Wijshoff, Ralph;
Haarburger, Christoph;
Muehlsteff, Jens;
Noordergraaf, Gerrit Jan
Abstract 12123: Periodicity-Analysis of a Photoplethysmography Signal May Support Pulse Detection in a Few-Second Compression-Pause During Cardiopulmonary Resuscitation
You can manage bookmarks using lists, please log in to your user account for this.
Media type:
E-Article
Title:
Abstract 12123: Periodicity-Analysis of a Photoplethysmography Signal May Support Pulse Detection in a Few-Second Compression-Pause During Cardiopulmonary Resuscitation
Contributor:
Wijshoff, Ralph;
Haarburger, Christoph;
Muehlsteff, Jens;
Noordergraaf, Gerrit Jan
Description:
<jats:p>
<jats:bold>Introduction:</jats:bold>
Pulse checks during cardiopulmonary resuscitation (CPR) are typically performed by manual palpation, which is unreliable and can interrupt compressions for longer than the recommended 10 s.
</jats:p>
<jats:p>
<jats:bold>Hypothesis:</jats:bold>
Presence of a spontaneous pulse can be detected in a compression-free photoplethysmography (PPG) signal within 5 s, by assessing the periodicity of the signal.
</jats:p>
<jats:p>
<jats:bold>Methods:</jats:bold>
We measured nasal PPG signals in 11 pigs during periods of 10-min baseline, arrest induction, 20-min 30:2 CPR, defibrillation, and 20-min post return of spontaneous circulation (ROSC), with aortic pressure and thoracic impedance as references.
</jats:p>
<jats:p>We annotated pulse presence and obtained compression-free windows of PPG signals from the ventilation pauses and by segmenting the baseline and post-ROSC phases in 4-s non-overlapping windows.</jats:p>
<jats:p>Pulse presence was detected by assessing signal periodicity in the autocorrelation function (ACF), which compared a window of the PPG signal to time shifts of this window. Windows with pulse had local maxima in the ACF when a time shift aligned the pulses in both windows (Fig. 1a). Pulseless windows had unstructured ACFs (Fig. 1b). We assessed periodicity via the maximum prominence of all peaks at non-zero shift in the ACF. Prominence is the amplitude relative to the largest surrounding minimum (Fig. 1a,b). We used the prominence and related features in a logistic regression classifier to quantify pulse absence and presence on a continuous scale from 0 to 1 (Fig. 1c). We trained and validated the classifier via five-fold cross-validation, regarding a classifier output of ≥0.5 as indicative of pulse presence.</jats:p>
<jats:p>
<jats:bold>Results:</jats:bold>
Sensitivity was 97.1% and specificity was 96.2%.
</jats:p>
<jats:p>
<jats:bold>Conclusions:</jats:bold>
Periodicity analysis allows for pulse detection in PPG signals within 5 s. This could be used during existing ventilation pauses to prepare for a faster pulse check or avoid an unnecessary pulse check.
</jats:p>
<jats:p>
<jats:bold>Fig. 1:</jats:bold>
(a) ACF for pulse presence. (b) ACF for pulse absence. (c) Classifier output.
</jats:p>
<jats:p>
<jats:graphic xmlns:xlink="http://www.w3.org/1999/xlink" orientation="portrait" position="float" xlink:href="g12123.jpeg" />
</jats:p>