def LFcomputing(condFile, geomFile, dipoleFile, electrodesFile, savedir):
"""
condFile = 'om_demo.cond'
geomFile = 'om_demo.geom'
dipoleFile = 'cortex.dip'
squidsFile = 'meg_squids.txt'
electrodesFile = 'eeg_electrodes.txt'
"""
# Load data
geom = om.Geometry()
geom.read(geomFile, condFile)
dipoles = om.Matrix()
dipoles.load(dipoleFile)
#squids = om.Sensors()
#squids.load(squidsFile)
electrodes = om.Sensors()
electrodes.load(electrodesFile)
# Compute forward problem
gaussOrder = 3
use_adaptive_integration = True
hm = om.HeadMat(geom, gaussOrder)
hminv = hm.inverse()
dsm = om.DipSourceMat(geom, dipoles, gaussOrder, use_adaptive_integration)
#ds2mm = om.DipSource2MEGMat (dipoles, squids)
#h2mm = om.Head2MEGMat (geom, squids)
h2em = om.Head2EEGMat(geom, electrodes)
#gain_meg = om.GainMEG (hminv, dsm, h2mm, ds2mm)
gain_eeg = om.GainEEG(hminv, dsm, h2em)
gain_eeg.save(savedir)
return gain_eeg
def inverse_head(self, gauss_order=3, inv_head_mat_file=None):
"""
Call OpenMEEG's HeadMat method to calculate a head matrix. The inverse
method of the head matrix is subsequently called to invert the matrix.
Optionaly saving the inverted matrix for later use.
Runtime ~8 hours, mostly in martix inverse as I just use a stock ATLAS
install which doesn't appear to be multithreaded (custom building ATLAS
should sort this)... Under Windows it should use MKL, not sure for Mac
For reg13+potato surfaces, saved file size: hminv ~ 5GB, ssm ~ 3GB.
"""
LOG.info("Computing HeadMat...")
head_matrix = om.HeadMat(self.om_head, gauss_order)
LOG.info("head_matrix: %d x %d" %
(head_matrix.nlin(), head_matrix.ncol()))
LOG.info("Inverting HeadMat...")
hminv = head_matrix.inverse()
LOG.info("inverse head_matrix: %d x %d" % (hminv.nlin(), hminv.ncol()))
if inv_head_mat_file is not None:
LOG.info("Saving inverse_head matrix as %s..." % inv_head_mat_file)
hminv.save(
os.path.join(OM_STORAGE_DIR,
inv_head_mat_file + OM_SAVE_SUFFIX)) #~5GB
return hminv
def make_forward_solution(info,
trans_fname,
src,
bem_model,
meg=True,
eeg=True,
mindist=0.0,
ignore_ref=False,
n_jobs=1,
verbose=None):
assert not meg # XXX for now
coord_frame = 'head'
trans = mne.read_trans(trans_fname)
head_trans, meg_trans, mri_trans = _prepare_trans(info, trans, coord_frame)
dipoles = _get_dipoles(src, mri_trans, head_trans)
eeg_electrodes, ch_names = _get_sensors(info, head_trans)
geom = _get_geom_files(bem_model, mri_trans, head_trans)
assert geom.is_nested()
assert geom.selfCheck()
# OpenMEEG
gauss_order = 3
use_adaptive_integration = True
# dipole_in_cortex = True
hm = om.HeadMat(geom, gauss_order)
hm.invert()
hminv = hm
dsm = om.DipSourceMat(geom, dipoles, gauss_order, use_adaptive_integration,
"Brain")
# For EEG
eeg_picks = mne.pick_types(info, meg=False, eeg=True)
# meg_picks = mne.pick_types(info, meg=True, eeg=False)
# seeg_picks = mne.pick_types(info, meg=False, seeg=True)
if eeg and len(eeg_picks) > 0:
h2em = om.Head2EEGMat(geom, eeg_electrodes)
eeg_leadfield = om.GainEEG(hminv, dsm, h2em)
eegfwd = _make_forward(eeg_leadfield, ch_names, info, src, trans_fname)
