njit-etd2003-081 - New Jersey Institute of Technology

njit-etd2003-081 - New Jersey Institute of Technology njit-etd2003-081 - New Jersey Institute of Technology

archives.njit.edu
from archives.njit.edu More from this publisher
20.01.2015 Views

81 length of the IBI signal depends on the number of heartbeats. Now, in order to take an 8192-point FFT of a signal that has only 6,000 samples, a technique called zero padding must be used. Essentially all samples from 6001 to 8192 are given the value of zero. The only effect this has on the spectrum is that it increases the frequency resolution. When the FFT is calculated, the spectrum is limited from 0 Hz up to 10 Hz. The frequency resolution is now 20/8192=0.00244 Hz. Once the power spectrum is obtained, it is smoothed twice by applying a modified Daniel Rectangular smoothing algorithm. Figure 3.8 illustrates the power spectrum of the IIBI signal in Figure 3.7. Figure 3.8 Power spectrum of the heart rate IIBI signal in Figure 3.7. It should also be noted that when performing the FFT, the IIBI signal is windowed by applying a split cosine bell taper. The taper begins at 20 percent from each end of the time signal. The application of the split cosine bell taper is done before the zero padding takes place so as to provide a smooth transition to zero rather than the window's abrupt changes from one to zero.

82 It should be noted that if the window were a simple rectangular function, the sharp one to zero changes are the cause of the side lobes in the FFT output sinc function (sin(x)/x). To minimize the spectral leakage caused by those sidelobes, one has to reduce the sidelobe amplitudes by using window functions other than the rectangular window. The purpose of the split cosine bell taper window is to reduce any additional spectral components that result from the shape of the window. Multiplying the time signal by the window function does the windowing. The split cosine bell taper for a 6,000-point long signal is shown in Figure 3.9. Figure 3.9 Plot of the split cosine bell taper used as a window for FFT calculation. Past research in power spectral analysis of heart rate variability correlates three distinct frequency regions peaks with certain physiological parameters [14] as illustrated by a more classical power spectrum of the IIBI signal in Figure 3.10. The very low frequency band (0.003-0.04 Hz) is associated with vasomotor control and temperature control. The low frequency band (0.04-0.15 Hz) is associated with baroreceptormediated blood pressure control. The high frequency band (0.15-0.5 Hz) has been linked with respiration.

82<br />

It should be noted that if the window were a simple rectangular function, the<br />

sharp one to zero changes are the cause <strong>of</strong> the side lobes in the FFT output sinc function<br />

(sin(x)/x). To minimize the spectral leakage caused by those sidelobes, one has to<br />

reduce the sidelobe amplitudes by using window functions other than the rectangular<br />

window. The purpose <strong>of</strong> the split cosine bell taper window is to reduce any additional<br />

spectral components that result from the shape <strong>of</strong> the window. Multiplying the time<br />

signal by the window function does the windowing. The split cosine bell taper for a<br />

6,000-point long signal is shown in Figure 3.9.<br />

Figure 3.9 Plot <strong>of</strong> the split cosine bell taper used as a window for FFT calculation.<br />

Past research in power spectral analysis <strong>of</strong> heart rate variability correlates three<br />

distinct frequency regions peaks with certain physiological parameters [14] as illustrated<br />

by a more classical power spectrum <strong>of</strong> the IIBI signal in Figure 3.10. The very low<br />

frequency band (0.003-0.04 Hz) is associated with vasomotor control and temperature<br />

control. The low frequency band (0.04-0.15 Hz) is associated with baroreceptormediated<br />

blood pressure control. The high frequency band (0.15-0.5 Hz) has been linked<br />

with respiration.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!