ECG power spectrum

[johnsold]

Luanna,

The size of the resampled data array could be larger or smaller than the original, depending on the sample rate applied. In my example I used a much faster sample rate than the RR intervals, but using various types of interpolation will allow other sample rates. Choose the rate based on desired frequency resolution (df = fs/N), array sizes, memory, processing time, etc.

Lynn

[hgurji]

Wow, thanks Lynn.  The resampling matrix you detailed really helps a lot.  Now I understand what the problem is.  Will a small dt affect the frequency spectrum of the RR intervals power spectra or should I stick with the inverse of my sampling frequency?  Again, much thanks for explaining the resampling section.

-hgurji 

[johnsold]

hgurji,

You need to use the dt of the data fed to the spectral analysis function (the resampled dt).   The FFT resolution is dependent on the timing of the samples it analyzes, not the rate at which you collected the data originally. Resampling simulates sampling at the new (usually faster) rate.  It does not create any new information, but does allow you to use tools designed for uniformly sampled data (FFT) with non-uniformly sampled data (RR Intervals).

Lynn

[hgurji]

Got it.  Thanks Lynn.  I appreciate your help.  

-hgurji 

[hgurji]

Hi Lynn,

After reading my previous post, it seems like I did not ask the right question.  Will a small dt that is inputted into the Align and resampled express VI affect the FFT power spectrum of the resampled RR intervals or does it not matter?   Again, thanks for your help.  It good to have people like yourself explain stuff like this to people like me.  Take care.

-hgurji 

[johnsold]

hgurji,

I am not sure I understand your question.

I have not worked with the express VI you mention, so I am not sure what it produces.

Generally, the resampling process will not change the spectral characteristics of the signal, because you do not change the signal.  That is, in the sense that a filter removes some spectral components, a filter changes the signal.  Resampling changes the representation of the signal but not the signal itself.

So if your average heart rate was 60 beats per minute and it varies with breathing at 5 breaths per minute, you would expect spectral components at 1 Hz and at 0.083 Hz.  The resolution (df) of the spectral output depends on the sampling rate (dt = 1/fs) of the resampled signal fed to the FFT.

If this did not answer your question, try to rephrase it or give an example.

Lynn 

[hgurji]

okay, I got it.  Thanks Lynn.  I appreciate your help.

-hgurji 

[Steve.J]

There is a very nice article on Heart Rave Variability analysis using LabVIEW on ni.com here:  http://zone.ni.com/devzone/cda/epd/p/id/5832

Also - Please note that we have re-opened the Life Sciences discussion forum for topics just like this here:  http://forums.ni.com/ni/board?board.id=LifeScience

Hopefully we can get more collaboration going among biomedical people using LabVIEW on this forum.

Thanks,

Steve