# if meg and len(meg_picks) > 0:
# h2em = om.Head2EEGMat(geom, eeg_electrodes)
# eeg_leadfield = om.GainEEG(hminv, dsm, h2em)
# megfwd = _make_forward(eeg_leadfield, ch_names, info, src, trans_fname)
# # merge forwards
# fwd = _merge_meg_eeg_fwds(_to_forward_dict(megfwd, megnames),
# _to_forward_dict(eegfwd, eegnames),
# verbose=False)
return eegfwd
dipoles.load(dipole_file) sensors = om.Sensors() sensors.load(squidsFile) patches = om.Sensors() patches.load(patches_file) ############################################################################### # Compute forward problem (Build Gain Matrices) gauss_order = 3 use_adaptive_integration = True dipole_in_cortex = True hm = om.HeadMat(geom, gauss_order) #hm.invert() # invert hm inplace (no copy) #hminv = hm hminv = hm.inverse() # invert hm with a copy ssm = om.SurfSourceMat(geom, mesh) ss2mm = om.SurfSource2MEGMat(mesh, sensors) dsm = om.DipSourceMat(geom, dipoles, gauss_order, use_adaptive_integration, "") ds2mm = om.DipSource2MEGMat(dipoles, sensors) h2mm = om.Head2MEGMat(geom, sensors) h2em = om.Head2EEGMat(geom, patches) gain_meg_surf = om.GainMEG(hminv, ssm, h2mm, ss2mm) gain_eeg_surf = om.GainEEG(hminv, ssm, h2em) gain_meg_dip = om.GainMEG(hminv, dsm, h2mm, ds2mm) gain_adjoint_meg_dip = om.GainMEGadjoint(geom, dipoles, hm, h2mm, ds2mm) gain_eeg_dip = om.GainEEG(hminv, dsm, h2em) gain_adjoint_eeg_dip = om.GainEEGadjoint(geom, dipoles, hm, h2em)
dipoles.load(dipole_file) sensors = om.Sensors() sensors.load(squidsFile) patches = om.Sensors() patches.load(patches_file) ############################################################################### # Compute forward problem (Build Gain Matrices) gauss_order = 3 use_adaptive_integration = True dipole_in_cortex = True hm = om.HeadMat(geom, gauss_order) #hm.invert() # invert hm inplace (no copy) #hminv = hm hminv = hm.inverse() # invert hm with a copy ssm = om.SurfSourceMat(geom, mesh) ss2mm = om.SurfSource2MEGMat(mesh, sensors) dsm = om.DipSourceMat(geom, dipoles, gauss_order, use_adaptive_integration, "") ds2mm = om.DipSource2MEGMat(dipoles, sensors) h2mm = om.Head2MEGMat(geom, sensors) h2em = om.Head2EEGMat(geom, patches) gain_meg_surf = om.GainMEG(hminv, ssm, h2mm, ss2mm) gain_eeg_surf = om.GainEEG(hminv, ssm, h2em) gain_meg_dip = om.GainMEG(hminv, dsm, h2mm, ds2mm) gain_adjoint_meg_dip = om.GainMEGadjoint(geom, dipoles, hm, h2mm, ds2mm) gain_eeg_dip = om.GainEEG(hminv, dsm, h2em) gain_adjoint_eeg_dip = om.GainEEGadjoint(geom, dipoles, hm, h2em)
import os
os.mkdir('tmp')
if not op.exists("leadfields"):
import os
os.mkdir('leadfields')
# Compute Leadfields
gauss_order = 3
use_adaptive_integration = True
dipole_in_cortex = True
if op.exists("tmp/hmi.mat"):
hminv = om.SymMatrix("tmp/hmi.mat")
print("HM inverse loaded from ", "tmp/hmi.mat")
else:
hm = om.HeadMat(geom, gauss_order)
hm.invert()
hm.save("tmp/hmi.mat")
hminv = hm
# hminv = hm.inverse() # to also test the adjoint method: comment the 3
# previous lines, and uncomment this line, and the two others containing
# 'adjoint'
if op.exists("tmp/dsm.mat"):
dsm = om.Matrix("tmp/dsm.mat")
print("DSM loaded from ", "tmp/dsm.mat")
else:
dsm = om.DipSourceMat(geom, dipoles, gauss_order, use_adaptive_integration,
"Brain")
dsm.save("tmp/dsm.mat")
