def test_gamma_map_vol_sphere():
"""Gamma MAP with a sphere forward and volumic source space."""
evoked = read_evokeds(fname_evoked, condition=0, baseline=(None, 0),
proj=False)
evoked.resample(50, npad=100)
evoked.crop(tmin=0.1, tmax=0.16) # crop to window around peak
cov = read_cov(fname_cov)
cov = regularize(cov, evoked.info, rank=None)
info = evoked.info
sphere = mne.make_sphere_model(r0=(0., 0., 0.), head_radius=0.080)
src = mne.setup_volume_source_space(subject=None, pos=30., mri=None,
sphere=(0.0, 0.0, 0.0, 80.0),
bem=None, mindist=5.0,
exclude=2.0)
fwd = mne.make_forward_solution(info, trans=None, src=src, bem=sphere,
eeg=False, meg=True)
alpha = 0.5
pytest.raises(ValueError, gamma_map, evoked, fwd, cov, alpha,
loose=0, return_residual=False)
pytest.raises(ValueError, gamma_map, evoked, fwd, cov, alpha,
loose=0.2, return_residual=False)
stc = gamma_map(evoked, fwd, cov, alpha, tol=1e-4,
xyz_same_gamma=False, update_mode=2,
return_residual=False)
assert_array_almost_equal(stc.times, evoked.times, 5)
# Compare orientation obtained using fit_dipole and gamma_map
# for a simulated evoked containing a single dipole
stc = mne.VolSourceEstimate(50e-9 * np.random.RandomState(42).randn(1, 4),
vertices=stc.vertices[:1],
tmin=stc.tmin,
tstep=stc.tstep)
evoked_dip = mne.simulation.simulate_evoked(fwd, stc, info, cov, nave=1e9,
use_cps=True)
dip_gmap = gamma_map(evoked_dip, fwd, cov, 0.1, return_as_dipoles=True)
amp_max = [np.max(d.amplitude) for d in dip_gmap]
dip_gmap = dip_gmap[np.argmax(amp_max)]
assert (dip_gmap[0].pos[0] in src[0]['rr'][stc.vertices])
dip_fit = mne.fit_dipole(evoked_dip, cov, sphere)[0]
assert (np.abs(np.dot(dip_fit.ori[0], dip_gmap.ori[0])) > 0.99)
def add_volume_stcs(stc1, stc2):
"""Adds two SourceEstimates together, allowing for different vertices."""
vertices = np.union1d(stc1.vertices, stc2.vertices)
assert stc1.data.shape[1] == stc2.data.shape[1]
assert stc1.tmin == stc2.tmin
assert stc1.tstep == stc2.tstep
data = np.zeros((len(vertices), stc1.data.shape[1]))
for i, vert in enumerate(vertices):
if vert in stc1.vertices[0]:
data[[i]] += stc1.data[stc1.vertices[0] == vert]
if vert in stc2.vertices[0]:
data[[i]] += stc2.data[stc2.vertices[0] == vert]
return mne.VolSourceEstimate(data, [vertices],
tmin=stc1.tmin,
tstep=stc1.tstep)
def compute_fwds_stc(position, perts, sphere):
pos = position.copy()
pos['rr'] = mne.transforms.apply_trans(head_mri_t, position['rr']) # invert back to mri
pos['nn'] = mne.transforms.apply_trans(head_mri_t, position['nn'])
src = mne.setup_volume_source_space(subject=subject, pos=pos, mri=None,
sphere=(0, 0, 0, 90), bem=None,
surface=None, mindist=1.0, exclude=0.0,
subjects_dir=None, volume_label=None,
add_interpolator=True, verbose=None)
fwd_pert = make_pert_forward_solution(raw_fname, trans=trans, src=src, bem=sphere, perts=perts,
meg=True, eeg=False, mindist=1.0, n_jobs=1)
fwd = mne.make_forward_solution(raw_fname, trans=trans, src=src, bem=sphere,
meg=True, eeg=False, mindist=1.0, n_jobs=1)
fwd_fixed = mne.convert_forward_solution(fwd, surf_ori=True, force_fixed=True,
use_cps=True)
fwd_pert_fixed = mne.convert_forward_solution(fwd_pert, surf_ori=True, force_fixed=True,
use_cps=True)
amplitude = 1e-5
stc = mne.VolSourceEstimate(amplitude * np.eye(1), [[0]], tmin=0., tstep=1)
return fwd_fixed, fwd_pert_fixed, stc
def fit_dips(min_rad, max_rad, nn, sphere, perts, sourcenorm):
testsources = dict(rr=[], nn=[])
nsources = max_rad - min_rad + 1
vertices = np.zeros((nsources, 1))
for i in range(min_rad, max_rad + 1):
ex, ey, ez = sourcenorm[0], sourcenorm[1], sourcenorm[2]
source = [.001*i*ex, .001*i*ey, .001*i*ez]
normal = [nn[0], nn[1], nn[2]]
testsources['rr'].append(source)
testsources['nn'].append(normal)
vertices[i - min_rad] = i
pos = dict(rr=[0], nn=[0])
pos['rr'] = mne.transforms.apply_trans(head_mri_t, testsources['rr']) # invert back to mri
pos['nn'] = mne.transforms.apply_trans(head_mri_t, testsources['nn'])
src = mne.setup_volume_source_space(subject=subject, pos=pos, mri=None,
sphere=(0, 0, 0, 90), bem=None,
surface=None, mindist=1.0, exclude=0.0,
subjects_dir=None, volume_label=None,
add_interpolator=True, verbose=None)
fwd_pert = make_pert_forward_solution(raw_fname, trans=trans, src=src, bem=sphere, perts=perts,
meg=True, eeg=False, mindist=1.0, n_jobs=1)
fwd = mne.make_forward_solution(raw_fname, trans=trans, src=src, bem=sphere,
meg=True, eeg=False, mindist=1.0, n_jobs=1)
fwd_fixed = mne.convert_forward_solution(fwd, surf_ori=True, force_fixed=True,
use_cps=True)
fwd_pert_fixed = mne.convert_forward_solution(fwd_pert, surf_ori=True, force_fixed=True,
use_cps=True)
amplitude = 1e-5
source = np.eye(nsources) * amplitude
stc = mne.VolSourceEstimate(source, vertices, tmin=0., tstep=1)
evoked = mne.simulation.simulate_evoked(fwd_fixed, stc, info, cov, use_cps=True,
iir_filter=None)
evoked_pert = mne.simulation.simulate_evoked(fwd_pert_fixed, stc, info, cov, use_cps=True,
iir_filter=None)
dip_fit_long = mne.fit_dipole(evoked, cov_fname, sphere, trans)[0]
dip_fit_pert = mne.fit_dipole(evoked_pert, cov_fname, sphere, trans)[0]
return dip_fit_long, dip_fit_pert, testsources
continue
###############################################################################
# Create dist stc from simulated data
###############################################################################
vert_sel = sel['vertex'].to_numpy()
data_dist_sel = sel['dist'].to_numpy()
data_eval_sel = sel['eval'].to_numpy()
data_ori_sel = sel['ori_error'].to_numpy()
data_dist = np.zeros(shape=(vertno.shape[0], 1))
data_dist[vert_sel, 0] = data_dist_sel + 0.001
vstc_dist = mne.VolSourceEstimate(data=data_dist,
vertices=vertno,
tmin=0,
tstep=1 / info['sfreq'],
subject='sample')
data_eval = np.zeros(shape=(vertno.shape[0], 1))
data_eval[vert_sel, 0] = data_eval_sel + 0.001
vstc_eval = mne.VolSourceEstimate(data=data_eval,
vertices=vertno,
tmin=0,
tstep=1 / info['sfreq'],
subject='sample')
data_ori = np.zeros(shape=(vertno.shape[0], 1))
data_ori[vert_sel, 0] = data_ori_sel + 0.001
vstc_ori = mne.VolSourceEstimate(data=data_ori,
vertices=vertno,
def test_mxne_vol_sphere():
"""(TF-)MxNE with a sphere forward and volumic source space"""
evoked = read_evokeds(fname_data, condition=0, baseline=(None, 0))
evoked.crop(tmin=-0.05, tmax=0.2)
cov = read_cov(fname_cov)
evoked_l21 = evoked.copy()
evoked_l21.crop(tmin=0.081, tmax=0.1)
info = evoked.info
sphere = mne.make_sphere_model(r0=(0., 0., 0.), head_radius=0.080)
src = mne.setup_volume_source_space(subject=None,
pos=15.,
mri=None,
sphere=(0.0, 0.0, 0.0, 80.0),
bem=None,
mindist=5.0,
exclude=2.0)
fwd = mne.make_forward_solution(info,
trans=None,
src=src,
bem=sphere,
eeg=False,
meg=True)
alpha = 80.
