Can you provide the original ECG you used with all the wavelet decomposition that you find for this particular ECG?

Which wavelet do you use for the decomposition?

The decomposition number can also be affected by the sample rate of your signal.

Why did you compute the sum of d3 d4 and d5? this summation may lead to a cancellation of two decomposed signal. For instance id d4=10 and d5=-10 at a given position.

Maxime Yochum beat me to it. I agree with all his comments. Most importantly the one referring to the sampling rate.. d3 in one sampling rate will be d6 for twice that sampling rate.

Ok Here is the ECG signal I used for the analysis.

Attached file 3 contains Matlab code I used to perform discrete wavelet transform of this ECG signal. 

I was committing one mistake earlier. I should not have summed up the detail coefficients.

Now I  have decomposed  ECG waveform up to level 5 using db4 wavelet and have reconstructed frequency localized version of the signal using wavelet coefficients at scales 3,4 and 5. Please execute this code and locate the QRS region in it. Also tell me whether I am doing right ?

Hoping for early response.

yes, provide the original ECG you used with all the wavelet decomposition that you find for this particular ECG?

Here is the original ECG

I think that any of the deflections from the baseline, no matter how big or small, are QRS complexes.

Richard