Life Science

Showing results for 
Search instead for 
Did you mean: 

ecg baseline wander


I am acquiring ECG data and have noticed significant baseline shift associated with each respiratory cycle.  After the signal has been acquired, my VI uses peak detection to find the R wave of each waveform, but because of the baseline wander, I get many false positive and false negative peaks.  Additionally, because I am trying to measure precise P wave time measurements and heart rate variability, applying FIR or IIR filtering techniques seem not to be the method of choice (secondarily to their manipulation of the basic ECG waveform itself).  I have read about Labview 8.6 signal processing toolkit, and have seen the detrend vi's.  Unfortunately, I only have access to labview 7.1.  With all this said, can anyone help me with this particular problem?  I know I am asking for a lot (baseline removal without distortion of ECG signal characteristics), but any help is greatly appreciated.  Thanks.



0 Kudos
Message 1 of 2

How are you acquiring the ECG signal (with what type of amplfier and electrodes, etc.)?  Is this animal or human research?


You almost certainly have some kind of AC coupling already (high-pass filter) - it just sounds like the time-constant is too long (filter setting is too low), leading to the wandering baseline (low frequency drift) due to movement artifact.  Most ECG is acquired with a 0.5Hz to 40Hz bandpass filter (though 100Hz is preferred to minimize attenuation of high-frequency features when possible).


I'm not clear on why you don't want to simply apply a FIR or IIR high-pass filter to remove the baseline drift.  It will slightly change the shape of the waveform, but it will not significantly affect the higher-frequency components that you are interested in measuring.  The intervals that you are measuring should not be affected (each beat complex will be affected in the same way, so the interbeat intervals will be unaffected).


If you absolutely cannot tolerate any further filtering, then you either have to use a more sophisticated detrending (wavelet based) and/or peak detection (there are many algorithms out there - just do a search) or minimize the baseline drift by recording from different sites or using different electrode technology.


For ECG signal processing ideas, refer to this great article on



National Instruments



0 Kudos
Message 2 of 2