assert_raises(ValueError,
mixed_norm,
evoked,
fwd,
cov,
alpha,
loose=0.0,
return_residual=False,
maxit=3,
tol=1e-8,
active_set_size=10)
assert_raises(ValueError,
mixed_norm,
evoked,
fwd,
cov,
alpha,
loose=0.2,
return_residual=False,
maxit=3,
tol=1e-8,
active_set_size=10)
# irMxNE tests
stc = mixed_norm(evoked_l21,
fwd,
cov,
alpha,
n_mxne_iter=1,
maxit=30,
tol=1e-8,
active_set_size=10)
assert_true(isinstance(stc, VolSourceEstimate))
assert_array_almost_equal(stc.times, evoked_l21.times, 5)
# Compare orientation obtained using fit_dipole and gamma_map
# for a simulated evoked containing a single dipole
stc = mne.VolSourceEstimate(50e-9 * np.random.RandomState(42).randn(1, 4),
vertices=stc.vertices[:1],
tmin=stc.tmin,
tstep=stc.tstep)
evoked_dip = mne.simulation.simulate_evoked(fwd,
stc,
info,
cov,
nave=1e9,
use_cps=True)
dip_mxne = mixed_norm(evoked_dip,
fwd,
cov,
alpha=80,
n_mxne_iter=1,
maxit=30,
tol=1e-8,
active_set_size=10,
return_as_dipoles=True)
amp_max = [np.max(d.amplitude) for d in dip_mxne]
dip_mxne = dip_mxne[np.argmax(amp_max)]
assert_true(dip_mxne.pos[0] in src[0]['rr'][stc.vertices])
dip_fit = mne.fit_dipole(evoked_dip, cov, sphere)[0]
assert_true(np.abs(np.dot(dip_fit.ori[0], dip_mxne.ori[0])) > 0.99)
# Do with TF-MxNE for test memory savings
alpha = 60. # overall regularization parameter
l1_ratio = 0.01 # temporal regularization proportion
stc, _ = tf_mixed_norm(evoked,
fwd,
cov,
maxit=3,
tol=1e-4,
tstep=16,
wsize=32,
window=0.1,
alpha=alpha,
l1_ratio=l1_ratio,
return_residual=True)
assert_true(isinstance(stc, VolSourceEstimate))
assert_array_almost_equal(stc.times, evoked.times, 5)
del fwd
##############################################################################
# Compute label time series and do envelope correlation
# -----------------------------------------------------
epochs.apply_hilbert() # faster to do in sensor space
stcs = apply_lcmv_epochs(epochs, filters, return_generator=True)
corr = envelope_correlation(stcs, verbose=True)
##############################################################################
# Compute the degree and plot it
# ------------------------------
degree = mne.connectivity.degree(corr, 0.15)
stc = mne.VolSourceEstimate(degree, src[0]['vertno'], 0, 1, 'bst_resting')
brain = stc.plot(src,
clim=dict(kind='percent', lims=[75, 85, 95]),
colormap='gnuplot',
subjects_dir=subjects_dir,
mode='glass_brain')
##############################################################################
# References
# ----------
# .. [1] Hipp JF, Hawellek DJ, Corbetta M, Siegel M, Engel AK (2012)
# Large-scale cortical correlation structure of spontaneous
# oscillatory activity. Nature Neuroscience 15:884–890
# .. [2] Khan S et al. (2018). Maturation trajectories of cortical
# resting-state networks depend on the mediating frequency band.
# Neuroimage 174:57–68
def add_source_to_raw(raw, fwd_disc_true, signal_vertex, signal_freq,
trial_length, n_trials, source_type):
"""
Add a new simulated dipole source to an existing raw. Operates on a copy of
the raw.
Parameters:
-----------
raw : instance of Raw
The raw data to add a new source to.
fwd_disc_true : instance of mne.Forward
The forward operator for the discrete source space created with
the true transformation file.
signal_vertex : int
The vertex where signal dipole is placed.
signal_freq : float
The frequency of the signal.
trial_length : float
Length of a single trial in samples.
n_trials : int
Number of trials to create.
source_type : 'chirp' | 'random'
Type of source signal to add.
Returns:
--------
raw : instance of Raw
The summation of the original raw and the simulated source.
stc_signal : instance of SourceEstimate
Source time courses of the new signal.
"""
sfreq = raw.info['sfreq']
trial_length = int((config.tmax - config.tmin) * sfreq)
times = np.arange(trial_length) / sfreq + config.tmin
src = fwd_disc_true['src']
signal_vert = src[0]['vertno'][signal_vertex]
data = np.zeros(len(times))
signal_part = times >= 0
if source_type == 'chirp':
data[signal_part] += generate_signal(times[signal_part],
signal_freq,
phase=0.5)
elif source_type == 'random':
data[signal_part] += generate_random(times[signal_part])
vertices = np.array([signal_vert])
stc_signal = mne.VolSourceEstimate(data=data[np.newaxis, :],
vertices=vertices,
tmin=times[0],
tstep=np.diff(times[:2])[0],
subject='sample')
raw_signal = simulate_raw_mne(raw.info,
stc_signal,
trans=None,
src=None,
bem=None,
forward=fwd_disc_true)
raw_signal = mne.concatenate_raws(
[raw_signal.copy() for _ in range(n_trials)])
raw = raw.copy()
raw_picks = mne.pick_types(raw.info, meg=True, eeg=False)
raw._data[raw_picks] += raw_signal._data[raw_picks]
return raw, stc_signal
def simulate_raw(info,
fwd_disc_true,
signal_vertex,
signal_freq,
n_trials,
noise_multiplier,
random_state,
n_noise_dipoles,
er_raw,
fn_stc_signal=None,
fn_simulated_raw=None,
fn_report_h5=None):
"""
Simulate raw time courses for two dipoles with frequencies
given by signal_freq1 and signal_freq2. Noise dipoles are
placed randomly in the whole cortex.
Parameters:
-----------
info : instance of Info | instance of Raw
The channel information to use for simulation.
fwd_disc_true : instance of mne.Forward
The forward operator for the discrete source space created with
the true transformation file.
signal_vertex : int
The vertex where signal dipole is placed.
signal_freq : float
The frequency of the signal.
n_trials : int
Number of trials to create.
noise_multiplier : float
Multiplier for the noise dipoles. For noise_multiplier equal to one
the signal and noise dipoles have the same magnitude.
random_state : None | int | instance of RandomState
If random_state is an int, it will be used as a seed for RandomState.
If None, the seed will be obtained from the operating system (see
RandomState for details). Default is None.
n_noise_dipoles : int
The number of noise dipoles to place within the volume.
er_raw : instance of Raw
Empty room measurement to be used as sensor noise.
fn_stc_signal : None | string
Path where the signal source time courses are to be saved. If None the file is not saved.
fn_simulated_raw : None | string
Path where the raw data is to be saved. If None the file is not saved.
fn_report_h5 : None | string
Path where the .h5 file for the report is to be saved.
Returns:
--------
raw : instance of Raw
Simulated raw file.
stc_signal : instance of SourceEstimate
Source time courses of the signal.
"""
sfreq = info['sfreq']
trial_length = int((config.tmax - config.tmin) * sfreq)
times = np.arange(trial_length) / sfreq + config.tmin
###############################################################################
# Simulate a single signal dipole source as signal
###############################################################################
# TODO: I think a discrete source space was used because mne.simulate_raw did not take volume source spaces -> test
src = fwd_disc_true['src']
signal_vert = src[0]['vertno'][signal_vertex]
data = np.asarray([generate_signal(times, freq=signal_freq)])
vertices = np.array([signal_vert])
stc_signal = mne.VolSourceEstimate(data=data,
vertices=[vertices],
tmin=times[0],
tstep=np.diff(times[:2])[0],
subject='sample')
if fn_stc_signal is not None:
set_directory(op.dirname(fn_stc_signal))
stc_signal.save(fn_stc_signal)
###############################################################################
# Create trials of simulated data
###############################################################################
# select n_noise_dipoles entries from rr and their corresponding entries from nn
raw_list = []
for i in range(n_trials):
# Simulate random noise dipoles
stc_noise = simulate_sparse_stc(src,
n_noise_dipoles,
times,
data_fun=generate_random,
random_state=random_state,
labels=None)
# Project to sensor space
stc = add_volume_stcs(stc_signal, noise_multiplier * stc_noise)
raw = simulate_raw_mne(info,
stc,
trans=None,
src=None,
bem=None,
forward=fwd_disc_true)
raw_list.append(raw)
print('%02d/%02d' % (i + 1, n_trials))
raw = mne.concatenate_raws(raw_list)
# Use empty room noise as sensor noise
raw_picks = mne.pick_types(raw.info, meg=True, eeg=False)
er_raw_picks = mne.pick_types(er_raw.info, meg=True, eeg=False)
raw._data[raw_picks] += er_raw._data[er_raw_picks, :len(raw.times)]
###############################################################################
# Save everything
###############################################################################
if fn_simulated_raw is not None:
set_directory(op.dirname(fn_simulated_raw))
raw.save(fn_simulated_raw, overwrite=True)
