I would like to find out if there is a manual or if you could point me to literature on the effect of different input parameters in mne.chpi.compute_chpi_head_pos ( info, chpi_locs, dist_limit=0.005, gof_limit=0.98, adjust_dig=False, verbose=None)
This is because I am using a few datasets with movement compensation performed with Elekta maxfilter (-movecomp with default parameters such as HPI-fit parameters for each every 200ms) and a few datasets where movement compensation is performed within MNE. What I am finding is some head position calculations with default goodness of fit (0.98) gives me an empty matrix and a "cannot determine the transformation" error for all coils. While, if I reduce the goodness of fit, or if I set the t_step_max=0.2 (which I am assuming is comparable to Maxfilter -movecomp) for calculation of chpi locations ( mne.chpi.compute_chpi_locs), I do get an head_pos array of a decent size.
I would like to find out if there is a manual or if you could point me to
literature on the effect of different input parameters in
mne.chpi.compute_chpi_head_pos ( info, chpi_locs, dist_limit=0.005,
gof_limit=0.98, adjust_dig=False, verbose=None)
These parameters should behave similarly to how they do in MaxFilter's head
position estimation, read more in the MaxFilter manual:
What I am finding is some head position calculations with default goodness
of fit (0.98) gives me an empty matrix and a ?cannot determine the
transformation? error for all coils. While, if I reduce the goodness of
fit, or if I set the t_step_max=0.2 (which I am assuming is comparable to
Maxfilter -movecomp) for calculation of chpi locations (
mne.chpi.compute_chpi_locs), I do get an head_pos array of a decent size.
You might need to set `t_window=0.2` as well. If MaxFilter always outputs
positions separated by 200 ms, then you might also need `t_step_min=0.2` as
well. But there should be some combination of parameters you can use to get
the head positions.
You can also use MaxFilter just for the head position estimation step by
calling it with the `-hp some_filename.pos` option, and then use MNE-Python
to do the movement compensation using these positions computed by MaxFilter.