Hello everyone!
I have an ECG signal sampled at 500 Hz and, from that, I would like to compute the RR interval and then Welch's PSD. All this aims to understand the sympathetic activation of the person.
I wanted to enhance the R peaks using the 'sym4' wavelet. However, I have some difficulties in understanding how to choose the proper level of decomposition.
Can someone help me with this topic? Is there a "standard" way to assess the proper level of decomposition?
Thank you in advance for your time!
Luca
The optimal level of wavelet decomposition is chosen from the power of 2 which equates your data.
Hi Sarita Azad thanks for replying!
Can you please explain your answer a little more in detail?
The level of decomposition depends on the frequency spectrum of the interested signal component. Here your signal is sampled at 500 Hz which means that the signal is band-limited to 250 Hz. Each level of decomposition separates out the upper half of frequency spectrum to the 'details' and the rest will be in the 'approximation'. ie. after each level, the approximation part contains spectral bands like 0-125 Hz, 62.5 Hz, 31.25 Hz, etc... Now depending upon the spectral characteristics of the signal component of your choice you can decide at which level of decomposition you will separate out the spectral components, after which further levels of decomposition is futile. Hope this clarifies your doubt .
The optimal level of wavelet decomposition basically depends on on the sampling frequency of the signal. It can be heuristically given by:
No. of Decomposition Level = fix [log2(Fs ) − 3]
where, Fs is the sampling frequency of the signal and fix gives a round-off value obtained from the inside bracket expression to its nearest integer.
There is an equation that can be used to determine the maximum level of decomposition.
Max Level = fix(log2(length_of_the_signal/(length_of_the_filter-1)))
Or you can simply use the inbuilt matlab function "wmaxlev(sizeX,wname)" to determine the level.
If you want to understand more about the working of the function you can always check the source code by typing the command "open('wmaxlev')"
Hope that works for you!
- Badges
- Science topic
- Similar topics
- Philosophy
- Philosophy Of Science
- Reasoning