Computerized artifact detection for ventilatory inductance plethysmographic apnea monitors.

Ventilatory inductive plethysmography allows noninvasive monitoring of patient ventilation. Patient movements unrelated to breathing introduce severe errors in ventilator inductive plethysmographic measurements and restrict its usefulness. The purpose of this research was to develop and test a microprocessor-based real-time digital signal processor that uses an adaptive filter to detect patient movements unrelated to breathing. The adaptive filter processor was tested for retrospective identification of artifacts in 20 male volunteers who performed the following specific movements between epochs of quiet, supine breathing: raising arms and legs (slowly, quickly, once, and several times), sitting up, breathing deeply and rapidly, and rolling from a supine to a lateral decubitus position. Flow was simultaneously measured directly with a pneumotachography attached to a mouthpiece. A multilinear regression was used to continuously calculate the calibration constants that relate the pneumotachographic and ventilatory inductive plethysmographic signals. Ventilatory inductive plethysmographic data were then processed, and results scored. There were a total of 166 movements. The calibration coefficients changed dramatically in 146 (88%) of the 166 movements. These movements would have significant errors on ventilatory inductive plethysmographic flow calculation. The changes lasted for the duration of the movements and returned to baseline within two to three breaths. The changes in the coefficients were five or more times larger than the variability around baseline during quiet, supine breathing. All of the total body movements and changes in breathing patterns were detected accurately. The filter detected 46 of 53 upper body movements, 34 of 36 lower body movements, 38 of 38 total body movements, and 19 of 19 breathing pattern changes where the calibration changed.(ABSTRACT TRUNCATED AT 250 WORDS)