.2501978

    2. [2]

      RAVI D, WONG C, DELIGIANNI F, et al. Deep learning for health informatics[J]. IEEE Journal of Biomedical and Health Informatics, 2017, 21(1): 4–21. doi: 10.1109/JBHI.2016.2636665

    3. [3]

      BOOSTANI R, KARIMZADEH F, and NAMI M. A comparative review on sleep stage classification methods in patients and healthy individuals[J]. Computer Methods and Programs in Biomedicine, 2017, 140: 77–91. doi: 10.1016/j.cmpb.2016.12.004

    4. [4]

      YUAN Ye, XUN Guangxu, JIA Kebin, et al. A multi-view deep learning framework for EEG seizure detection[J]. IEEE Journal of Biomedical and Health Informatics, 2019, 23(1): 83–94. doi: 10.1109/JBHI.2018.2871678

    5. [5]

      ACAR E, LEVIN-SCHWARTZ Y, CALHOUN V D, et al. Tensor-based fusion of EEG and fMRI to understand neurological changes in schizophrenia[C]. 2017 IEEE International Symposium on Circuits and Systems, Baltimore, USA, 2017: 1–4.

    6. [6]

      JIA Xiaowei, LI Kang, LI Xiaoyi, et al. A novel semi-supervised deep learning framework for affective state recognition on EEG signals[C]. 2014 IEEE International Conference on Bioinformatics and Bioengineering, Boca Raton, USA, 2014: 30–37.

    7. [7]

      LÄNGKVIST M, KARLSSON L, and LOUTFI A. A review of unsupervised feature learning and deep learning for time-series modeling[J]. Pattern Recognition Letters, 2014, 42: 11–24. doi: 10.1016/j.patrec.2014.01.008

    8. [8]

      HOLZINGER A. Machine Learning for Health Informatics[M]. Cham: Springer, 2016: 161–182.

    9. [9]

      SUPRATAK A, LI Ling, and GUO Yike. Feature extraction with stacked autoencoders for epileptic seizure detection[C]. The 36th Annual International Conference of the IEEE Engineering in Medicine and Biology Society, Chicago, USA, 2014: 4184–4187.

    10. [10]

      YAN Bo, WANG Yong, LI Yuheng, et al. An EEG signal classification method based on sparse auto-encoders and support vector machine[C]. 2016 IEEE/CIC International Conference on Communications in China, Chengdu, China, 2016: 1–6.

    11. [11]

      LIN Qin, YE Shuqun, HUANG Xiumei, et al. Classification of epileptic EEG signals with stacked sparse autoencoder based on deep learning[C]. The 12th International Conference on Intelligent Computing, Lanzhou, China, 2016: 802–810.

    12. [12]

      YANG Jianli, BAI Yang, LI Guojun, et al. A novel method of diagnosing premature ventricular contraction based on sparse auto-encoder and softmax regression[J]. Bio-medical Materials and Engineering, 2015, 26(S1): S1549–S1558. doi: 10.3233/BME-151454

    13. [13]

      XUN Guangxu, JIA Xiaowei, and ZHANG Aidong. Detecting epileptic seizures with electroencephalogram via a context-learning model[J]. BMC Medical Informatics and Decision Making, 2016, 16(Suppl 2): 70. doi: 10.1186/s12911-016-0310-7

    14. [14]

      LI Xiaoyi, JIA Xiaowei, XUN Guangxu, et al. Improving EEG feature learning via synchronized facial video[C]. 2015 IEEE International Conference on Big Data, Santa Clara, USA, 2015: 843–848.