# Plot the simulated raw data in the report
if fn_report_h5 is not None:
from matplotlib import pyplot as plt
set_directory(op.dirname(fn_report_h5))
fn_report_html = fn_report_h5.rsplit('.h5')[0] + '.html'
now = datetime.now()
with mne.open_report(fn_report_h5) as report:
fig = plt.figure()
plt.plot(times, generate_signal(times, freq=10))
plt.xlabel('Time (s)')
ax = fig.axes[0]
add_text_next_to_xlabel(fig, ax,
now.strftime('%m/%d/%Y, %H:%M:%S'))
report.add_figs_to_section(fig,
now.strftime('Signal time course'),
section='Sensor-level',
replace=True)
fig = raw.plot()
# axis 1 contains the xlabel
ax = fig.axes[1]
add_text_next_to_xlabel(fig, ax,
now.strftime('%m/%d/%Y, %H:%M:%S'))
report.add_figs_to_section(fig,
now.strftime('Simulated raw'),
section='Sensor-level',
replace=True)
report.save(fn_report_html, overwrite=True, open_browser=False)
raw._annotations = mne.annotations.Annotations([], [], [])
return raw, stc_signal
def run():
t0 = time.time()
parser = get_optparser(__file__)
parser.add_option("--raw",
dest="raw_in",
help="Input raw FIF file",
metavar="FILE")
parser.add_option("--pos",
dest="pos",
default=None,
help="Position definition text file. Can be 'constant' "
"to hold the head position fixed",
metavar="FILE")
parser.add_option("--dipoles",
dest="dipoles",
default=None,
help="Dipole definition file",
metavar="FILE")
parser.add_option("--cov",
dest="cov",
help="Covariance to use for noise generation. Can be "
"'simple' to use a diagonal covariance, or 'off' to "
"omit noise",
metavar="FILE",
default='simple')
parser.add_option("--duration",
dest="duration",
default=None,
help="Duration of each epoch (sec). If omitted, the last"
" time point in the dipole definition file plus 200 ms "
"will be used",
type="float")
parser.add_option("-j",
"--jobs",
dest="n_jobs",
help="Number of jobs to"
" run in parallel",
type="int",
default=1)
parser.add_option("--out",
dest="raw_out",
help="Output raw filename",
metavar="FILE")
parser.add_option("--plot-dipoles",
dest="plot_dipoles",
help="Plot "
"input dipole positions",
action="store_true")
parser.add_option("--plot-raw",
dest="plot_raw",
help="Plot the resulting "
"raw traces",
action="store_true")
parser.add_option("--plot-evoked",
dest="plot_evoked",
help="Plot evoked "
"data",
action="store_true")
parser.add_option("-p",
"--plot",
dest="plot",
help="Plot dipoles, raw, "
"and evoked",
action="store_true")
parser.add_option("--overwrite",
dest="overwrite",
help="Overwrite the"
"output file if it exists",
action="store_true")
options, args = parser.parse_args()
raw_in = options.raw_in
pos = options.pos
raw_out = options.raw_out
dipoles = options.dipoles
n_jobs = options.n_jobs
plot = options.plot
plot_dipoles = options.plot_dipoles or plot
plot_raw = options.plot_raw or plot
plot_evoked = options.plot_evoked or plot
overwrite = options.overwrite
duration = options.duration
cov = options.cov
# check parameters
if not (raw_out or plot_raw or plot_evoked):
raise ValueError('data must either be saved (--out) or '
'plotted (--plot-raw or --plot_evoked)')
if raw_out and op.isfile(raw_out) and not overwrite:
raise ValueError('output file exists, use --overwrite (%s)' % raw_out)
if raw_in is None or pos is None or dipoles is None:
parser.print_help()
sys.exit(1)
s = 'Simulate raw data with head movements'
print('\n%s\n%s\n%s\n' % ('-' * len(s), s, '-' * len(s)))
# setup the simulation
with printer('Reading dipole definitions'):
if not op.isfile(dipoles):
raise IOError('dipole file not found:\n%s' % dipoles)
dipoles = np.loadtxt(dipoles, skiprows=1, dtype=float)
n_dipoles = dipoles.shape[0]
if dipoles.shape[1] != 8:
raise ValueError('dipoles must have 8 columns')
rr = dipoles[:, :3] * 1e-3
nn = dipoles[:, 3:6]
t = dipoles[:, 6:8]
duration = t.max() + 0.2 if duration is None else duration
if (t[:, 0] > t[:, 1]).any():
raise ValueError('found tmin > tmax in dipole file')
if (t < 0).any():
raise ValueError('found t < 0 in dipole file')
if (t > duration).any():
raise ValueError('found t > duration in dipole file')
amp = np.sqrt(np.sum(nn * nn, axis=1)) * 1e-9
mne.surface._normalize_vectors(nn)
nn[(nn == 0).all(axis=1)] = (1, 0, 0)
src = mne.SourceSpaces([
dict(rr=rr,
nn=nn,
inuse=np.ones(n_dipoles, int),
coord_frame=FIFF.FIFFV_COORD_HEAD)
])
for key in ['pinfo', 'nuse_tri', 'use_tris', 'patch_inds']:
src[0][key] = None
trans = {
'from': FIFF.FIFFV_COORD_HEAD,
'to': FIFF.FIFFV_COORD_MRI,
'trans': np.eye(4)
}
if (amp > 100e-9).any():
print('')
warnings.warn('Largest dipole amplitude %0.1f > 100 nA' %
(amp.max() * 1e9))
if pos == 'constant':
print('Holding head position constant')
pos = None
else:
with printer('Loading head positions'):
pos = mne.get_chpi_positions(pos)
with printer('Loading raw data file'):
with warnings.catch_warnings(record=True):
raw = mne.io.Raw(raw_in,
preload=False,
allow_maxshield=True,
verbose=False)
if cov == 'simple':
print('Using diagonal covariance for brain noise')
elif cov == 'off':
print('Omitting brain noise in the simulation')
cov = None
else:
with printer('Loading covariance file for brain noise'):
cov = mne.read_cov(cov)
with printer('Setting up spherical model'):
bem = mne.bem.make_sphere_model('auto',
'auto',
raw.info,
verbose=False)
# check that our sources are reasonable
rad = bem['layers'][0]['rad']
r0 = bem['r0']
outside = np.sqrt(np.sum((rr - r0)**2, axis=1)) >= rad
n_outside = outside.sum()
if n_outside > 0:
print('')
raise ValueError(
'%s dipole%s outside the spherical model, are your positions '
'in mm?' % (n_outside, 's were' if n_outside != 1 else ' was'))
with printer('Constructing source estimate'):
tmids = t.mean(axis=1)
t = np.round(t * raw.info['sfreq']).astype(int)
t[:, 1] += 1 # make it inclusive
n_samp = int(np.ceil(duration * raw.info['sfreq']))
data = np.zeros((n_dipoles, n_samp))
for di, (t_, amp_) in enumerate(zip(t, amp)):
data[di, t_[0]:t_[1]] = amp_ * np.hanning(t_[1] - t_[0])
stc = mne.VolSourceEstimate(data, np.arange(n_dipoles), 0,
1. / raw.info['sfreq'])
# do the simulation
print('')
raw_mv = simulate_raw(raw,
stc,
trans,
src,
bem,
cov=cov,
head_pos=pos,
chpi=True,
n_jobs=n_jobs,
verbose=True)
print('')
if raw_out:
with printer('Saving data'):
raw_mv.save(raw_out, overwrite=overwrite)
# plot results -- must be *after* save because we low-pass filter
if plot_dipoles:
with printer('Plotting dipoles'):
fig, axs = plt.subplots(1, 3, figsize=(10, 3), facecolor='w')
fig.canvas.set_window_title('Dipoles')
meg_info = mne.pick_info(
raw.info, mne.pick_types(raw.info, meg=True, eeg=False))
helmet_rr = [
ch['coil_trans'][:3, 3].copy() for ch in meg_info['chs']
]
helmet_nn = np.zeros_like(helmet_rr)
helmet_nn[:, 2] = 1.
surf = dict(rr=helmet_rr,
nn=helmet_nn,
coord_frame=FIFF.FIFFV_COORD_DEVICE)
helmet_rr = mne.surface.transform_surface_to(
surf, 'head', meg_info['dev_head_t'])['rr']
p = np.linspace(0, 2 * np.pi, 40)
x_sphere, y_sphere = rad * np.sin(p), rad * np.cos(p)
for ai, ax in enumerate(axs):
others = np.setdiff1d(np.arange(3), [ai])
ax.plot(helmet_rr[:, others[0]],
helmet_rr[:, others[1]],
marker='o',
linestyle='none',
alpha=0.1,
markeredgecolor='none',
markerfacecolor='b',
zorder=-2)
ax.plot(x_sphere + r0[others[0]],
y_sphere + r0[others[1]],
color='y',
alpha=0.25,
zorder=-1)
ax.quiver(rr[:, others[0]],
rr[:, others[1]],
amp * nn[:, others[0]],
amp * nn[:, others[1]],
angles='xy',
units='x',
color='k',
alpha=0.5)
ax.set_aspect('equal')
ax.set_xlabel(' - ' + 'xyz'[others[0]] + ' + ')
ax.set_ylabel(' - ' + 'xyz'[others[1]] + ' + ')
ax.set_xticks([])
ax.set_yticks([])
plt.setp(list(ax.spines.values()), color='none')
plt.tight_layout()
if plot_raw or plot_evoked:
with printer('Low-pass filtering simulated data'):
events = mne.find_events(raw_mv, 'STI101', verbose=False)
b, a = signal.butter(4,
40. / (raw.info['sfreq'] / 2.),
'low',
analog=False)
raw_mv.filter(None,
40.,
method='iir',
iir_params=dict(b=b, a=a),
verbose=False,
n_jobs=n_jobs)
if plot_raw:
with printer('Plotting raw data'):
raw_mv.plot(clipping='transparent', events=events, show=False)
if plot_evoked:
with printer('Plotting evoked data'):
picks = mne.pick_types(raw_mv.info, meg=True, eeg=True)
events[:, 2] = 1
evoked = mne.Epochs(raw_mv, events, {
'Simulated': 1
}, 0, duration, None, picks).average()
evoked.plot_topomap(np.unique(tmids), show=False)
print('\nTotal time: %0.1f sec' % (time.time() - t0))
sys.stdout.flush()
if any([plot_dipoles, plot_raw, plot_evoked]):
plt.show(block=True)
def apply_mft(fwdname,
datafile,
evocondition=None,
meg='mag',
exclude='bads',
mftpar=None,
calccdm=None,
cdmcut=0.,
cdmlabels=None,
subject=None,
save_stc=True,
verbose=False):
""" Apply MFT to specified data set.
