njit-etd2003-081 - New Jersey Institute of Technology

5
breathing room air (RA) or with oxygen supplement (02) during rest and exercise, that
alter baroreflex activity, in order to assess COPD related changes in baroreflex response.
When HRV is analyzed in the frequency domain, the power spectrum of HRV
does not show its temporal changes. There are many physiological situations of interest
where heart rate changes rapidly over time and the monitoring of these temporal changes
may be very important. Time frequency representations are perfect candidates to
monitor these temporal-spectral changes. Time frequency analysis is performed on the
HRV data to show vagal tone and the sympatho-vagal balance as a function of time. In
this part of the study, several time frequency representations such as the short time
Fourier transform, the smoothed pseudo Wigner-Ville, the Choi-William, the Born-
Jordan-Cohen are used in comparison with wavelet distributions (Morlet, Meyer,
Daubechies, Mexican Hat and Haar) for analyzing HRV and physiological signals.
Coherence spectral plots are graphical representations of the coherence between
two signals. Coherence can be viewed as correlation coefficients in the frequency
domain [5]. It is a measure of the linear dependence between two signals, normalized to
values between zero and one. The applications of coherence in past studies have been
directed toward the analysis of the electroencephalogram or EEG signals. In this study,
however, coherence is used not for EEG signals but for the combinations between heart
rate, blood pressure and respiration signals. The coherence spectral plots in the whole
HRV frequency range (0.04 to 0.7 Hz) are presented and the average weighted sum, the
weighted coherence values, in the low frequency band (LF: 0.04 to 0.15 Hz) and in the
high frequency band (HF: 0.15 to 0.4 Hz) are obtained for further statistical analyses
(i.e. principal component analysis, PCA and cluster analysis, CA). These values are