Unconstrained measurement of cardiorespiratory and body movind episodes for infant monitoring

我们提出了一个简单的系统，用于测量呼吸率（RR），心率（HR），和身体运动事件的婴儿健康监测。该系统由八个聚偏氟乙烯（PVDF）电缆传感器，一个变化放大器，和测量软件与CCD相机单元。电缆传感器水平分布到床表面，电缆之间的距离约为8cm，覆盖身体的上半部。传感器通过呼吸活动检测胸腔位移，通过心跳检测弹道运动。系统软件基于相应输出的功率谱密度的幅度搜索用于测量呼吸和心跳的最佳传感器。专门构造的FIR数字滤波器提取由呼吸和心脏动作引起的信号。然后将滤波后的心脏信号通过全整流积分滤波器。计算滤波后信号的自相关函数（ACF）并求其微分，以检测各个ACF中的峰值。如果受试者改变姿势，则软件通过CCD相机捕获带有时间戳的身体位置的图像，并调用例程以进行最佳传感器选择。为验证该方法的有效性，将该监护仪测得的HR和RR与同时测量呼吸流量和ECG波形的结果进行了比较。在大多数情况下，该系统成功地实时提取了呼吸和心脏动作。平均HR和RR的个体误差均小于10%。这些实验结果表明，所提出的系统是可行的，在家庭或医院的组合长期监测。

We proposed a simple system for measuring respiratory rate (RR), heart rate (HR), and body moving episodes for infant health monitoring. The system consists of eight polyvinylidene fluoride (PVDF) cable sensor, a change amplifier, and measuring software with CCD camera unit. The cable sensors were horizontall distributed to the bed surface with an inter-cable distance of about 8-cm, covering the upper half of the body. The sensors detect thorax displacement by respiratory activity and ballistic movement by heartbeat. The system software searches the optimal sensors for measuring respiration and heartbeat based on the amplitude of the power spectral density of the respective outputs. The FIR digital filters that were specifically constructed extract the signals caused by respiratory and cardiac actions. The filtered cardiac signal was then passed through a full rectified integrate filter. The autocorrelation functions (ACFs) of the filtered signals were computed and differentiated to detect the peaks in the respective ACFs. If a subject changes a posture, the software captures the image of the body position with a time stamp by a CCD camera and calls the routine for the optimal sensor seletion. To demonstrate the validity of this method, HR and RR obtained by this monitor were compared with those measured simultaneously with respiratory flow and ECG wave. In most cases, the system successfully extracted the respiratory and cardiac action on a real-time basis. The individual errors of average HR and RR were less than 10%. These experimental results demonstrate that the proposed system is feasible for the combined long-term monitoring at home or hospital.