基于体动射频信号的睡眠分期  

Identification of Sleep Stages Based on Body Movement Signals from RFMS

作　　者：徐礼胜[1,2] 薄红瑞 赵金野 窦元珠 

机构地区：[1]东北大学中荷生物医学与信息工程学院,辽宁沈阳110169 [2]东北大学教育部医学影像计算重点实验室,辽宁沈阳110169 [3]熙康阿尔卑斯科技有限公司,辽宁沈阳110179

出　　处：《东北大学学报（自然科学版）》2016年第8期1079-1083,共5页Journal of Northeastern University(Natural Science)

基　　金：国家自然科学基金资助项目(61374015;61202258);辽宁省自然科学基金资助项目(201102067);教育部高等学校博士学科点专项科研基金资助项目(20110042120037);中央高校基本科研业务费专项资金资助项目(N110219001;N130404016)

摘　　要：为了减轻传统接触式睡眠生理监测系统对人体造成的负担,设计了一种基于微波技术的非接触式睡眠生理信号采集与分析系统,提出一种基于体动射频信号的睡眠分期识别算法.通过小波变换对射频运动传感器(RFMS)采集的体动信号进行预处理,再计算出体动信号的能量值,最后通过判别式处理和阈值法实现了睡眠分期:醒觉期、浅睡期、中睡期、深睡期.实验采集分析了8个实验者为期46天的睡眠生理信号,同时同步采集视频信息、TANITA水床睡眠信息、接触式呼吸脉搏信号.与视频结果比较发现醒觉期正确率达到90%;与TANITA水床睡眠结果相比,本系统的结果与其吻合程度达到70%;与不同睡眠状态下呼吸率、心率的变化相比,本系统的结果吻合度达到80%.To reduce the burden of the traditional sleep monitoring system on human body,a noncontact sleep monitoring system based on microwave technique was developed,and a sleep stages recognition algorithm was proposed based on body movement signals from radio frequency motion sensors（RFMSs）.The RFMS signals were processed by wavelet transform,then,the energy of signals was calculated by integral algorithm.Finally,the sleep stages such as wakefulness,light sleep,moderate sleep,deep sleep were recognized by discriminant analysis and threshold processing.The RFMS signals from 8 normal subjects,whose sleep were monitored for 46 days,were acquired,with their video information,TANITA sleep information,pulse wave,and respiration signals gathered synchronously.Compared with videos,the staging results of this system match 90% in wakefulness stage.In comparison with the results of TANITA,the results of the RFMS match 70% in other sleep stages.The results match 80% in comparison with the change modes of respiratory rate and heart rate at different sleep stages.

关 键 词：非接触 睡眠监测 睡眠分期 体动 射频 

分 类 号：TP274.1[自动化与计算机技术—检测技术与自动化装置]