Strong corruption of electrocardiograms caused by cardiopulmonary resuscitation reduces efficiency of two-channel methods for removing motion artefacts in non-shockable rhythms
There is no author summary for this article yet. Authors can add summaries to their articles on ScienceOpen to make them more accessible to a non-specialist audience.
Abstract
Cardiopulmonary resuscitation (CPR) artefact removal methods provide satisfactory
results when the rhythm is shockable but fail on non-shockable rhythms. We investigated
the influence of the corruption level on the performance of four different two-channel
methods for CPR artefact removal.
395 artefact-free ECGs and 13 pure CPR artefacts with corresponding blood pressure
readings as a reference channel were selected. Using a simplified additive data model
we generated CPR-corrupted signals at different signal-to-noise ratio (SNR) levels
from -10 to +10 dB. The algorithms were optimized on learning data with respect to
SNR improvement and then applied to testing data. Sensitivity and specificity were
derived from the shock/no-shock advice of an automated external defibrillator before
CPR corruption and after artefact removal.
Sensitivity for the filtered data (>95%) was significantly superior to that for the
unfiltered data (76%), p<0.001. However, specificity was similar for the filtered
and unfiltered data (<90% vs 89.3%). For large artefacts (-10 dB) specificity decreased
below 70%. No important difference in the performance of the four algorithms was found.
Using a simplified data model we showed that, when the ECG rhythm is non-shockable,
two-channel methods could not reduce CPR artefacts without affecting the rhythm analysis
for shock recommendation. The reason could be poor reconstruction when the artefacts
are large. However, poor reconstruction was not a hindrance to re-identifying shockable
rhythms. Future investigations should both include the refinement of filter methods
and also focus on reducing motion artefacts already at the recording stage.