Separation of cardiogenic oscillations from airflow waveforms using singular spectrum analysis. (June 2022)

• Record Type: Journal Article
• Authors: Pagano, Parwane P.; Ciaccio, Edward J.; Garan, Hasan
• Is Part Of: Computer methods and programs in biomedicine. Volume 220(2022)
• Publication Date: 2022-06
• Subjects: Cardiogenic oscillation -- Mechanical ventilation -- Pressure support ventilation -- Ventilator autotriggering -- Signal separation -- Singular spectrum analysis; Medicine -- Computer programs -- Periodicals; Biology -- Computer programs -- Periodicals; Computers -- Periodicals; Medicine -- Periodicals; Médecine -- Logiciels -- Périodiques; Biologie -- Logiciels -- Périodiques; Biology -- Computer programs; Medicine -- Computer programs; Periodicals; Electronic journals; 610.28
• DOI: 10.1016/j.cmpb.2022.106803
• Languages: English
• ISSNs: 0169-2607

Abstract

Highlights: Airflow fluctuations caused by cardiac contraction can trigger inappropriate ventilator pressure support in anesthesia machines and intensive care unit mechanical ventilators. Singular spectrum analysis (SSA) was implemented to remove cardiogenic oscillations from ventilator airflow waveforms recorded from intubated, mechanically ventilated patients under general anesthesia. SSA is effective in extracting higher amplitude respiratory excursions while excluding lower amplitude cardiogenic oscillations and noise from the airflow signal. Abstract: Background and Objective: Airflow fluctuations caused by cardiac contraction can trigger inappropriate ventilator pressure support in anesthesia machines and intensive care unit mechanical ventilators. Removal of this cardiogenic artifact from the airflow signal would improve ventilator function. The application of singular spectrum analysis (SSA) to remove cardiogenic oscillations from ventilator airflow signals recorded from intubated, mechanically ventilated patients under general anesthesia was evaluated in this study. Methods: Airflow (liters/minute) and CO2 (mmHg) data were collected at a sampling rate of 125 Hz from the intraoperative monitoring systems using special-purpose software. Simultaneous electrocardiogram signals (mV) were also collected at a sampling rate of 250 Hz. One-dimensional SSA was performed offline on normalized airflow signals using a window length sufficient to span one period of typical respiratory variation. The main components of the airflow waveform are respiratory excursions and cardiogenic oscillations, with respiratory excursions more slowly varying and of higher magnitude. The smooth respiratory waveform was formed from elementary reconstructed series corresponding to the highest singular values obtained with SSA analysis. The quality of respiratory waveform extraction with SSA was determined by calculating the weighted correlation between the selected elementary reconstructed series. Results: Airflow data was recorded from 6 patients. The respiratory component of the airflow signal without cardiogenic oscillations was reconstructed from elementary series corresponding to singular values of highest magnitude. The weighted correlations obtained were greater than 0.96 in the majority of patients studied. Cardiogenic oscillations were reconstructed from elementary reconstructed series corresponding to singular values of lower magnitude. Conclusions: SSA is effective in extracting higher amplitude respiratory excursions while excluding lower amplitude cardiogenic oscillations and noise from the airflow signal. This study demonstrates that suppression of the cardiogenic artefact with SSA is computationally feasible to augment ventilator performance.