基于MEMS的帕金森病震颤实时评估系统

A real- time system based on MEMS to evaluate tremor in Parkinson′s disease

针对如何全面、客观地提取出帕金森病人震颤信息的问题,构建了一套基于MEMS惯性传感器的帕金森病震颤实时分类与量化评估系统。将3组惯性传感器单元(IMU)分别固定在被测对象的大腿、胸腔和手腕上,上位机中基于LabVIEW平台设计特定的算法实现对4种特定人体姿态的识别及震颤信号的分析,提取特征参数。设计了二叉决策树特征分类器,利用特征信号对分类器进行特征训练。算法验证试验结果表明,系统针对4种特定人体姿态和两种震颤状态具有较高的识别率,震颤评估参数具有一定的合理性,能够辅助医生给出更加客观的诊断结论。

In order to extract the tremor information comprehensively in patients with Parkinson ′ s disease and assist doctors to give more scientific clinical diagnosis conclusion, a real- time system based on MEMS to aid clinical classification and quantification of tremor in Parkinson ＇ s disease is established. Three sets of Inertial Measurement Units（ IMU） are proposed to be fixed on the object＇s thigh, chest and wrist. A series of algorithms based on Lab VIEW platform in the computer are introduced to identify four kinds of human gestures, achieve the analysis of the tremor signal and extract the character parameters. These parameters are used to design and train a binary decision tree classifier. The algorithm verification test shows that the system has good recognition rate for static posture and tremor type and the tremor evaluation parameters are reasonable.