回复： Frequency domain analysis of HRV.

[Shabbychev]

I need to do the Frequency domain analysis on HRV. I have the input signal generated from the Simulate ECG source.vi from the Biomedical Toolkit. The VI samples at 500Hz evenly, so is interpolation needed? Then when i put the signal through FFT PSD, the y value output is weird on the PSD graph. Also i know that i need to split it into VLF LF and HF bands, which i need help also. Example block diagram most appreciated. Need to clarify urgently!!

Front panel
Block diagram

[johnsold]

Shabbychev,

Looking at the images you posted raises a few questions:

First, you are taking the spectrum of the ECG signal not the HR signal. The multiple peaks in the spectrum are probably the spectral content of the P waves, QRS complexes, and T waves.

For spectral analysis of the ECG (which is NOT the same as spectral analysis of HR) no interpolation is needed because the ECG is uniformly sampled as you mentioned. When you have an array of RR intervals (which is almost what you get at the output of the for loop following the peak detector), you have a non-uniformly sampled data set. The standard spectral analysis routines cannot handle that data correctly. That data can be resampled or interpolated to produce a uniformly sampled data set, but you must use some care to make sure that you do not introduce spurious signals.

The shift registers in the for loops should probably be initialized. Otherwise they will retain the last value from the previous iteration of outer while loop.

The for loop containing the 1/x function is unnecessary. Most numeric functions are polymorphic, meaning they will accept real, double, arrays, and some other data types.

Lynn

[ZJ.Gu]

Hi,

Do you know the page in Using LabVIEW for Heart Rate Variability Analysis? It has a free startup kit application and you can have a look. The frequency domain analysis for human has some standards in frequency bands selection and measure calculations. In that page, you can find most of them and apply the calculation in LabVIEW using the startup kit.

If you have more questions, please contact:

Zhijun Gu 

帖子被Zhijun Gu在07-31-2008 09:14 PM时编辑过了

[Shabbychev]

Yes i have seen that site, but the problem is that i am unable to split them into bands of VLF, LF, and HF. I know that interpolation works but i do not know how to use it.

[ZJ.Gu]

Hi,

From the first message of this topic, you just use the ECG signal to estimate the PSD. However, for HRV analysis, you should use the RR interval signals to estimate PSD. From your screenshots, you have extracted R peaks and RR intervals. What you need is to resample the RR intervals to evenly-sampled series. In LabVIEW, you can use the functions in palette " Interpolation and Extrapolation", If you have Advanced Signal Processing Toolkit, it has a time series resampling function which could accept unevenly-sampled signals.

Zhijun Gu

[jmc1]

Hello,

Respect to HRV Analysis Startup Kit, when I use the Frequency Domain Analysis with FFT, what are the two parameters needed to resample a RR interval signal? I suppose that one of them is the sampling rate used to obtain an equally time spaced RR interval signal, but what is the meaning of the other?

Furthermore, can I use a RR temporal serie (sampled RR interval serie) as input of the application?

Thanks in advance!

[ZJ.Gu]

Hi,

There is a parameter called "interpolated rate" which is the sampling rate you mentioned.

Unfortunately, the startup kit could only support the original RR interval signals not interpolated RR interval signals.

[jmc1]

Hi, thanks for your quick response!

Other questions related to Frequency Domain Analysis (FFT).

FFT setting:

- What is the parameter "frequency bins" for? Should be related to the sampling rate used for ECG digitalization?

- The values of LF Power and HF Power are expressed in ms2?

Window setting: 

- Could you briefly explain the meaning of the three parameters of this tab? If I choose in "window type": None, the length and overlap values are needed?

Finally,

- did you validate your results?

- can I use the kit to analyze mean aortic pressure signals (mean value during a heartbeat of aortic pressure)?

Thanks!

[Steve.J]

Shabbychev and JMC1 - please see and consider joining our NI Biomedical User Group on ni.com - there are more articles and a discussion forum there that has many interested users, all using LabVIEW in biomedical applications.

thanks,

Steve

[Coq_Rouge]

---

S.E.Johnson wrote:

there are more articles and a discussion forum there that has many interested users, all using LabVIEW in biomedical applications.

---

But I think the owners has to pay more attention to it. It can not reach popularity on its own..