Downstream Processing#

Having extracted the required anatomical components, and computed the sensor covariance and M/EEG to anatomy co-registration, the inverse model may be calculated as follows. Of note, as the inverse transformation and the following intermediate byproducts depend on the sensor-position, these have to be recomputed in case the sensor covariance or M/EEG co-registration was recomputed (cf. section Sensor Position Dependent Factors).

The following code segment computes the inverse model, employing the previously calculated intermediate results.

import finnpy.src_rec.bem_mdl
import finnpy.src_rec.cort_mdl
import finnpy.src_rec.fwd_mdl
import finnpy.src_rec.inv_mdl
import finnpy.src_rec.subj_to_fsavg

bem_mdl = finnpy.src_rec.bem_mdl.run(fs_path, anatomy_path, subj_name, signal_type, coreg, conductivity = None, tgt_icosahedron_level = 4):
cort_mdl = finnpy.src_rec.cort_mdl.get(anatomy_path, subj_name, signal_type, coreg, bem_mdl)

fwd_mdl = finnpy.src_rec.fwd_mdl.compute(bem_mdl, cort_mdl, coreg, signal_type, rec_info, _mag_factor=1e-7, conductivity = None):
fwd_mdl = finnpy.src_rec.fwd_mdl.restrict(cort_mdl, fwd_sol, coreg)
inv_mdl = finnpy.src_rec.inv_mdl.compute(sen_cov, fwd_sol, signal_type, rec_info, method = "dSPM")

subj_to_fsavg_mdl = compute(cort_mdl, anatomy_path, subj_name, fs_path, overwrite)
fs_path points towards the FreeSurfer installation. The FreeSurfer folder should contain the ‘bin’ folder, license.txt, and sources.sh (among other files/directories).
anatomy_path contains sub-folders with anatomy folders for all subjects. FreeSurfer will fail if a subjects folder already exists.
subj_name is the name of the subject.
signal_type is be either ‘EEG’ or ‘MEG’.
coreg is the output of the co-registration run from Sensor Position Dependent Factors.
conductivity describes the conductivity of the individual layers. Defaults to 0.3 for MEG and (0.3, 0.006, 0.3) for EEG.
tgt_icosahedron_level determines the level of downscaling. Defaults to 4.
rec_info if the signal_type is ‘MEG’ the path to the *.fif file is required. In case the signal_type is ‘EEG’ a tuple of EEG setup (‘1005’ or ‘1020’) and channel names is required.
_mag_factor is a statical scaling factor for MEG analysis.
method is the method used to compute the sensor noise normalization. Currently only supports ‘dSPM’.

Let’s take a closer look at this code example.

bem_mdl = finnpy.src_rec.bem_mdl.run(fs_path, anatomy_path, subj_name, signal_type, coreg, conductivity = None, tgt_icosahedron_level = 4):

Calculates position and current sharing of the individual virtual cortical dipoles.

cort_mdl = finnpy.src_rec.cort_mdl.get(anatomy_path, subj_name, signal_type, coreg, bem_mdl)

Loads cortical data and removes any data points that are outside the inner skull BEM model component.

fwd_mdl = finnpy.src_rec.fwd_mdl.compute(bem_mdl, cort_mdl, coreg, signal_type, rec_info, _mag_factor=1e-7, conductivity = None):
fwd_mdl = finnpy.src_rec.fwd_mdl.restrict(cort_mdl, fwd_sol, coreg)
inv_mdl = finnpy.src_rec.inv_mdl.compute(sen_cov, fwd_sol, signal_type, rec_info, method = "dSPM")

Computation of the forward model proper, restriction to surface orthogonal dipoles only, and subsequent inversion.

Next steps#

This concludes the computation of the inverse matrix. It’s application is demonstrated in section Model Application.

On this page

This Page