How to remove EMG artifacts from ECG?

Hello,
I built a circuit that aquires ECG from electrodes on chest. When the subject is stable, I can get ECG with enough accuracy. But when the subject moves or takes a deep breath, I am getting a ECG mixed with EMG.

I though my filters are okay, but apparently they're not.

In short, I m stuck.

Do you have any suggestions?
I'll be appreciated.

Thank you.

Regards,

You forgot to post your schematic so we are just guessing what you did.
Manufacturers of instrumentation amplifier ICs recommend negative feedback of common mode DC voltages to the patient's right leg which cancels AC hum pickup and DC from muscles that are in both sensor pads.
Maybe you built your own instrumentation amplifier and its resistors do not match causing it to have poor common-mode rejection.

Audioguru said:

You forgot to post your schematic so we are just guessing what you did.
Manufacturers of instrumentation amplifier ICs recommend negative feedback of common mode DC voltages to the patient's right leg which cancels AC hum pickup and DC from muscles that are in both sensor pads.
Maybe you built your own instrumentation amplifier and its resistors do not match causing it to have poor common-mode rejection.

Click to expand...

Actually that is my referance circuit. I just changed AD705J with its equivalent(OP07) and added a notch filter to output.

And my 0.03Hz HPF has 1000uF cap and 5k6 res.

I can't see any differance.

Hi Aeytr,
I have had a few ECG tests, one was a portable unit that recorded my ECG all night long. They all used many sensors, not just three so maybe the additional sensors help to cancel muscle activity.

Hi,
Thank you for your useful response. I am a little unexperienced. Actually I dont even know how to add more electrodes. I guess, I should add more AD620XX and then use MUX but that is jus a guess. Do you have any books or something like that based on adding more electrodes?

Thank you again.
Regards.

I have never made an ECG circuit and I have no books about it. I have never seen a circuit with all the sensors like I had connected to me. My doctor tells me to hold still during the ECG test.

Audioguru said:

I have never made an ECG circuit and I have no books about it. I have never seen a circuit with all the sensors like I had connected to me. My doctor tells me to hold still during the ECG test.

Click to expand...

Understood.
Thanks for the replies.

Audioguru said:

Hi Aeytr,
I have had a few ECG tests, one was a portable unit that recorded my ECG all night long. They all used many sensors, not just three so maybe the additional sensors help to cancel muscle activity.

Click to expand...

The additional sensors are for the V1-V6 derivations, for inspecting left/right ventricular abnormalities (ST elevation etc.) at higher resolution than by extremity electrodes. They have nothing to do with canceling muscle-related noise. I am no electronics engineer but I am a licensed medical doctor (I haven't read an ECG since medical school 35 years ago though!).

Muscle artifacts lie in the range of 35-45 Hz.
In general, a 2nd order notch filter suffice.

aagaag said:

The additional sensors are for the V1-V6 derivations, for inspecting left/right ventricular abnormalities (ST elevation etc.) at higher resolution than by extremity electrodes. They have nothing to do with canceling muscle-related noise. I am no electronics engineer but I am a licensed medical doctor (I haven't read an ECG since medical school 35 years ago though!).

Click to expand...

Thanks for information. I'll jus adjust my filters better.

Thanks again.

andre_teprom said:

Muscle artifacts lie in the range of 35-45 Hz.
In general, a 2nd order notch filter suffice.

Click to expand...

I have 2nd order notch filter for both mains electricity and EMG. I'll check my calculations again.

Thank you.

I made lots of ECG/EMG/EEG amplifiers in the mid '70's as I elected to take Biomedical and for a festival week, we demonstrated on girls how to measure heartbeats. But even cooler was that electrodes on the head placed on each temple could detect lateral eye position.

Your problem isn't EMG but rather galvanic skin response (GSR). The electrode conductor, gel electrolyte and skin salts all modulate the charge on the interface so any motion from a galvanic action. ALso consider Q=CV. If Q is constant and you squeeze the gap between skin electrode C increases which affects V instantly. But water has a huge dielectric constant ( ~60) compared to oil which can be as low as 2. I dont know what gels are, so removing all skin salt and moisture with rubbing alcohal, proper silver coated electrodes and low dielectric constant gel with a constant electrode pressure using some attachment method will reduce your motion artifact noise.

Better quality pads improve signal and reduce noise as well as constant pressure attachment with tape or whatever or pads that are self-adhesive with silver micro-particles.

Higher CMMR and shielding with 5 lead measurement methods also improve SNR by getting a better common mode signal to cancel with the differential signal.

For EMG signals, we used only 2 electrodes to capture muscle amplitude and frequency signals to control prosthetics quite easily in university, although we did not get into force feedback at the time. Now it is standard practice.

EEG signals are the smallest ( uV) and often require bypassing the insulation layer of the hair and skin and get some blood to the surface with massage but can be done without cutting hair if you use a baseball cap to prevent scalp motion and piezo effects of moving electrodes of charged dielectric.

SunnySkyguy said:

we demonstrated on girls how to measure heartbeats.

Click to expand...

Then you are also a dirty old man like me.