Automating the correction of flow integration drift during whole-body plethysmography

Prolonged measurement of total body volume variations (deltaVb) with whole-body, flow-based plethysmography (WBP) results in a drift of the signal due to changes in temperature and humidity inside the plethysmograph and to numerical integration of the flow to obtain deltaVb. This drift has been previously corrected with the application of a wavelet- based filter using visual inspection of the signal to select the optimal filter level (Uva et al. Front. Physiol. 6:411, 2016), thus introducing potential operator bias. To exclude the latter we compared this approach with a newly developed automatic method based on (1) correction for actual changes in temperature and humidity inside the plethysmograph (algorithm TH) and (2) automatic selection of the wavelet filter level based on comparison between deltaVb and intra-thoracic and abdominal pressure variations measured simultaneously (algorithm WAV). The Pearson's correlation coefficient between deltaVb and the changes in volume of the chest wall (deltaVcw) simultaneously obtained by optoelectronic plethysmography (OEP) was calculated after correction of deltaVb with TH and WAV applied separately, TH and WAV applied consecutively (TH+WAV), manual selection of a wavelet filter based on visual inspection (MAN) or no correction (CTRL). The correlation between deltaVb and deltaVcw increased marginally with WAV, TH+WAV and MAN compared to CTRL (P <; 0.01). Conversely, TH alone yielded a lower correlation (P <; 0.01). It follows that while the automated wavelet filter level selection method (WAV) represents an effective, operator-independent method for the correction of deltaVb, whether or not it is combined with specific correction for changes in thermodynamic conditions inside the plethysmograph, the manual method (MAN) yields satisfactory results without the constraints of intra-thoracic and abdominal pressure measurement.