北京生物医学工程

基于多元多尺度模糊熵的帕金森步态信号分类

Signal classification method for Parkinson gait based on multivariate multiscale fuzzy entropy

作者：王旭尧徐永红

单位：燕山大学生物医学工程研究所(河北秦皇岛066004)

分类号：R318.04

出版年·卷·期（页码）：2016·35·5（489-496）

摘要：

目的传统多元多尺度熵在量化有限长数据时会造成部分数据丢失，同时传统算法对阈值的过分依赖也会造成整个系统产生不稳定的现象，二者皆会使最终结果产生较大的误差，因此本文提出一种多元多尺度模糊熵算法。方法对传统多元多尺度样本熵的粗粒化方式进行改进，采用滑动均值滤波使粗粒化后各尺度上的时间序列与原始时间序列长度一致，减小了所计算多元多尺度熵的离散性。此外，本文算法在保持多元样本熵中硬阈值优点的同时，通过定义模糊隶属度函数来统计两复合延迟向量距离略大于阈值的情况。结果本算法既降低了传统方法对阈值的依赖性，又很好地解决了传统阈值所导致的不稳定现象。最后用仿真数据对该算法进行验证，并将其应用于帕金森患者步态复杂度的评价和分类。结论实验结果表明多元多尺度模糊熵的识别效果明显优于传统多元多尺度熵。

Objective Traditional multivariate multiscale entropy may cause some loss of data when quantifying finite length data. Meanwhile, the over-reliance of traditional algorithm on threshold may cause the entire system produces an unstable phenomenon. Both of the two problems can lead large errors in the results. Therefore, this paper presents multivariate multiscale fuzzy entropy. Methods The method improves coarse-grained way of the traditional algorithm and makes coarse-grained time series equal to the length of original time series on each scale by sliding mean filter, and reduces the discreteness of multivariate multiscale entropy. In addition, the algorithm maintains the advangtage of hard threshold in traditional method and counts the distance of two composite delay vectors slightly greater than threshold value by defining fuzzy membership function. Results This method not only reduces the dependence of threshold in traditional algorithm, but also solves the instability caused by traditional threshold. Finally, the algorithm is validated in the simulation data and is applied to the evaluation and classification on gait complexity of Parkinson patients. Conclusions The recognition effect of multivariate multiscale fuzzy entropy is better than traditional multivariate multiscale entropy.

参考文献：

［1］中华医学会神经病学分会运动障碍及帕金森病学组. 帕金森病的诊断［J］.中华神经科杂志，2006, 39(6):408-409.

［2］Jankovic J. Parkinson’s disease: clinical features and diagnosis［J］.J Neurol Neurosurgery Psychiatry, 2008, 79(4):368-376.

［3］袁立伟，王健.帕金森病患者的步态特征［J］.中国康复医学杂志,2010,25(6):586-588.

［4］Michael T, Martin A , Anke H ,et al. Gait and upper limb variability in Parkinson’s disease patients with and without freezing of gait［J］.J Neurol, 2014, 261(2):330-342.

［5］Jorik N , Arno M , Janssen S, et al.Short rapid steps to provoke freezing of gait in parkinso-n’s disease［J］. J Neurol, 2014, 261(9): 1763-1767.

［6］刘晓婷,曹玉珍.基于加速度的人体步态信息多尺度熵研究［D］.天津：天津大学，2008:19-30.

［7］Babu GS, Suresh S, Sangumathi KU, et al. A projection based learning meta-cognitive RBF network classifier for effective diagnosis of Parkinson’s disease［J］. ISNN, 2012, 7368(2):611-620.

［8］Luca P, Sabato M, Guido A, et al. Classification of early-mild subjects with parkinsons’s disease by using sensor-based measures of posture gait and transitions［J］. AIME, 2013, 7885(6)： 176-180.

［9］Acharya U, Molinari F, Sree S, et al. Automated diagnosis of epileptic EEG using entropies［J］. Biomedical Signal Processing and Control, 2012,7(1):401-408.

［10］徐永红，崔杰，洪文学,等.基于改进多元多尺度熵的癫痫脑电信号自动分类［J］.生物医学工程学杂志，2015,32(2):256-262.

Xu Yonghong, Cui Jie, Hong Wenxue, et al. Automatic classification of epileptic electroencephalogram signal based on improved multivariate multiscale entropy［J］. Journal of Biomedical Engineering,2015,32(2):256-262.

［11］Huang NE, Shen Z, Long SR, et al. The empirical mode decomposition and the Hilbert spectrum for nonlinear and non-stationary time series analysis［J］. Proceedings of the Royal Society Mathematical, Physical and Engineering Sciences, 1998, 454(1971): 903-995.

［12］Costa M, Peng CK, Goldberger AL, et al. Multiscale entropy analysis of human gait dynamics［J］. Physica A, 2003, 330: 53-60.

［13］Abed K, Mohammad R. HMM for classification of Parkinson’s disease based on the raw gait data［J］. J Med Syst, 2014, 38(147):2-6.

［14］Yunfeng W, Krishnan S. Statistical analysis of gait rhythm in patients with Parkinson’s disease［J］. Neural Systems and Rehabilitation Engineering, 2010, 18(2):150-158.

服务与反馈：

【文章下载】【加入收藏】

提示：您还未登录，请登录！点此登录