Impairing Factors in Remote-PPG Pulse Transit Time Measurements on the Face

Andreia Moco, Sander Stuijk, Mark van Gastel, Gerard de Haan; Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) Workshops, 2018, pp. 1358-1366

The time it takes for a pulse wave to propagate between two arterial sites--i.e., the pulse transit time (PTT)--has received considerable attention as marker of aortic stiffness and as surrogate for blood pressure. However, obtrusiveness and manual intervention requirements render conventional PTT measurement methods inappropriate for ubiquitous monitoring. In this regard, high-speed camera systems are interesting alternatives. Recognizably, a technical breakthrough would be estimating PTT with a relatively inexpensive RGB camera pointed at the face only. A simple means to do this is determining the phase shift (PS) between photoplethysmographic (PPG) signals extracted at collocated skin pixels. In this paper, we show that the validity of this approach is threatened by skin variability. We analysed simultaneous video recordings of the neck and face in 21 subjects (ages, 33 +-11 yrs). These were used to extract PPG signals at the face and skin motion (sMOT) signals at the vicinity of the carotid artery. Using sMOT as reference signal, we show that the pressure wave undergoes delay and frequency leakage as it propagates across the arterial tree; the extent of propagation distortion is subject-dependent and place PPG-based estimations at a disadvantage in comparison with PTTs measured at the arterial level. Awareness is further raised for the site-dependency of PS outcomes by the provision of facial PPG-phase maps and collocated PPG signals. Lastly, impairments due to waveform dissimilarity are demonstrated under exercise-induced PTT changes. In conclusion, PS is unsuitable for PTT measurements at the face.