Detection of movement artifact in recorded pulse oximeter saturation

Movement artifact (MA) must be detected when analysing recordings of pulse oximeter saturation (SpO(2)). Visual analysis of individual pulse waveforms is the safest, but also the most tedious, method for this purpose. We wanted to test the reliability of a computer algorithm (Edentec Motion Annotation System), based on a comparison between pulse and heart rate, for MA detection. Ten 12-h recordings of SpO(2), pulse waveforms and heart rate from ten preterm infants were analysed for the presence of MA on the pulse waveform signal. These data were used to determine the sensitivity and specificity of the computer algorithm, and of the oximeter itself, in detecting MA. Recordings were divided into segments of 2.5 s duration to compare the movement identification methods. Of the segments 31% +/- 6% (mean +/- SD) contained MA. The computer algorithm identified 95% +/- 3% of these segments, the pulse oximeter only 18% +/- 11%. Specificity was 85% +/- 4% and 99% +/- 0%, respectively. SpO(2) was less than or equal to 80% in 3% +/- 1% of segments. 88% +/- 7% of the pulse waveform signal showed MA during this time, leaving a significant potential for erroneous identification of hypoxaemia. Recordings of SpO(2) do not allow a reliable identification of MA. Conclusion Without additional information about movement artifact, a significant proportion of recording time of pulse oximeter signal may be regarded as demonstrating hypoxaemia which, in fact, simply reflects poor measurement conditions. The computer algorithm used in this study identified periods of movement artifact reliably.