Parameters
----------
fwdname: name of forward solution file
datafile: name of datafile (ave or raw)
evocondition: condition in case of evoked input file
meg: meg-channels to pick ['mag']
exclude: meg-channels to exclude ['bads']
mftpar: dictionary with parameters for MFT algorithm
calccdm : str | None
where str can be 'all', 'both', 'left', 'right'
cdmcut : (rel.) cut to use in cdm-calculations [0.]
cdmlabels: list of labels to analyse
entries for 'cdmlabels', 'jlglabels', 'jtotlabels'
in qualdata are returned, containing cdm,
longitudinal and total current for each label.
subject : str | None
The subject name. While not necessary, it is safer to set the
subject parameter to avoid analysis errors.
verbose: control variable for verbosity
False='CRITICAL','WARNING',True='INFO','DEBUG'
or 'chatty'='verbose' (>INFO,<DEBUG)
Returns
qualmft: dictionary with relerr,rdmerr,mag-arrays and
cdm-arrays (if requested)
stcdata: stc with ||cdv|| at fwdmag['source_rr']
(type corresponding to forward solution)
"""
twgbl0 = time.time()
tcgbl0 = time.clock()
# Use mftparm as local copy of mftpar to keep that ro.
mftparm = {}
if mftpar:
mftparm.update(mftpar)
mftparm.setdefault('iter', 8)
mftparm.setdefault('currexp', 1)
mftparm.setdefault('prbfct', 'uniform')
mftparm.setdefault('prbcnt')
mftparm.setdefault('prbhw')
mftparm.setdefault('regtype', 'PzetaE')
mftparm.setdefault('zetareg', 1.00)
mftparm.setdefault('solver', 'lu')
mftparm.setdefault('svrelcut', 5.e-4)
if mftparm['solver'] == 'svd':
use_svd = True
use_lud = False
svrelcut = mftparm['svrelcut']
elif mftparm['solver'] == 'lu' or mftparm['solver'] == 'ludecomp':
use_lud = True
use_svd = False
else:
raise ValueError(
">>>>> mftpar['solver'] must be either 'svd' or 'lu[decomp]'")
if mftparm['prbfct'].lower() == 'gauss':
if not mftparm['prbcnt'].all() or not mftparm['prbhw'].all():
raise ValueError(
">>>>> 'prbfct'='Gauss' requires 'prbcnt' and 'prbhw' entries")
elif mftparm['prbfct'].lower() != 'uniform' and mftparm['prbfct'].lower(
) != 'flat':
raise ValueError(">>>>> unrecognized keyword for 'prbfct'")
if mftparm['prbcnt'] == None and mftparm['prbhw'] == None:
prbcnt = np.array([0.0, 0.0, 0.0], ndmin=2)
prbdhw = np.array([0.0, 0.0, 0.0], ndmin=2)
else:
prbcnt = np.reshape(mftparm['prbcnt'],
(len(mftparm['prbcnt'].flatten()) / 3, 3))
prbdhw = np.reshape(mftparm['prbhw'],
(len(mftparm['prbhw'].flatten()) / 3, 3))
if prbcnt.shape != prbdhw.shape:
raise ValueError(
">>>>> mftpar['prbcnt'] and mftpar['prbhw'] must have same size")
verbosity = 1
if verbose == False or verbose == 'CRITICAL':
verbosity = -1
elif verbose == 'WARNING':
verbosity = 0
elif verbose == 'chatty' or verbose == 'verbose':
verbose = 'INFO'
verbosity = 2
elif verbose == 'DEBUG':
verbosity = 3
if verbosity >= 0:
print "meg-channels = ", meg
print "exclude-channels = ", exclude
print "mftpar['iter' ] = ", mftparm['iter']
print "mftpar['currexp' ] = ", mftparm['currexp']
print "mftpar['regtype' ] = ", mftparm['regtype']
print "mftpar['zetareg' ] = ", mftparm['zetareg']
print "mftpar['solver' ] = ", mftparm['solver']
print "mftpar['svrelcut'] = ", mftparm['svrelcut']
print "mftpar['prbfct' ] = ", mftparm['prbfct']
print "mftpar['prbcnt' ] = ", mftparm['prbcnt']
print "mftpar['prbhw' ] = ", mftparm['prbhw']
if mftparm['prbcnt'] != None or mftparm['prbhw'] != None:
for icnt in xrange(prbcnt.shape[0]):
print " pos(prbcnt[%d]) = " % (icnt + 1), prbcnt[icnt]
print " dhw(prbdhw[%d]) = " % (icnt + 1), prbdhw[icnt]
if calccdm:
print "calccdm = '%s' with rel. cut = %5.2f" % (calccdm, cdmcut)
if calccdm and (cdmcut < 0. or cdmcut >= 1.):
raise ValueError(">>>>> cdmcut must be in [0,1)")
# Msg will be written by mne.read_forward_solution()
fwd = mne.read_forward_solution(fwdname, verbose=verbose)
# Block off fixed_orientation fwd-s for now:
if fwd['source_ori'] == FIFF.FIFFV_MNE_FIXED_ORI:
raise ValueError(
">>>>> apply_mft() cannot handle fixed-orientation fwd-solutions")
# Select magnetometer channels:
fwdmag = mne.io.pick.pick_types_forward(fwd,
meg=meg,
ref_meg=False,
eeg=False,
exclude=exclude)
lfmag = fwdmag['sol']['data']
n_sens, n_loc = lfmag.shape
n_srcspace = len([s['vertno'] for s in fwdmag['src']])
if verbosity >= 2:
print "Leadfield size : n_sen x n_loc = %d x %d" % (n_sens, n_loc)
print "Number of source spaces = %d" % n_srcspace
if cdmlabels is not None:
if verbosity >= 1:
print "########## Searching for label(s) in source space(s)..."
tc0 = time.clock()
tw0 = time.time()
numcdmlabels = 0
labvrtstot = 0
labvrtsusd = 0
if cdmlabels is not None:
invmri_head_t = mne.transforms.invert_transform(
fwdmag['info']['mri_head_t'])
mrsrcpnt = np.zeros(fwdmag['source_rr'].shape)
mrsrcpnt = mne.transforms.apply_trans(invmri_head_t['trans'],
fwdmag['source_rr'])
offsets = [0]
for s in fwdmag['src']:
offsets = np.append(offsets, [offsets[-1] + s['nuse']])
labinds = []
ilab = 0
for label in cdmlabels:
ilab = ilab + 1
labvrts = []
# Find src corresponding to this label (match by position)
# (Assume surface-labels are in head-cd, vol-labels in MR-cs)
isrc = 0
for s in fwdmag['src']:
isrc += 1
labvrts = label.get_vertices_used(vertices=s['vertno'])
numlabvrts = len(labvrts)
if numlabvrts == 0:
continue
if not np.all(s['inuse'][labvrts]):
print "isrc = %d: label='%s' (np.all(s['inuse'][labvrts])=False)" % (
isrc, label.name)
continue
# labindx: indices of used label-vertices in this src-space + offset2'source_rr'
# iinlabused: indices of used label-vertices in this label
labindx = np.searchsorted(s['vertno'],
labvrts) + offsets[isrc - 1]
iinlabused = np.searchsorted(label.vertices, labvrts)
if s['type'] == 'surf':
if not np.allclose(mrsrcpnt[labindx, :],
label.pos[iinlabused]):
continue # mismatch
else:
if not np.allclose(fwdmag['source_rr'][labindx, :],
label.pos[iinlabused]):
continue # mismatch
if verbosity >= 1:
print "%3d %30s %7s: %5d verts %4d used" % \
(ilab, label.name, label.hemi, len(label.vertices), numlabvrts)
break # from src-space-loop
if len(labvrts) > 0:
labvrtstot += len(label.vertices)
labvrtsusd += len(labvrts)
labinds.append(labindx)
numcdmlabels = len(labinds)
else:
warnings.warn(
'NO vertex found for label \'%s\' in any source space' %
label.name)
if verbosity >= 1:
print "--> sums: %5d verts %4d used" % (labvrtstot, labvrtsusd)
tc1 = time.clock()
tw1 = time.time()
print "prep. labels took %.3f" % (
1000. * (tc1 - tc0)), "ms (%.3f s walltime)" % (tw1 - tw0)
if datafile.rfind('-ave.fif') > 0 or datafile.rfind('-ave.fif.gz') > 0:
if verbosity >= 0:
print "Reading evoked data from %s" % datafile
if evocondition is None:
#indatinfo = mne.io.read_info(datafile)
indathndl = mne.read_evokeds(datafile,
baseline=(None, 0),
verbose=verbose)
if len(indathndl) > 1:
raise ValueError(
">>>>> need to specify a condition for this datafile. Aborting-"
)
picks = mne.io.pick.pick_types(indathndl[0].info,
meg=meg,
ref_meg=False,
eeg=False,
stim=False,
exclude=exclude)
data = indathndl[0].data[picks, :]
else:
indathndl = mne.read_evokeds(datafile,
condition=evocondition,
baseline=(None, 0),
verbose=verbose)
#if len(indathndl) > 1:
# raise ValueError(">>>>> need to specify a condition for this datafile. Aborting-")
picks = mne.io.pick.pick_types(indathndl.info,
meg=meg,
ref_meg=False,
eeg=False,
stim=False,
exclude=exclude)
data = indathndl.data[picks, :]
elif datafile.rfind('-raw.fif') > 0 or datafile.rfind('-raw.fif.gz') > 0:
if verbosity >= 0:
print "Reading raw data from %s" % datafile
indathndl = mne.io.Raw(datafile, preload=True, verbose=verbose)
picks = mne.io.pick.pick_types(indathndl.info,
meg=meg,
ref_meg=False,
eeg=False,
stim=False,
exclude=exclude)
data = indathndl._data[picks, :]
else:
raise ValueError(
">>>>> datafile is neither 'ave' nor 'raw'. Aborting-")
if verbosity >= 3:
print "data.shape = ", data.shape
if n_sens != data.shape[0]:
raise ValueError(
">>>>> Mismatch in #channels for forward (%d) and data (%d) files. Aborting."
