Extraction of respiratory activity from PPG and BP signals using Principal Component Analysis

Abstract

In high risk situations such as cardiac arrhythmias, ambulatory monitoring, stress tests, sleep disorder investigations and post-operative hypoxemia situations, monitoring of respiratory activity would be mandatory. Electrocardiogram (ECG), blood pressure (BP) and photoplethysmographic (PPG) signals can be used for extraction of respiratory activity, and will eventually eliminate the use of additional respiratory sensor. Using a simple and standard non-parametric mathematical technique, Principal Component Analysis (PCA), the respiratory related information is extracted from complex data sets such as PPG and BP signals. The respiratory induced variations (RIV) of PPG and BP signals are described by coefficients of computed principal components. Singular value ratio (SVR) trend is used to find the periodicity, which is one of the crucial parameters in forming the data sets for PCA. Test results on MIMIC data base clearly indicated a strong correlation between the extracted and actual respiratory signals. Statistical measures in both time and frequency domains such as Relative Correlation Coefficient (RCC) and Magnitude Squared Coherence (MSC) respectively and Accuracy Rate (AR) are calculated to demonstrate the fact, that respiratory signal is present in the form of first principal components.