    15. [15]

      YUAN Ye, XUN Guangxu, SUO Qiuling, et al. Wave2Vec: Deep representation learning for clinical temporal data[J]. Neurocomputing, 2019, 324: 31–42. doi: 10.1016/j.neucom.2018.03.074

    16. [16]

      YUAN Ye, XUN Guangxu, JIA Kebin, et al. A multi-context learning approach for EEG epileptic seizure detection[J]. BMC Systems Biology, 2018, 12(6): 47–57. doi: 10.1186/s12918-018-0626-2

    17. [17]

      ZHANG Junming, WU Yan, BAI Jing, et al. Automatic sleep stage classification based on sparse deep belief net and combination of multiple classifiers[J]. Transactions of the Institute of Measurement and Control, 2016, 38(4): 435–451. doi: 10.1177/0142331215587568

    18. [18]

      YULITA I N, FANANY M I, and ARYMURTHY A M. Sequence-based sleep stage classification using conditional neural fields[J]. arXiv preprint arXiv:1610.01935 , 2016.

    19. [19]

      LÄNGKVIST M, KARLSSON L, and LOUTFI A. Sleep stage classification using unsupervised feature learning[J]. Advances in Artificial Neural Systems, 2012, 2012: 107046. doi: 10.1155/2012/107046

    20. [20]

      MASCI J, MEIER U, CIREŞAN D, et al. Stacked convolutional auto-encoders for hierarchical feature extraction[C]. The 21st International Conference on Artificial Neural Networks, Espoo, Finland, 2011: 52–59.

    21. [21]

      HINTON G E and SALAKHUTDINOV R R. Reducing the dimensionality of data with neural networks[J]. Science, 2006, 313(5786): 504–507. doi: 10.1126/science.1127647

    22. [22]

      MIKOLOV T, SUTSKEVER I, CHEN Kai, et al. Distributed representations of words and phrases and their compositionality[C]. The 26th International Conference on Neural Information Processing Systems, Lake Tahoe, USA, 2013: 3111–3119.

    23. [23]

      CHOI E, BAHADORI M T, SEARLES E, et al. Multi-layer representation learning for medical concepts[C]. The 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, San Francisco, USA, 2016: 1495–1504.

    24. [24]

      SHOEB A H. Application of machine learning to epileptic seizure onset detection and treatment[D]. [Ph.D. dissertation], Massachusetts Institute of Technology, 2009.

    25. [25]

      GOLDBERGER A L, AMARAL L A N, GLASS L, et al. PhysioBank, PhysioToolkit, and PhysioNet: Components of a new research resource for complex physiologic signals[J]. Circulation, 2000, 101(23): E215–E220. doi: 10.1161/01.CIR.101.23.e215

    26. [26]

      FAWCETT T. An introduction to ROC analysis[J]. Pattern Recognition Letters, 2006, 27(8): 861–874. doi: 10.1016/j.patrec.2005.10.010

    27. [27]

      DAVIS J and GOADRICH M. The relationship between precision-recall and ROC curves[C]. The 23rd International Conference on Machine Learning, Pittsburgh, USA, 2006: 233–240.

    28. [28]

      HE Haibo and GARCIA E A. Learning from imbalanced data[J]. IEEE Transactions on Knowledge and Data Engineering, 2009, 21(9): 1263–1284. doi: 10.1109/TKDE.2008.239

    29. [29]

      ZEILER M D. ADADELTA: An adaptive learning rate method[J]. arXiv preprint arXiv:1212.5701, 2012.

    1. [1]

      李祖贺, 樊养余, 王凤琴. YUV空间中基于稀疏自动编码器的无监督特征学习. 电子与信息学报, 2016, 38(1): 29-37.

    2. [2]

      李寰宇, 毕笃彦, 杨源, 查宇飞, 覃兵, 张立朝. 基于深度特征表达与学习的视觉跟踪算法研究. 电子与信息学报, 2015, 37(9): 2033-2039.

    3. [3]

      杨宏宇, 王峰岩. 基于深度卷积神经网络的气象雷达噪声图像语义分割方法. 电子与信息学报, 2019, 41(10): 2373-2381.