% (n_sens, data.shape[0]))
tptotwall = 0.
tptotcpu = 0.
nptotcall = 0
tltotwall = 0.
tltotcpu = 0.
nltotcall = 0
tpcdmwall = 0.
tpcdmcpu = 0.
npcdmcall = 0
if verbosity >= 1:
print "########## Calculate initial prob-dist:"
tw0 = time.time()
tc0 = time.clock()
if mftpar['prbfct'] == 'Gauss':
wtmp = np.zeros(n_loc / 3)
for icnt in xrange(prbcnt.shape[0]):
testdiff = fwdmag['source_rr'] - prbcnt[icnt, :]
testdiff = testdiff / prbdhw[icnt, :]
testdiff = testdiff * testdiff
testsq = np.sum(testdiff, 1)
wtmp += np.exp(-testsq)
wdist0 = wtmp / (np.sum(wtmp) * np.sqrt(3.))
elif mftpar['prbfct'] == 'flat' or mftpar['prbfct'] == 'uniform':
if verbosity >= 2:
print "Setting initial w=const !"
wdist0 = np.ones(n_loc / 3) / (float(n_loc) / np.sqrt(3.))
else:
raise ValueError(
">>>>> mftpar['prbfct'] must be 'Gauss' or 'uniform'/'flat'")
wdist3 = np.repeat(wdist0, 3)
if verbosity >= 3:
wvecnorm = np.sum(
np.sqrt(
np.sum(np.reshape(wdist3, (wdist3.shape[0] / 3, 3))**2,
axis=1)))
print "sum(||wvec(i)||) = ", wvecnorm
tc1 = time.clock()
tw1 = time.time()
if verbosity >= 1:
print "calc(wdist0) took %.3f" % (
1000. * (tc1 - tc0)), "ms (%.3f s walltime)" % (tw1 - tw0)
if verbosity >= 1:
print "########## Calculate P-matrix, incl. weights:"
tw0 = time.time()
tc0 = time.clock()
lfw = lfmag * np.repeat(np.sqrt(wdist0), 3)
pmat0 = np.einsum('ik,jk->ij', lfw, lfw)
# Avoiding sqrt is expensive!
# pmat0 = np.einsum('ik, k, jk -> ij', lfmag, wdist3, lfmag)
tc1 = time.clock()
tw1 = time.time()
tptotwall += (tw1 - tw0)
tptotcpu += (tc1 - tc0)
nptotcall += 1
if verbosity >= 1:
print "calc(lf*w*lf.T) took ", 1000. * (
tc1 - tc0), "ms (%.3f s walltime)" % (tw1 - tw0)
# Normalize P:
pmax = np.amax([np.abs(np.amax(pmat0)), np.abs(np.amin(pmat0))])
if verbosity >= 3:
print "pmax(init) = ", pmax
pscalefct = 1.
while pmax > 1.0:
pmax /= 2.
pscalefct /= 2.
while pmax < 0.5:
pmax *= 2.
pscalefct *= 2.
#print ">>>>> Keeping scale factor eq 1"
#pscalefct = 1.
pmat0 = pmat0 * pscalefct
if verbosity >= 3:
print "pmax(fin.) = ", np.amax(
[np.abs(np.amax(pmat0)),
np.abs(np.amin(pmat0))])
# Regularize P:
if mftparm['regtype'] == 'PzetaE':
zetatrp = mftparm['zetareg'] * np.trace(pmat0) / float(pmat0.shape[0])
if verbosity >= 3:
print "Use PzetaE-regularization with zeta*tr(P)/ncol(P) = %12.5e" % zetatrp
ptilde0 = pmat0 + zetatrp * np.identity(pmat0.shape[0])
elif mftparm['regtype'] == 'classic' or mftparm['regtype'] == 'PPzetaP':
zetatrp = mftparm['zetareg'] * np.trace(pmat0) / float(pmat0.shape[0])
if verbosity >= 3:
print "Use PPzetaP-regularization with zeta*tr(P)/ncol(P) = %12.5e" % zetatrp
ptilde0 = np.dot(pmat0, pmat0) + zetatrp * pmat0
else:
raise ValueError(
">>>>> mftpar['regtype'] must be 'PzetaE' or 'classic''")
# decompose:
if use_lud is True:
LU0, P0 = scipy.linalg.lu_factor(ptilde0)
#rhstmp = np.zeros([LU0.shape[1]])
#xtmp = np.empty([LU0.shape[1]])
#xtmp = scipy.linalg.lu_solve((LU0,P0),rhstmp)
if verbosity >= 3:
# Calculate condition number:
#(sign, lndetbf) = np.linalg.slogdet(ptilde0)
lndettr = np.sum(np.log(np.abs(np.diag(LU0))))
#print "lndet(ptilde0) = %8.3f =?= %8.3f = sum(log(|diag(LU0)|))" % (lndetbf,lndettr)
# log(prod(a_i, i=1,n)) for a_i = sqrt(sum(ptilde0_ij^2, j=1,n))
denom = np.sum(np.log(np.sqrt(np.sum(ptilde0 * ptilde0, axis=0))))
lncondno = lndettr - denom
print "ln(condno) = %8.3f, K_H = 10^(%8.3f) = %8.3f" % (
lncondno, lncondno / np.log(10.), np.exp(lncondno))
print "(K_H < 0.01 : bad, K_H > 0.1 : good)"
if use_svd is True:
U, s, V = np.linalg.svd(ptilde0, full_matrices=True)
if verbosity >= 2:
print ">>> SV range %e ... %e" % (np.amax(s), np.amin(s))
dtmp = s.max() * svrelcut
s *= (abs(s) >= dtmp)
sinv = [
1. / s[k] if s[k] != 0. else 0. for k in xrange(ptilde0.shape[0])
]
if verbosity >= 2:
print ">>> With rel-cutoff=%e %d out of %d SVs remain" % \
(svrelcut,np.array(np.nonzero(sinv)).shape[1],len(sinv))
if verbosity >= 3:
stat = np.allclose(ptilde0, np.dot(U, np.dot(np.diag(s), V)))
print ">>> Testing svd-result: %s" % stat
if not stat:
print " (Maybe due to SV-cutoff?)"
print ">>> Setting ptildeinv=(U diag(sinv) V).tr"
ptilde0inv = np.transpose(np.dot(U, np.dot(np.diag(sinv), V)))
if verbosity >= 3:
stat = np.allclose(np.identity(ptilde0.shape[0]),
np.dot(ptilde0inv, ptilde0))
if stat:
print ">>> Testing ptilde0inv-result (shld be unit-matrix): ok"
else:
print ">>> Testing ptilde0inv-result (shld be unit-matrix): failed"
print np.transpose(np.dot(ptilde0inv, ptilde0))
print ">>>"
if verbosity >= 1:
print "########## Create stc data and qual data arrays:"
qualdata = {
'relerr': np.zeros(data.shape[1]),
'rdmerr': np.zeros(data.shape[1]),
'mag': np.zeros(data.shape[1])
}
if calccdm is not None:
if verbosity >= 0 and \
n_srcspace ==1 and (calccdm == 'left' or calccdm == 'right'):
print ">>>Warning>> cdm-results may differ from what you expect."
ids = data.shape[1]
if calccdm == 'all':
(qualdata['cdmall'], qualdata['jlgall']) = (np.zeros(ids),
np.zeros(ids))
(qualdata['cdmleft'], qualdata['jlgleft']) = (np.zeros(ids),
np.zeros(ids))
(qualdata['cdmright'], qualdata['jlgright']) = (np.zeros(ids),
np.zeros(ids))
elif calccdm == 'both':
(qualdata['cdmleft'], qualdata['jlgleft']) = (np.zeros(ids),
np.zeros(ids))
(qualdata['cdmright'], qualdata['jlgright']) = (np.zeros(ids),
np.zeros(ids))
elif calccdm == 'left':
(qualdata['cdmleft'], qualdata['jlgleft']) = (np.zeros(ids),
np.zeros(ids))
elif calccdm == 'right':
(qualdata['cdmright'], qualdata['jlgright']) = (np.zeros(ids),
np.zeros(ids))
elif calccdm == 'glob':
(qualdata['cdmall'], qualdata['jlgall']) = (np.zeros(ids),
np.zeros(ids))
if qualdata.has_key('cdmleft'):
fwdlhinds = np.where(fwdmag['source_rr'][:, 0] < 0.)[0]
if qualdata.has_key('cdmright'):
fwdrhinds = np.where(fwdmag['source_rr'][:, 0] > 0.)[0]
if cdmlabels is not None and numcdmlabels > 0:
qualdata['cdmlabels'] = np.zeros((numcdmlabels, data.shape[1]))
qualdata['jlglabels'] = np.zeros((numcdmlabels, data.shape[1]))
qualdata['jtotlabels'] = np.zeros((numcdmlabels, data.shape[1]))
stcdata = np.zeros([n_loc / 3, data.shape[1]])
if verbosity >= 2:
print "Reading %d slices of data to calc. cdv:" % data.shape[1]
if data.shape[1] > 1000:
print " "
for islice in xrange(data.shape[1]):
wdist = np.copy(wdist0)
wdist3 = np.repeat(wdist, 3)
pmat = np.copy(pmat0)
ptilde = np.copy(ptilde0)
if use_svd is True:
ptildeinv = np.copy(ptilde0inv)
if use_lud is True:
LU = np.copy(LU0)
P = np.copy(P0)
slice = pscalefct * data[:, islice]
if mftparm['regtype'] == 'PzetaE':
mtilde = np.copy(slice)
else:
mtilde = np.dot(pmat, slice)
acoeff = np.empty([ptilde.shape[0]])
if use_svd is True:
for irow in xrange(ptilde.shape[0]):
acoeff[irow] = np.dot(ptildeinv[irow, :], mtilde)
if use_lud is True:
acoeff = scipy.linalg.lu_solve((LU, P), mtilde)
cdv = np.zeros(n_loc)
cdvnorms = np.zeros(n_loc / 3)
for krow in xrange(lfmag.shape[0]):
lfwtmp = lfmag[krow, :] * wdist3
cdv += acoeff[krow] * lfwtmp
tlw0 = time.time()
tlc0 = time.clock()
for mftiter in xrange(mftparm['iter']):
# MFT iteration loop:
cdvecs = np.reshape(cdv, (cdv.shape[0] / 3, 3))
cdvnorms = np.sqrt(np.sum(cdvecs**2, axis=1))
wdist = np.power(cdvnorms, mftparm['currexp']) * wdist0
wdistsum = np.sum(wdist)
wdist = wdist / wdistsum
wdist3 = np.repeat(wdist, 3)
# Calculate new P-matrix, incl. weights:
tw0 = time.time()
tc0 = time.clock()
lfw = lfmag * np.repeat(np.sqrt(pscalefct * wdist), 3)
pmat = np.einsum('ik,jk->ij', lfw, lfw)
tc1 = time.clock()
tw1 = time.time()
tptotwall += (tw1 - tw0)
tptotcpu += (tc1 - tc0)
nptotcall += 1
# Regularize P:
if mftparm['regtype'] == 'PzetaE':
ptilde = pmat + zetatrp * np.identity(pmat.shape[0])
else:
ptilde = np.dot(pmat, pmat) + zetatrp * pmat
# decompose:
if use_svd is True:
U, s, V = np.linalg.svd(ptilde, full_matrices=True)
dtmp = s.max() * svrelcut
s *= (abs(s) >= dtmp)
sinv = [
1. / s[k] if s[k] != 0. else 0.
