Estimation of Respiratory Rate From ECG, Photoplethysmogram, and Piezoelectric Pulse Transducer Signals: A Comparative Study of Time-Frequency Methods

• Published in: IEEE transactions on biomedical engineering Vol. 57; no. 5; pp. 1099 - 1107
• Main Authors: Dash, Shishir, Shelley, Kirk H., Silverman, David G., Chon, Ki H.
• Format: Journal Article
• Language: English
• Published: New York, NY IEEE 01-05-2010; Institute of Electrical and Electronics Engineers; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Algorithms; Amplitude modulation; Biological and medical sciences; Complex demodulation (CDM); Demodulation; Detection algorithms; Diagnosis, Computer-Assisted - methods; ECG; Electrocardiography; Electrocardiography - methods; Electrocardiography. Vectocardiography; Electrodiagnosis. Electric activity recording; Flexible manufacturing systems; Humans; Investigative techniques, diagnostic techniques (general aspects); Medical sciences; Methods; Micro-Electrical-Mechanical Systems; Oximetry - methods; Photoplethysmography - methods; piezoelectric sensors; Piezoelectric transducers; pulse oximeter; Reproducibility of Results; respiratory rate; Respiratory Rate - physiology; Sensitivity and Specificity; Signal analysis; Signal detection; Time frequency analysis; Transducers; Wavelet analysis; wavelets (WTs); Human; piezoelectric sensors; Breathing rate; Piezoelectric sensor; Estimation; ECG; Algorithm; wavelets (WTs); Statistical method; Electrodiagnosis; Temporal frequency; respiratory rate; Electrocardiography; Signal processing; Photoplethysmography; pulse oximeter; Complex demodulation (CDM); Comparative study; Demodulation; Biomedical engineering
• ISSN: 0018-9294; 1558-2531
• DOI: 10.1109/TBME.2009.2038226

We compare the performance of two different time-frequency-based breathing rate (BR) detection algorithms when used on three different physiological signals: the ECG, the photoplethysmogram (PPG), and the piezoelectric pulse transducer (PZO) signal. Studies carried out over the past have shown the existence of amplitude and/or FMs due to respiration in physiological signals, such as those mentioned. In a recent study, we analyzed the PPG signal and detected the FM and amplitude modulation effect that controlled breathing had on it, and inferred the rate of respiration using the time-frequency spectrum (TFS) (via a wavelet (WT) or complex demodulation (CDM) approach). We showed that such TFS BR detection methods were very accurate and consistently outperformed the exclusively time-domain autoregressive modeling (AR) method, especially in the real-time (data length of 1 min) case. We now explore the possibility of using these methods on the ECG and the finger PZO signal, of which only the former has been previously used with some success to derive BR. Testing performed on 15 healthy human subjects for a range of BR and two body positions showed that though the PPG signal gave the most consistently high performance, the ECG and PZO also proved to be reasonably accurate over longer time segments. Furthermore, the CDM approach was on average either better than or comparable to the WT method in terms of both accuracy and repeatability of the detection.