Wang Qian, Yang Ping, Zhang Yuanting
Key Laboratory of Biomedical information and Health Engineering, Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences, China.
Annu Int Conf IEEE Eng Med Biol Soc. 2010;2010:959-62. doi: 10.1109/IEMBS.2010.5627581.
The pulstile components of photoplethysmography (PPG) contain valuable information about a subject's cardiovascular and metabolic systems. Pulse rate is one of the most significant vital signs that can be extracted from PPG signals. However, patient movement, especially movement at the measurement sites, such as fingers, can disturb the PPG's light path significantly, resulting in corrupted measurements. In this paper, a method is proposed for removing motion artifacts from PPG recordings. In this method, the Empirical Mode Decomposition (EMD) and Hilbert transform are used together to decompose PPG recordings into instantaneous frequency series on different scales of resolution. Motion artifacts and physiological signals are separated based on these series. The proposed method was used to recover PPG signals recorded in an experiment, where motion artifacts were intentionally introduced by finger bending. By using our method, the signal-to-noise ratio was increased from 0.078 dB of the contaminated signals to 0.318 dB, and the true detection rate of heartbeats was improved from 59.2% to 96.6%. The results demonstrated that the EMD combined with Hilbert transform has great potential in reducing motion artifacts in PPG signals and can improve the accuracy of heartbeat detection.
光电容积脉搏波描记法（PPG）的搏动成分包含有关受试者心血管和代谢系统的有价值信息。脉搏率是可以从PPG信号中提取的最重要生命体征之一。然而，患者的运动，尤其是测量部位（如手指）的运动，会严重干扰PPG的光路，导致测量结果失真。本文提出了一种从PPG记录中去除运动伪影的方法。在该方法中，经验模态分解（EMD）和希尔伯特变换一起用于将PPG记录分解为不同分辨率尺度上的瞬时频率序列。基于这些序列分离运动伪影和生理信号。所提出的方法用于恢复在实验中记录的PPG信号，在该实验中通过手指弯曲故意引入运动伪影。通过使用我们的方法，信噪比从受污染信号的0.078 dB提高到0.318 dB，心跳的真实检测率从59.2%提高到96.6%。结果表明，EMD与希尔伯特变换相结合在减少PPG信号中的运动伪影方面具有很大潜力，并且可以提高心跳检测的准确性。