for k in xrange(ptilde.shape[0])
]
ptildeinv = np.transpose(np.dot(U, np.dot(np.diag(sinv), V)))
for irow in xrange(ptilde.shape[0]):
acoeff[irow] = np.dot(ptildeinv[irow, :], mtilde)
if use_lud is True:
LU, P = scipy.linalg.lu_factor(ptilde)
acoeff = scipy.linalg.lu_solve((LU, P), mtilde)
cdv = np.einsum('ji,i,j->i', lfmag, wdist3, acoeff)
tc1 = time.clock()
tw1 = time.time()
tltotwall += (tw1 - tlw0)
tltotcpu += (tc1 - tlc0)
nltotcall += 1
cdvecs = np.reshape(cdv, (cdv.shape[0] / 3, 3))
cdvnorms = np.sqrt(np.sum(cdvecs**2, axis=1))
#(relerr,rdmerr,mag) = compare_est_exp(ptilde,acoeff,mtilde)
(relerr, rdmerr, mag) = compare_est_exp(pmat, acoeff, slice)
qualdata['relerr'][islice] = relerr
qualdata['rdmerr'][islice] = rdmerr
qualdata['mag'][islice] = mag
tc0 = time.clock()
tw0 = time.time()
if qualdata.has_key('cdmall'):
(qualdata['cdmall'][islice],
qualdata['jlgall'][islice]) = scan_cdm_w_cut(cdv, cdmcut)
if qualdata.has_key('cdmleft'):
(qualdata['cdmleft'][islice],qualdata['jlgleft'][islice]) = \
scan_cdm_w_cut(cdvecs[fwdlhinds,:],cdmcut)
if qualdata.has_key('cdmright'):
(qualdata['cdmright'][islice],qualdata['jlgright'][islice]) = \
scan_cdm_w_cut(cdvecs[fwdrhinds,:],cdmcut)
if qualdata.has_key('cdmlabels'):
for ilab in xrange(numcdmlabels):
(qualdata['cdmlabels'][ilab,islice],qualdata['jlglabels'][ilab,islice]) = \
scan_cdm_w_cut(cdvecs[labinds[ilab],:],cdmcut)
qualdata['jtotlabels'][ilab,islice] = \
calc_jtotal_w_cut(cdvecs[labinds[ilab],:],cdmcut)
tc1 = time.clock()
tw1 = time.time()
tpcdmwall += (tw1 - tw0)
tpcdmcpu += (tc1 - tc0)
npcdmcall += 1
# Write final cdv to file:
for iloc in xrange(n_loc / 3):
stcdata[iloc, islice] = cdvnorms[iloc]
del wdist
if verbosity >= 2 and islice > 0 and islice % 1000 == 0:
print "\r%6d out of %6d slices done." % (islice, data.shape[1])
if verbosity >= 2 and data.shape[1] > 1000:
print "Done."
vertices = [s['vertno'] for s in fwdmag['src']]
tstep = 1. / indathndl.info['sfreq']
tmin = indathndl.times[0]
if len(vertices) == 1:
stc_mft = mne.VolSourceEstimate(stcdata,
vertices=fwdmag['src'][0]['vertno'],
tmin=tmin,
tstep=tstep,
subject=subject)
elif len(vertices) == 2:
vertices = [s['vertno'] for s in fwdmag['src']]
stc_mft = mne.SourceEstimate(stcdata,
vertices=vertices,
tmin=tmin,
tstep=tstep,
subject=subject)
else:
vertices = np.concatenate(([s['vertno'] for s in fwdmag['src']]))
stc_mft = mne.VolSourceEstimate(stcdata,
vertices=vertices,
tmin=tmin,
tstep=tstep,
subject=subject)
stcdatamft = stc_mft.data
print "##### Results:"
for islice in xrange(data.shape[1]):
print "slice=%4d: relerr=%9.3e rdmerr=%9.3e mag=%5.3f cdvmax=%9.2e" % \
(islice, qualdata['relerr'][islice], qualdata['rdmerr'][islice], qualdata['mag'][islice],\
np.amax(stcdatamft[:, islice]))
stat = np.allclose(stcdata, stcdatamft, atol=0., rtol=1e-07)
if stat:
print "stcdata from mft-SR and old calc agree."
else:
print "stcdata from mft-SR and old calc DIFFER."
if save_stc:
# save Surface stc.
print "##### Trying to save stc:"
stcmft_fname = os.path.join(
os.path.dirname(datafile),
os.path.basename(datafile).split('-')[0]) + "mft"
print "stcmft basefilename: %s" % stcmft_fname
stc_mft.save(stcmft_fname, verbose=True)
print "##### done."
twgbl1 = time.time()
tcgbl1 = time.clock()
if verbosity >= 1:
print "calc(lf*w*lf.T) took total %9.2f s CPU-time (%9.2f s walltime)" % (
tptotcpu, tptotwall)
print "calc(lf*w*lf.T) took per call %9.2fms CPU-time (%9.2fms walltime)" % \
(1000.*tptotcpu/float(nptotcall),1000.*tptotwall/float(nptotcall))
print "scan_cdm calls took total %9.2f s CPU-time (%9.2f s walltime)" % (
tpcdmcpu, tpcdmwall)
print "scan_cdm calls took per call %9.2fms CPU-time (%9.2fms walltime)" % \
(1000.*tpcdmcpu/float(npcdmcall),1000.*tpcdmwall/float(npcdmcall))
print "iteration-loops took total %9.2f s CPU-time (%9.2f s walltime)" % (
tltotcpu, tltotwall)
print "iteration-loops took per call %9.2fms CPU-time (%9.2fms walltime)" % \
(1000.*tltotcpu/float(nltotcall),1000.*tltotwall/float(nltotcall))
print "Total mft-call took total %9.2f s CPU-time (%9.2f s walltime)" % (
(tcgbl1 - tcgbl0), (twgbl1 - twgbl0))
return (fwdmag, qualdata, stc_mft)
# normalize to unit length
tangential = (tangential.T * (1. / np.linalg.norm(tangential, axis=1))).T
nn = tangential
###############################################################################
# Simulate a single signal dipole source as signal
###############################################################################
signal_vertex = src[0]['vertno'][config.vertex]
data = np.asarray([generate_signal(times, freq=config.signal_freq)])
vertices = np.array([signal_vertex])
stc_signal = mne.VolSourceEstimate(data=data,
vertices=vertices,
tmin=0,
tstep=1 / info['sfreq'],
subject='sample')
stc_signal.save(vfname.stc_signal(noise=config.noise, vertex=config.vertex))
###############################################################################
# Create discrete source space based on voxels in volume source space
###############################################################################
if not op.exists(vfname.fwd_discrete):
pos = {'rr': rr, 'nn': nn}
# make discrete source space
src_disc = mne.setup_volume_source_space(subject='sample',
def rsa_source_level(stcs,
dsm_model,
src,
spatial_radius=0.04,
temporal_radius=0.1,
stc_dsm_metric='correlation',
stc_dsm_params=dict(),
rsa_metric='spearman',
y=None,
n_folds=1,
sel_vertices=None,
tmin=None,
tmax=None,
n_jobs=1,
verbose=False):
"""Perform RSA in a searchlight pattern across the source space.
The output is a source estimate where the "signal" at each source point is
the RSA, computed for a patch surrounding the source point.
Parameters
----------
stcs : list of mne.SourceEstimate | list of mne.VolSourceEstimate
For each item, a source estimate for the brain activity.
dsm_model : ndarray, shape (n, n) | (n * (n - 1) // 2,) | list of ndarray
The model DSM, see :func:`compute_dsm`. For efficiency, you can give it
in condensed form, meaning only the upper triangle of the matrix as a
vector. See :func:`scipy.spatial.distance.squareform`. To perform RSA
against multiple models at the same time, supply a list of model DSMs.
Use :func:`compute_dsm` to compute DSMs.
src : instance of mne.SourceSpaces
The source space used by the source estimates specified in the `stcs`
parameter.
spatial_radius : float | None
The spatial radius of the searchlight patch in meters. All source
points within this radius will belong to the searchlight patch. Set to
None to only perform the searchlight over time, flattening across
sensors. Defaults to 0.04.
temporal_radius : float | None
The temporal radius of the searchlight patch in seconds. Set to None to
only perform the searchlight over sensors, flattening across time.
Defaults to 0.1.
stc_dsm_metric : str
The metric to use to compute the DSM for the source estimates. This can
be any metric supported by the scipy.distance.pdist function. See also
the ``stc_dsm_params`` parameter to specify and additional parameter
for the distance function. Defaults to 'correlation'.
stc_dsm_params : dict
Extra arguments for the distance metric used to compute the DSMs.
Refer to :mod:`scipy.spatial.distance` for a list of all other metrics
and their arguments. Defaults to an empty dictionary.
rsa_metric : str
The RSA metric to use to compare the DSMs. Valid options are:
* 'spearman' for Spearman's correlation (the default)
* 'pearson' for Pearson's correlation
* 'kendall-tau-a' for Kendall's Tau (alpha variant)
* 'partial' for partial Pearson correlations
* 'partial-spearman' for partial Spearman correlations
* 'regression' for linear regression weights
Defaults to 'spearman'.
y : ndarray of int, shape (n_items,) | None
For each source estimate, a number indicating the item to which it
belongs. When ``None``, each source estimate is assumed to belong to a
different item. Defaults to ``None``.
n_folds : int | None
Number of folds to use when using cross-validation to compute the
evoked DSM metric. Specify ``None``, to use the maximum number of folds
possible, given the data.
