I am not sure if you can remove noise from the data. Usually the mweasurement contains of a signal over a given time and then the noise at no signal. These can be compared, usually in the frequency domain and plotted together. There are routines for adjusting for the noise, but it is only possible down to a given s/n difference, I guess about -6dB. The noise, if it has an equal frequency distribution as the signal part, will influence the signal down to - 10 dB using 1 dB digit resolution. If the noise is all above 100 Hz and the signal below it can just be low pass filtered for convenience. The noise in the signal region is difficult to deal with if they cant be separated in the measurements.