基于惯性传感器的帕金森患者行走能力评估系统

摘要

针对如何客观地评估帕金森病人在行走方面的能力问题,构建了一套基于惯性传感器的帕金森病人行走能力评估系统.将惯性传感器单元分别固定在被测对象的背部、腰部采集测试者在运动过程中的运动数据,采用Kalman融合算法对运动数据计算后提取出特征参数.该文设计了多类SVM分类器,利用特征信号对分类器进行特征训练.测试结果表明,对帕金森病人和正常人的识别以及对帕金森病人的行走能力分级具有较高的识别率,能够辅助医生给出更加客观的诊断结论.%In order to evaluate the ability of Parkinson's patients to walk comprehensively, a system based on MEMS to aid clinical quantification of ability in Parkinson's is established. The inertial units are respectively fixed on the back and the waist of subject to be measured. The Kalman fusion algorithm is used to extract the characteristic parameters of accelerometer and gyroscope data. SVM classifier is designed to train and test the classifier by the feature. The results show that the system possesses a high recognition rate for Parkinson's patients and normal subjects and for the classification of the walking ability of patients with Parkinson's disease. So, this system can aid doctors to give more object diagnostic conclusion.