Defaults to 1 (no cross-validation).
sel_vertices : list of int | None
When set, searchlight patches will only be generated for the subset of
vertices/voxels with the given indices. Defaults to ``None``, in which
case patches for all vertices/voxels are generated.
tmin : float | None
When set, searchlight patches will only be generated from subsequent
time points starting from this time point. This value is given in
seconds. Defaults to ``None``, in which case patches are generated
starting from the first time point.
tmax : float | None
When set, searchlight patches will only be generated up to and
including this time point. This value is given in seconds. Defaults to
``None``, in which case patches are generated up to and including the
last time point.
n_jobs : int
The number of processes (=number of CPU cores) to use. Specify -1 to
use all available cores. Defaults to 1.
verbose : bool
Whether to display a progress bar. In order for this to work, you need
the tqdm python module installed. Defaults to False.
Returns
-------
stc : SourceEstimate | VolSourceEstimate | list of SourceEstimate | list of VolSourceEstimate
The correlation values for each searchlight patch. When spatial_radius
is set to None, there will only be one vertex. When temporal_radius is
set to None, there will only be one time point. When multiple models
have been supplied, a list will be returned containing the RSA results
for each model.
See Also
--------
compute_dsm
""" # noqa E501
# Check for compatibility of the source estimated and the model features
one_model = type(dsm_model) is np.ndarray
if one_model:
dsm_model = [dsm_model]
# Check for compatibility of the evokeds and the model features
for dsm in dsm_model:
n_items = _n_items_from_dsm(dsm)
if len(stcs) != n_items and y is None:
raise ValueError(
'The number of source estimates (%d) should be equal to the '
'number of items in `dsm_model` (%d). Alternatively, use '
'the `y` parameter to assign evokeds to items.' %
(len(stcs), n_items))
if y is not None and len(np.unique(y)) != n_items:
raise ValueError(
'The number of items in `dsm_model` (%d) does not match '
'the number of items encoded in the `y` matrix (%d).' %
(n_items, len(np.unique(y))))
_check_compatible(stcs, src)
dist = _get_distance_matrix(src, dist_lim=spatial_radius, n_jobs=n_jobs)
if temporal_radius is not None:
# Convert the temporal radius to samples
temporal_radius = int(temporal_radius // stcs[0].tstep)
if temporal_radius < 1:
raise ValueError('Temporal radius is less than one sample.')
sel_samples = _tmin_tmax_to_indices(stcs[0].times, tmin, tmax)
# Perform the RSA
X = np.array([stc.data for stc in stcs])
patches = searchlight(X.shape,
dist=dist,
spatial_radius=spatial_radius,
temporal_radius=temporal_radius,
sel_series=sel_vertices,
sel_samples=sel_samples)
data = rsa_array(X,
dsm_model,
patches,
data_dsm_metric=stc_dsm_metric,
data_dsm_params=stc_dsm_params,
rsa_metric=rsa_metric,
y=y,
n_folds=n_folds,
n_jobs=n_jobs,
verbose=verbose)
# Pack the result in a SourceEstimate object
if spatial_radius is not None:
vertices = stcs[0].vertices
else:
if src.kind == 'volume':
vertices = [np.array([1])]
else:
vertices = [np.array([1]), np.array([])]
tmin = _construct_tmin(stcs[0].times, sel_samples, temporal_radius)
tstep = stcs[0].tstep
if one_model:
if src.kind == 'volume':
return mne.VolSourceEstimate(data,
vertices,
tmin,
tstep,
subject=stcs[0].subject)
else:
return mne.SourceEstimate(data,
vertices,
tmin,
tstep,
subject=stcs[0].subject)
else:
if src.kind == 'volume':
return [
mne.VolSourceEstimate(data[:, :, i],
vertices,
tmin,
tstep,
subject=stcs[0].subject)
for i in range(data.shape[-1])
]
else:
return [
mne.SourceEstimate(data[:, :, i],
vertices,
tmin,
tstep,
subject=stcs[0].subject)
for i in range(data.shape[-1])
]
def apply_mft(fwdname,
datafile,
evocondition=None,
meg='mag',
exclude='bads',
mftpar=None,
subject=None,
save_stc=True,
verbose=False):
""" Apply MFT to specified data set.
Parameters
----------
fwdname: name of forward solution file
datafile: name of datafile (ave or raw)
evocondition: condition in case of evoked input file
meg: meg-channels to pick ['mag']
exclude: meg-channels to exclude ['bads']
mftpar: dictionary with parameters for MFT algorithm
subject : str | None
The subject name. While not necessary, it is safer to set the
subject parameter to avoid analysis errors.
verbose: control variable for verbosity
False='CRITICAL','WARNING',True='INFO','DEBUG'
or 'chatty'='verbose' (>INFO,<DEBUG)
Returns
-------
qualmft: dictionary with relerr,rdmerr,mag-arrays
stcdata: stc with ||cdv|| at fwdmag['source_rr']
(type corresponding to forward solution)
"""
twgbl0 = time.time()
tcgbl0 = time.clock()
if mftpar is None:
mftpar = {
'prbfct': 'uniform',
'prbcnt': None,
'prbhw': None,
'iter': 8,
'currexp': 1,
'regtype': 'PzetaE',
'zetareg': 1.00,
'solver': 'lu',
'svrelcut': 5.e-4
}
if mftpar['solver'] == 'svd':
use_svd = True
use_lud = False
svrelcut = mftpar['svrelcut']
elif mftpar['solver'] == 'lu' or mftpar['solver'] == 'ludecomp':
use_lud = True
use_svd = False
else:
raise ValueError(
">>>>> mftpar['solver'] must be either 'svd' or 'lu[decomp]'")
if mftpar['prbcnt'] == None and mftpar['prbhw'] == None:
prbcnt = np.array([0.0, 0.0, 0.0], ndmin=2)
prbdhw = np.array([0.0, 0.0, 0.0], ndmin=2)
else:
prbcnt = np.reshape(mftpar['prbcnt'],
(len(mftpar['prbcnt'].flatten()) / 3, 3))
prbdhw = np.reshape(mftpar['prbhw'],
(len(mftpar['prbhw'].flatten()) / 3, 3))
if prbcnt.shape != prbdhw.shape:
raise ValueError(
">>>>> mftpar['prbcnt'] and mftpar['prbhw'] must have same size")
verbosity = 1
if verbose == False or verbose == 'CRITICAL':
verbosity = -1
elif verbose == 'WARNING':
verbosity = 0
elif verbose == 'chatty' or verbose == 'verbose':
verbose = 'INFO'
verbosity = 2
elif verbose == 'DEBUG':
verbosity = 3
if verbosity >= 0:
print "meg-channels = ", meg
print "exclude-channels = ", exclude
print "mftpar['iter' ] = ", mftpar['iter']
print "mftpar['regtype' ] = ", mftpar['regtype']
print "mftpar['zetareg' ] = ", mftpar['zetareg']
print "mftpar['solver' ] = ", mftpar['solver']
print "mftpar['svrelcut'] = ", mftpar['svrelcut']
print "mftpar['prbfct' ] = ", mftpar['prbfct']
print "mftpar['prbcnt' ] = ", mftpar['prbcnt']
print "mftpar['prbhw' ] = ", mftpar['prbhw']
if mftpar['prbcnt'] != None or mftpar['prbhw'] != None:
for icnt in xrange(prbcnt.shape[0]):
print " pos(prbcnt[%d]) = " % (icnt + 1), prbcnt[icnt]
print " dhw(prbdhw[%d]) = " % (icnt + 1), prbdhw[icnt]
# Msg will be written by mne.read_forward_solution()
fwd = mne.read_forward_solution(fwdname, verbose=verbose)
# Block off fixed_orientation fwd-s for now:
if fwd['source_ori'] == FIFF.FIFFV_MNE_FIXED_ORI:
raise ValueError(
">>>>> apply_mft() cannot handle fixed-orientation fwd-solutions")
# Select magnetometer channels:
fwdmag = mne.io.pick.pick_types_forward(fwd,
meg=meg,
ref_meg=False,
eeg=False,
exclude=exclude)
lfmag = fwdmag['sol']['data']
n_sens, n_loc = lfmag.shape
if verbosity >= 2:
print "Leadfield size : n_sen x n_loc = %d x %d" % (n_sens, n_loc)
if datafile.rfind('-ave.fif') > 0 or datafile.rfind('-ave.fif.gz') > 0:
if verbosity >= 0:
print "Reading evoked data from %s" % datafile
if evocondition is None:
#indatinfo = mne.io.read_info(datafile)
indathndl = mne.read_evokeds(datafile,
baseline=(None, 0),
verbose=verbose)
if len(indathndl) > 1:
raise ValueError(
">>>>> need to specify a condition for this datafile. Aborting-"
)
picks = mne.io.pick.pick_types(indathndl[0].info,
meg=meg,
ref_meg=False,
eeg=False,
stim=False,
exclude=exclude)
data = indathndl[0].data[picks, :]
else:
indathndl = mne.read_evokeds(datafile,
condition=evocondition,
baseline=(None, 0),
verbose=verbose)
#if len(indathndl) > 1:
# raise ValueError(">>>>> need to specify a condition for this datafile. Aborting-")
picks = mne.io.pick.pick_types(indathndl.info,
meg=meg,
ref_meg=False,
eeg=False,
stim=False,
exclude=exclude)
data = indathndl.data[picks, :]
elif datafile.rfind('-raw.fif') > 0 or datafile.rfind('-raw.fif.gz') > 0:
if verbosity >= 0:
print "Reading raw data from %s" % datafile
indathndl = mne.io.Raw(datafile, preload=True, verbose=verbose)
picks = mne.io.pick.pick_types(indathndl.info,
meg=meg,
ref_meg=False,
eeg=False,
stim=False,
exclude=exclude)
data = indathndl._data[picks, :]
else:
raise ValueError(
">>>>> datafile is neither 'ave' nor 'raw'. Aborting-")
if verbosity >= 3:
print "data.shape = ", data.shape
if n_sens != data.shape[0]:
raise ValueError(
">>>>> Mismatch in #channels for forward (%d) and data (%d) files. Aborting."
