Research Article
Dynamics Reconstruction of Remote Photoplethysmography
@INPROCEEDINGS{10.1007/978-3-030-99194-4_8, author={Lin He and Kazi Shafiul Alam and Jiachen Ma and Richard Povinelli and Sheikh Iqbal Ahamed}, title={Dynamics Reconstruction of Remote Photoplethysmography}, proceedings={Pervasive Computing Technologies for Healthcare. 15th EAI International Conference, Pervasive Health 2021, Virtual Event, December 6-8, 2021, Proceedings}, proceedings_a={PERVASIVEHEALTH}, year={2022}, month={3}, keywords={Remote photoplethysmography Phase space reconstruction Heart rate Heart rate variability}, doi={10.1007/978-3-030-99194-4_8} }- Lin He
Kazi Shafiul Alam
Jiachen Ma
Richard Povinelli
Sheikh Iqbal Ahamed
Year: 2022
Dynamics Reconstruction of Remote Photoplethysmography
PERVASIVEHEALTH
Springer
DOI: 10.1007/978-3-030-99194-4_8
Abstract
Photoplethysmography based medical devices are widely used for cardiovascular status monitoring. In recent years, many algorithms have been developed to achieve cardiovascular monitoring results comparable to the medical device from remote photoplethysmography (rPPG). rPPG is usually collected from the region of interest of the subject face and has been used for heart rate detection. Though there were many works on the study of chaos dynamics of PPG, very few are on the characteristics of the rPPG signal. The main purpose of this study is to discover rPPG dynamics from nonlinear signal processing techniques, which may provide insight for improving the accuracy of cardiovascular status monitoring. Univ. Bourgogne Franche-Comté Remote PhotoPlethysmoGraphy dataset is used for the experiment. The results show rPPG is considered as chaotic. The best-estimated embedding dimension for the rPPG signal is between 3 to 4. The time delay is 10 for an interpolated 240 Hz rPPG signal. The interpolation process will increase the complexity level and reduce the correlation dimension of the rPPG. The bandpass filtering process will reduce the complexity level and the correlation dimension of the rPPG. Introducing the features derived from reconstructed phase space such as Lyanpunov exponent, correlation dimension and approximate entropy, could improve the accuracy of heart rate variability detection from rPPG.
