Mikel Leturiondo

Influence of chest compression artefact on capnogram-based ventilation detection during out-of-hospital cardiopulmonary resuscitation

BACKGROUND Capnography has been proposed as a method for monitoring the ventilation rate during cardiopulmonary resuscitation (CPR). A high incidence (above 70%) of capnograms distorted by chest compression induced oscillations has been previously reported in out-of-hospital (OOH) CPR. The aim of the study was to better characterize the chest compression artefact and to evaluate its influence on the performance of a capnogram-based ventilation detector during OOH CPR. METHODS Data from the MRx monitor-defibrillator were extracted from OOH cardiac arrest episodes. For each episode, presence of chest compression artefact was annotated in the capnogram. Concurrent compression depth and transthoracic impedance signals were used to identify chest compressions and to annotate ventilations, respectively. We designed a capnogram-based ventilation detection algorithm and tested its performance with clean and distorted episodes. RESULTS Data were collected from 232 episodes comprising 52 654 ventilations, with a mean (±SD) of 227 (±118) per episode. Overall, 42% of the capnograms were distorted. Presence of chest compression artefact degraded algorithm performance in terms of ventilation detection, estimation of ventilation rate, and the ability to detect hyperventilation. CONCLUSION Capnogram-based ventilation detection during CPR using our algorithm was compromised by the presence of chest compression artefact. In particular, artefact spanning from the plateau to the baseline strongly degraded ventilation detection, and caused a high number of false hyperventilation alarms. Further research is needed to reduce the impact of chest compression artefact on capnographic ventilation monitoring.

Enhancement of capnogram waveform in the presence of chest compression artefact during cardiopulmonary resuscitation

BACKGROUND Current resuscitation guidelines emphasize the use of waveform capnography to help guide rescuers during cardiopulmonary resuscitation (CPR). However, chest compressions often cause oscillations in the capnogram, impeding its reliable interpretation, either visual or automated. The aim of the study was to design an algorithm to enhance waveform capnography by suppressing the chest compression artefact. METHODS Monitor-defibrillator recordings from 202 patients in out-of-hospital cardiac arrest were analysed. Capnograms were classified according to the morphology of the artefact. Ventilations were annotated using the transthoracic impedance signal acquired through defibrillation pads. The suppression algorithm is designed to operate in real-time, locating distorted intervals and restoring the envelope of the capnogram. We evaluated the improvement in automated ventilation detection, estimation of ventilation rate, and detection of excessive ventilation rates (over-ventilation) using the capnograms before and after artefact suppression. RESULTS A total of 44 267 ventilations were annotated. After artefact suppression, sensitivity (Se) and positive predictive value (PPV) of the ventilation detector increased from 91.9/89.5% to 98.0/97.3% in the distorted episodes (83/202). Improvement was most noticeable for high-amplitude artefact, for which Se/PPV raised from 77.6/73.5% to 97.1/96.1%. Estimation of ventilation rate and detection of over-ventilation also upgraded. The suppression algorithm had minimal impact in non-distorted data. CONCLUSION Ventilation detection based on waveform capnography improved after chest compression artefact suppression. Moreover, the algorithm enhances the capnogram tracing, potentially improving its clinical interpretation during CPR. Prospective research in clinical settings is needed to understand the feasibility and utility of the method.

Monitoring chest compression quality during cardiopulmonary resuscitation: Proof-of-concept of a single accelerometer-based feedback algorithm

Background The use of real-time feedback systems to guide rescuers during cardiopulmonary resuscitation (CPR) significantly contributes to improve adherence to published resuscitation guidelines. Recently, we designed a novel method for computing depth and rate of chest compressions relying solely on the spectral analysis of chest acceleration. That method was extensively tested in a simulated manikin scenario. The purpose of this study is to report the results of this method as tested in human out-of-hospital cardiac arrest (OHCA) cases. Materials and methods The algorithm was evaluated retrospectively with seventy five OHCA episodes recorded by monitor-defibrillators equipped with a CPR feedback device. The acceleration signal and the compression signal computed by the CPR feedback device were stored in each episode. The algorithm was continuously applied to the acceleration signals. The depth and rate values estimated every 2-s from the acceleration data were compared to the reference values obtained from the compression signal. The performance of the algorithm was assesed in terms of the sensitivity and positive predictive value (PPV) for detecting compressions and in terms of its accuracy through the analysis of measurement error. Results The algorithm reported a global sensitivity and PPV of 99.98% and 99.79%, respectively. The median (P75) unsigned error in depth and rate was 0.9 (1.7) mm and 1.0 (1.7) cpm, respectively. In 95% of the analyzed 2-s windows the error was below 3.5 mm and 3.1 cpm, respectively. Conclusions The CPR feedback algorithm proved to be reliable and accurate when tested retrospectively with human OHCA episodes. A new CPR feedback device based on this algorithm could be helpful in the resuscitation field.

Enhancing ventilation detection during cardiopulmonary resuscitation by filtering chest compression artifact from the capnography waveform

Background During cardiopulmonary resuscitation (CPR), there is a high incidence of capnograms distorted by chest compression artifact. This phenomenon adversely affects the reliability of automated ventilation detection based on the analysis of the capnography waveform. This study explored the feasibility of several filtering techniques for suppressing the artifact to improve the accuracy of ventilation detection. Materials and methods We gathered a database of 232 out-of-hospital cardiac arrest defibrillator recordings containing concurrent capnograms, compression depth and transthoracic impedance signals. Capnograms were classified as non-distorted or distorted by chest compression artifact. All chest compression and ventilation instances were also annotated. Three filtering techniques were explored: a fixed-coefficient (FC) filter, an open-loop (OL) adaptive filter, and a closed-loop (CL) adaptive filter. The improvement in ventilation detection was assessed by comparing the performance of a capnogram-based ventilation detection algorithm with original and filtered capnograms. Results Sensitivity and positive predictive value of the ventilation algorithm improved from 91.9%/89.5% to 97.7%/96.5% (FC filter), 97.6%/96.7% (OL), and 97.0%/97.1% (CL) for the distorted capnograms (42% of the whole set). The highest improvement was obtained for the artifact named type III, for which performance improved from 77.8%/74.5% to values above 95.5%/94.5%. In addition, errors in the measurement of ventilation rate decreased and accuracy in the detection of over-ventilation increased with filtered capnograms. Conclusions Capnogram-based ventilation detection during CPR was enhanced after suppressing the artifact caused by chest compressions. All filtering approaches performed similarly, so the simplicity of fixed-coefficient filters would take advantage for a practical implementation.