Reduction of motion artifacts using a two-frequency impedance plethysmograph and adaptive filtering.

We measured transthoracic impedance in nine presumed healthy adult subjects with a two-frequency plethysmograph at 57 kHz and 185 kHz. The measurement protocol included periods of normal breathing without motion and periods of motion without breathing. We analyzed the cross-correlation and the ratio between the signals at both frequencies for all the different maneuvers. The correlation coefficient was between 0.97 and 1 for breathing, the minimal cross-correlation (0.81) was for simulated obstructive apnea. We found that the amplitude ratio between the two-frequency signals was different for normal breathing and for motion. Based on these results, we designed and tested an adaptive filter to increase the signal-to-artifact ratio (SAR). The increase in SAR (mean +/- standard deviation) compared with the signal at 57 kHz was: 183% +/- 117% for arm movement, 133% +/- 93% for leg movement, and 34% +/- 62% for simulated obstructive apnea.