Hello,

For ECG noise filtering by wavelet decomposition, it is important to choose not only the wavelet type but also the level of decomposition. A comparison is given by the following reference:

Gavrovska et al “Selection of wavelets décomposition levels in ECG filtering” , IEEE, 2009.

A yet more recent reference is:

He et al “Optimal base wavelet selection for ECG noise reduction using a comprehensive entropy criterion”, Entropy, MDPI, 2015.

Thanks a lot

MATLAB has a function which decomposes time-domain signals fully automatically, including into wavelets. It is in with their "toolkit" for A.I. on

signals. Beyond converting the data into wavelets, it also identifies which

components of the signal hold all of the information.

They have some demo lectures on it

Nominally, it is a commercially sold product, not free. But virtually any of their A.I. products can be had as demos for free.

If nothing else, try to locate the training videos, to get a sense

of what good advanced tools can do. It is important to decide

whether you want to do a lot of your own mathematical

programming, whether you want to leverage your work

with existing software libraries, or whether you want to

try to use a fully-integrated product. All are fully viable

approaches.

It is also worth clarifying in your own mind - do you want to

use wavelets simply as a denoising step before other

processing, or do you wish to do broader processing

and signal analysis based on wavelets. Once again,

both answers are totally reasonable.

Good luck!