    4. [4]

      王鑫, 李可, 宁晨, 黄凤辰. 基于深度卷积神经网络和多核学习的遥感图像分类方法. 电子与信息学报, 2019, 41(5): 1098-1105.

    5. [5]

      李寰宇, 毕笃彦, 查宇飞, 杨源. 一种易于初始化的类卷积神经网络视觉跟踪算法. 电子与信息学报, 2016, 38(1): 1-7.

    6. [6]

      吴泽民, 王军, 胡磊, 田畅, 曾明勇, 杜麟. 基于卷积神经网络与全局优化的协同显著性检测. 电子与信息学报, 2018, 40(12): 2896-2904.

    7. [7]

      伍家松, 达臻, 魏黎明, SENHADJILotfi, 舒华忠. 基于分裂基-2/(2a)FFT算法的卷积神经网络加速性能的研究. 电子与信息学报, 2017, 39(2): 285-292.

    8. [8]

      王海, 蔡英凤, 贾允毅, 陈龙, 江浩斌. 基于深度卷积神经网络的场景自适应道路分割算法. 电子与信息学报, 2017, 39(2): 263-269.

    9. [9]

      程帅, 曹永刚, 孙俊喜, 赵立荣, 刘广文, 韩广良. 基于增强群跟踪器和深度学习的目标跟踪. 电子与信息学报, 2015, 37(7): 1646-1653.

    10. [10]

      董书琴, 张斌. 基于深度特征学习的网络流量异常检测方法. 电子与信息学报, 2019, 41(0): 1-9.

    11. [11]

      赵海涛, 程慧玲, 丁仪, 张晖, 朱洪波. 基于深度学习的车联边缘网络交通事故风险预测算法研究. 电子与信息学报, 2020, 42(1): 50-57.

    12. [12]

      王斐, 吴仕超, 刘少林, 张亚徽, 魏颖. 基于脑电信号深度迁移学习的驾驶疲劳检测. 电子与信息学报, 2019, 41(9): 2264-2272.

    13. [13]

      杜兰, 魏迪, 李璐, 郭昱辰. 基于半监督学习的SAR目标检测网络. 电子与信息学报, 2020, 42(1): 154-163.

    14. [14]

      王星, 周一鹏, 周东青, 陈忠辉, 田元荣. 基于深度置信网络和双谱对角切片的低截获概率雷达信号识别. 电子与信息学报, 2016, 38(11): 2972-2976.

    15. [15]

      吕晓琪, 吴凉, 谷宇, 张明, 李菁. 基于深度卷积神经网络的低剂量CT肺部去噪. 电子与信息学报, 2018, 40(6): 1353-1359.

    16. [16]

      郭晨, 简涛, 徐从安, 何友, 孙顺. 基于深度多尺度一维卷积神经网络的雷达舰船目标识别. 电子与信息学报, 2019, 41(6): 1302-1309.

    17. [17]

      秦华标, 曹钦平. 基于FPGA的卷积神经网络硬件加速器设计. 电子与信息学报, 2019, 41(11): 2599-2605.

    18. [18]

      王巍, 周凯利, 王伊昌, 王广, 袁军. 基于快速滤波算法的卷积神经网络加速器设计. 电子与信息学报, 2019, 41(11): 2578-2584.

    19. [19]

      程帅, 孙俊喜, 曹永刚, 刘广文, 韩广良. 多示例深度学习目标跟踪. 电子与信息学报, 2015, 37(12): 2906-2912.

    20. [20]

      刘勤让, 刘崇阳. 利用参数稀疏性的卷积神经网络计算优化及其FPGA加速器设计. 电子与信息学报, 2018, 40(6): 1368-1374.

