ECG filter probs..

[Joaana]

Hi guys....

I am actually developing a Heart rate variability measuring meter. I need to accquire ECG in real time which i am able to do but with loot of added noise. I have attached the pics of the ECG ( doest not much look like it !!!) dat i am able to accquire. Can sme 1  tell me how can i filter it out.....Notch with 50 Hz is one solution i know can u suggest me any thing better .

The Band pass filter in my ckt has a BW of 0.01 to 106 Hz jus sufficient for the ECG signal of 0.05 -120 Hz. But i suppose my circuit is practically even very high freqs(shud i go for a Low Pass Filter)...can ny 1 plz  help me out. 

[Ali84]

What kind of amplification circuitry are you using? And what are you using as electrodes?

[johnsold]

Joanna,

From the waveforms it looks as though you have some power line interference. This is quite common in ECG applications and is the major reason why differential input or instrumentation amplifiers are usually used. If you cannot eliminate the interference before capturing the data, a notch filter may be a good choice.

Other possibilities: (1) Try to set the threshold for identifying the peaks above the interference. In the images you posted the signal to noise ratio appears high enough that this might work fairly well. You could test for peaks that occur too close together to be valid RR intervals. This won't be perfect but should flag the most serious errors.

(2) Try a high pass filter with the cutoff above the power line frequency. Since you are interested in heart rate variability, you only need the locations of the QRS complexes to calculate RR intervals. If the QRS complexes are comparable in width to the power line frequency waveform, this will not work well.

(3) Use another channel and try to capture the power line interference only. Then scale that and subtract it from the data channels. This creates a pseudo-differential signal and may help. Phase shifts can be a problem for good cancellation.

Lynn