On-the-fly respiratory cycle estimation method based on photoplethysmography waveform morphology analysis

It is a well-known fact that the raw photoplethysmography signal contains a slowly variable respiratory component which can be extracted by a low-frequency filtration or spectral decomposition. However, these methods do not allow to determine the current phase of the respiratory cycle, i.e. inhalation, exhalation, or plateau phase. And yet these often subtle differences in the characteristics of the individual phases of breath cycle and their interactions have a clinically significant diagnostic value. During the work, it was noted that the chest movements during respiratory cycle contribute to the shape of the PPG curve – especially on the descending part of the photoplethysmographic waveform. This effect was used to develop a method for evaluating a phase of respiratory cycle on-the-fly. The proposed method is based on the temporal variation of the three characteristic points (tPA, tPB and tPC) of the normalized PPG curve. The tPA designator is the reference point for an algorithm that normalises the amplitude of the PPG signal. Points tPB and tPC (in particular their variability over time) are the basis for further analysis that leads to extraction of a phase of respiratory cycle. Strong correlation between the respiratory curve and the temporal variation of the tPB and tPC points allows to assess the respiratory cycle phase on-the-fly.