  • 图 1  本文提出的多级上下文深度卷积自编码器结构图

    图 2  不同特征表示模型在CHB-MIT和UCD数据库上的ROC和PR曲线

    图 3  不同特征学习模型在CHB-MIT数据库上对不同超参数配置的影响

    图 4  不同特征学习模型在UCD数据库上对不同超参数配置对的影响

    表 1  多元卷积自编码模块具体配置参数

    编码单元卷积层非线性变换池化层
    元内编码单元$1 \times 3 \times 16$ReLU$1 \times 2$
    元间编码单元$C \times 3 \times 8$ReLU$1 \times 2$
    解码单元反卷积层非线性变换反池化层
    元间解码单元$C \times 3 \times 8$ReLU$1 \times 2$
    元内解码单元$1 \times 3 \times 16$ReLU$1 \times 2$
    下载: 导出CSV

    表 2  CHB-MIT数据库上的方法比较结果

    方法AUC-ROCAUC-PRF1分子准确率
    PCA0.8291 ± 0.04340.7021 ± 0.08720.6421 ± 0.02230.8768 ± 0.0223
    SAE0.5934 ± 0.03770.4180 ± 0.11890.0668 ± 0.04150.7987 ± 0.0309
    CAE0.8657 ± 0.03050.7646 ± 0.08810.6277 ± 0.12460.8690 ± 0.0267
    Med2Vec0.8155 ± 0.11810.5870 ± 0.16700.6066 ± 0.23630.8351 ± 0.0359
    Skip-gram+0.9090 ± 0.03560.7467 ± 0.15400.6288 ± 0.20400.8898 ± 0.0173
    CtxFusionEEG0.9287 ± 0.03060.7833 ± 0.11470.7202 ± 0.14850.9025 ± 0.0104
    Wave2Vec0.9035 ± 0.03710.8839 ± 0.02610.8267 ± 0.01840.9210 ± 0.0099
    m-CAE0.8946 ± 0.04010.8727 ± 0.01890.8417 ± 0.01310.9324 ± 0.0058
    mCtx-CAE0.9372 ± 0.04950.8980 ± 0.03330.8493 ± 0.01910.9412 ± 0.0110
    下载: 导出CSV

    表 3  UCD数据库上的方法比较结果

    方法AUC-ROCAUC-PRF1分数准确率
    PCA0.8177 ± 0.01420.5764 ± 0.01720.5204 ± 0.02750.6193 ± 0.0638
    SAE0.7068 ± 0.13720.4965 ± 0.09510.2760 ± 0.18150.4917 ± 0.1364
    CAE0.8386 ± 0.03760.5710 ± 0.04290.5180 ± 0.07010.6208 ± 0.0961
    Med2Vec0.7479 ± 0.07960.4836 ± 0.10460.3997 ± 0.13610.5619 ± 0.0619
    Skip-gram+0.8010 ± 0.09920.5406 ± 0.09950.4342 ± 0.17310.5884 ± 0.1077
    CtxFusionEEG0.7941 ± 0.14850.6358 ± 0.07090.5171 ± 0.19940.6375 ± 0.1074
    Wave2Vec0.8161 ± 0.05070.5984 ± 0.06980.5268 ± 0.06610.6408 ± 0.0723
    m-CAE0.8446 ± 0.03610.5727 ± 0.02150.5600 ± 0.04820.6562 ± 0.0767
    mCtx-CAE0.8648 ± 0.02580.6423 ± 0.04520.5655 ± 0.02280.6734 ± 0.0562
    下载: 导出CSV
  • 加载中
图(4)表(3)
计量
  • PDF下载量:  40
  • 文章访问数:  1385
  • HTML全文浏览量:  597
文章相关
  • 通讯作者:  贾克斌, kebinj@bjut.edu.cn
  • 收稿日期:  2019-03-07
  • 录用日期:  2019-08-17
  • 网络出版日期:  2019-08-28
  • 刊出日期:  2020-02-01
通讯作者: 陈斌, bchen63@163.com
  • 1. 

    沈阳化工大学材料科学与工程学院 沈阳 110142

  1. 本站搜索
  2. 百度学术搜索
  3. 万方数据库搜索
  4. CNKI搜索

/

返回文章