Photoplethysmography signal quality assessment using attractor reconstruction analysis

Abstract

The Photoplethysmography (PPG) has already proved its importance as an auxiliary vital signal measurement technique. The PPG is predominantly a quasi-periodic signal for small timescales. This property allows the reconstructing of a phase space and, therefore, the attractor of PPG segments which presents good insights on dynamics of physiological systems. However, the assessment of PPG waveforms is a difficult task due to motion artifacts, biological tissue differences and other sources of noise. In the same direction, protocols for PPG signals acquisition have its development in a slower step. To contribute against this scenario, this work proposes and describes a method using the attractor reconstruction analysis to provide a quality assessment of PPG signals. The method presented good results in classifying into good and bad quality PPG signals segments achieving F1-score of 97%. The method is totally analytical, thus, it is suitable for real-time applications. No training process is required, it is independent of the subject physiology and no calibration is needed.