Fine calibration and SSS shielding actor

The fine calibration file is provided in the sss_cal.dat file and is used by the MaxWell filter software. The fine calibration improves by a factor of 10 or 100 the “SSS shielding factor”, i.e. how well SSS removes artifacts from the signal.

Note

In MNE-Python’s implementation of mne.preprocessing.maxwell_filter(), the fine calibration file is provided in the argument calibration.

The “default” fine calibration file in the sample dataset:

from mne_wiki.datasets import sample

fine_cal_file = sample.data_path() / "calibration" / "sss_cal.dat"

This default fine calibration file is computed during the annual maintenance of the MEG and is stable over time. However, it can also be re-computed from an empty-room recoridng.

Computing the fine calibration

MNE-Python provides the function mne.preprocessing.compute_fine_calibration() to compute the fine calibration from empty-room data.

Note

MNE-Python’s docstring mentions that all channels should be good, probably because of the channel selection occurring under-the-hood and to the creation of an output structure which has to include all channels. However, it seems that bad or noisy channels have little effect on the fine calibration. Thus, MNE-Python’s docstring can be understood as “do not mark bad channels as bad, just let the function handle them”.

The service tools also include scripts to compute the fine calibration and the SSS shielding factor.

$ /neuro/dacq/tools/bin/meg_calib -v -f empty_room_68.fif

SSS shielding factor

The SSS shielding factor corresponds to ratio of the norm of the channel type vector (shape (102,) for mags, shape (204,) for grads) before and after SSS, as a function of time (the norm is estimated at every timepoint). xfilter can estimate the SSS shielding factor with (1) default fine calibration and (2) a custom fine calibration file.

$ /neuro/dacq/tools/bin/xfilter -v -f empty_room_68.fif -sf
$ /neuro/dacq/tools/bin/xfilter -v -f empty_room_68.fif -sf -cal sss_cal.dat

You can then compare the *.xfilter.txt reports, e.g. with:

$ more *.xfilter.txt

The argument xwav can be added to generate a FIFF file containing the uncorrelated noise waveforms (uncorrelated between channel waveforms), i.e. channel noise waveforms.