It would help to define what you describe as artifact as we would not want to remove true signal, only noise. The signal must be described to suggest a solution. Simple electronic filters are generally used for this and plenty of information on their basic design and use can be found in the product manuals used with similar devices.

Hello William Letter , thank you for your answer!

The objective of our study was to determine the effect of a massage system in a car seat on the electromyogram (EMG) of back muscles. We can observe rather static activity. The EMG-rashes in our data are small. Therefore, the ECG artifact is very dominant in the raw data.

I found one article that suggests using a fourth-order Butterworth filter with a 30 Hz cutoff. What do you think about this method?

To suggest an appropriate electronic filter, the actual signal (and settings used to collect it) would need to be reviewed (the 'scientific method' is used). Speculating is a waste of time. Have someone with electronic signal analysis experience use a scope to view the data, then suggest a filter for your application. Many commercial system used to review these types of data streams already have optional filters which can be selectively used to remove various types of noise (and if the same types of noise exist in your signal, then you may want to start with the ones they suggest too. Be careful though. The reason these filters are optional is because they often degrade the real signal).

Also, if the massage system used interferes with the signal data obtained, then turn it off, collect the data, then turn it back on as needed (note: it is not clear from your post when you collect data for analysis, so this may or may not apply).

Filtering poorly collected data is probably not the answer. It might help to put the EMG electrodes closely together, in line with the fiber direction of your muscle under study. They could then pick up the same amount of ECG signal. This would be canceled out in your differential amplifier for the EMG.

Besides the use of Low pass Filter, Clipping ,adaptive substracton technique seem to be good option

Thank you Alok Kumar Bharti !

Do you have any experience with the software AcqKnowledge by BIOPAC? Do you know how to apply an adaptive substraction technique on the data using AcqKnowledge?

Personaly I have not used the adaptive substracton technique . It is better than Filter ing or clipping