% (n_sens, data.shape[0]))
tptotwall = 0.
tptotcpu = 0.
nptotcall = 0
tltotwall = 0.
tltotcpu = 0.
nltotcall = 0
tpcdmwall = 0.
tpcdmcpu = 0.
npcdmcall = 0
if verbosity >= 1:
print "########## Calculate initial prob-dist:"
tw0 = time.time()
tc0 = time.clock()
if mftpar['prbfct'] == 'Gauss':
wtmp = np.zeros(n_loc / 3)
for icnt in xrange(prbcnt.shape[0]):
testdiff = fwdmag['source_rr'] - prbcnt[icnt, :]
testdiff = testdiff / prbdhw[icnt, :]
testdiff = testdiff * testdiff
testsq = np.sum(testdiff, 1)
wtmp += np.exp(-testsq)
wdist0 = wtmp / (np.sum(wtmp) * np.sqrt(3.))
elif mftpar['prbfct'] == 'flat' or mftpar['prbfct'] == 'uniform':
if verbosity >= 2:
print "Setting w=const !"
wdist0 = np.ones(n_loc / 3) / (float(n_loc) / np.sqrt(3.))
else:
raise ValueError(
">>>>> mftpar['prbfct'] must be 'Gauss' or 'uniform'/'flat'")
wdist3 = np.repeat(wdist0, 3)
if verbosity >= 3:
wvecnorm = np.sum(
np.sqrt(
np.sum(np.reshape(wdist3, (wdist3.shape[0] / 3, 3))**2,
axis=1)))
print "sum(||wvec(i)||) = ", wvecnorm
tc1 = time.clock()
tw1 = time.time()
if verbosity >= 1:
print "calc(wdist0) took %.3f" % (
1000. * (tc1 - tc0)), "ms (%.3f s walltime)" % (tw1 - tw0)
if verbosity >= 1:
print "########## Calculate P-matrix, incl. weights:"
tw0 = time.time()
tc0 = time.clock()
wdist3rt = np.repeat(np.sqrt(wdist0), 3)
lfw = lfmag * wdist3rt
pmat0 = np.einsum('ik,jk->ij', lfw, lfw)
tc1 = time.clock()
tw1 = time.time()
tptotwall += (tw1 - tw0)
tptotcpu += (tc1 - tc0)
nptotcall += 1
if verbosity >= 1:
print "calc(lf*w*lf.T) took ", 1000. * (
tc1 - tc0), "ms (%.3f s walltime)" % (tw1 - tw0)
# Normalize P:
pmax = np.amax([np.abs(np.amax(pmat0)), np.abs(np.amin(pmat0))])
if verbosity >= 3:
print "pmax(init) = ", pmax
pscalefct = 1.
while pmax > 1.0:
pmax /= 2.
pscalefct /= 2.
while pmax < 0.5:
pmax *= 2.
pscalefct *= 2.
#print ">>>>> Keeping scale factor eq 1"
#pscalefct = 1.
pmat0 = pmat0 * pscalefct
if verbosity >= 3:
print "pmax(fin.) = ", np.amax(
[np.abs(np.amax(pmat0)),
np.abs(np.amin(pmat0))])
# Regularize P:
if mftpar['regtype'] == 'PzetaE':
zetatrp = mftpar['zetareg'] * np.trace(pmat0) / float(pmat0.shape[0])
if verbosity >= 3:
print "Use PzetaE-regularization with zeta*tr(P)/ncol(P) = %12.5e" % zetatrp
ptilde0 = pmat0 + zetatrp * np.identity(pmat0.shape[0])
elif mftpar['regtype'] == 'classic' or mftpar['regtype'] == 'PPzetaP':
zetatrp = mftpar['zetareg'] * np.trace(pmat0) / float(pmat0.shape[0])
if verbosity >= 3:
print "Use PPzetaP-regularization with zeta*tr(P)/ncol(P) = %12.5e" % zetatrp
ptilde0 = np.dot(pmat0, pmat0) + zetatrp * pmat0
else:
raise ValueError(
">>>>> mftpar['regtype'] must be 'PzetaE' or 'classic''")
# decompose:
if use_lud is True:
LU0, P0 = scipy.linalg.lu_factor(ptilde0)
#rhstmp = np.zeros([LU0.shape[1]])
#xtmp = np.empty([LU0.shape[1]])
#xtmp = scipy.linalg.lu_solve((LU0,P0),rhstmp)
if verbosity >= 3:
# Calculate condition number:
#(sign, lndetbf) = np.linalg.slogdet(ptilde0)
lndettr = np.sum(np.log(np.abs(np.diag(LU0))))
#print "lndet(ptilde0) = %8.3f =?= %8.3f = sum(log(|diag(LU0)|))" % (lndetbf,lndettr)
# log(prod(a_i, i=1,n)) for a_i = sqrt(sum(ptilde0_ij^2, j=1,n))
denom = np.sum(np.log(np.sqrt(np.sum(ptilde0 * ptilde0, axis=0))))
lncondno = lndettr - denom
print "ln(condno) = %8.3f, K_H = 10^(%8.3f) = %8.3f" % (
lncondno, lncondno / np.log(10.), np.exp(lncondno))
print "(K_H < 0.01 : bad, K_H > 0.1 : good)"
if use_svd is True:
U, s, V = np.linalg.svd(ptilde0, full_matrices=True)
dtmp = s.max() * svrelcut
s *= (abs(s) >= dtmp)
sinv = [
1. / s[k] if s[k] != 0. else 0. for k in xrange(ptilde0.shape[0])
]
if verbosity >= 2:
print ">>> With rel-cutoff=%e %d out of %d SVs remain" % \
(svrelcut,np.array(np.nonzero(sinv)).shape[1],len(sinv))
if verbosity >= 3:
stat = np.allclose(ptilde0, np.dot(U, np.dot(np.diag(s), V)))
print ">>> Testing svd-result: %s" % stat
if not stat:
print " (Maybe due to SV-cutoff?)"
print ">>> Setting ptildeinv=(U diag(sinv) V).tr"
ptilde0inv = np.transpose(np.dot(U, np.dot(np.diag(sinv), V)))
if verbosity >= 3:
stat = np.allclose(np.identity(ptilde0.shape[0]),
np.dot(ptilde0inv, ptilde0))
if stat:
print ">>> Testing ptilde0inv-result (shld be unit-matrix): ok"
else:
print ">>> Testing ptilde0inv-result (shld be unit-matrix): failed"
print np.transpose(np.dot(ptilde0inv, ptilde0))
print ">>>"
if verbosity >= 1:
print "########## Create stc data and qual data arrays:"
qualdata = {
'relerr': np.zeros(data.shape[1]),
'rdmerr': np.zeros(data.shape[1]),
'mag': np.zeros(data.shape[1])
}
stcdata = np.zeros([n_loc / 3, data.shape[1]])
stcdata1 = [np.zeros([s['nuse'], data.shape[1]]) for s in fwdmag['src']]
stcinds = np.zeros((n_loc / 3, 2), dtype=int)
stcinds1 = np.zeros((n_loc / 3), dtype=int)
offsets = np.append([0], [s['nuse'] for s in fwdmag['src']])
iblck = -1
nmatch = 0
for s in fwdmag['src']:
iblck = iblck + 1
for kvert0 in xrange(s['nuse']):
kvert1 = offsets[iblck] + kvert0
if np.all(
np.equal(fwdmag['source_rr'][kvert1],
s['rr'][s['vertno'][kvert0]])):
stcinds[kvert1][0] = iblck
stcinds[kvert1][1] = kvert0
nmatch = nmatch + 1
if verbosity >= 3:
print "Found %d matches in creating source_rr/rr index table." % nmatch
if verbosity >= 2:
print "Reading slices of data to calc. cdv:"
if data.shape[1] > 1000:
print " "
for islice in xrange(data.shape[1]):
wdist = np.copy(wdist0)
wdist3 = np.repeat(wdist, 3)
pmat = np.copy(pmat0)
ptilde = np.copy(ptilde0)
if use_svd is True:
ptildeinv = np.copy(ptilde0inv)
if use_lud is True:
LU = np.copy(LU0)
P = np.copy(P0)
slice = pscalefct * data[:, islice]
if mftpar['regtype'] == 'PzetaE':
mtilde = np.copy(slice)
else:
mtilde = np.dot(pmat, slice)
acoeff = np.empty([ptilde.shape[0]])
if use_svd is True:
for irow in xrange(ptilde.shape[0]):
acoeff[irow] = np.dot(ptildeinv[irow, :], mtilde)
if use_lud is True:
acoeff = scipy.linalg.lu_solve((LU, P), mtilde)
cdv = np.zeros(n_loc)
cdvnorms = np.zeros(n_loc / 3)
for krow in xrange(lfmag.shape[0]):
lfwtmp = lfmag[krow, :] * wdist3
cdv += acoeff[krow] * lfwtmp
tlw0 = time.time()
tlc0 = time.clock()
for mftiter in xrange(mftpar['iter']):
# MFT iteration loop:
cdvecs = np.reshape(cdv, (cdv.shape[0] / 3, 3))
cdvnorms = np.sqrt(np.sum(cdvecs**2, axis=1))
wdist = np.power(cdvnorms, mftpar['currexp']) * wdist0
wdistsum = np.sum(wdist)
wdist = wdist / wdistsum
wdist3 = np.repeat(wdist, 3)
# Calculate new P-matrix, incl. weights:
tw0 = time.time()
tc0 = time.clock()
wdist3rt = np.repeat(np.sqrt(pscalefct * wdist), 3)
lfw = lfmag * wdist3rt
pmat = np.einsum('ik,jk->ij', lfw, lfw)
tc1 = time.clock()
tw1 = time.time()
tptotwall += (tw1 - tw0)
tptotcpu += (tc1 - tc0)
nptotcall += 1
# Regularize P:
if mftpar['regtype'] == 'PzetaE':
ptilde = pmat + zetatrp * np.identity(pmat.shape[0])
else:
ptilde = np.dot(pmat, pmat) + zetatrp * pmat
# decompose:
if use_svd is True:
U, s, V = np.linalg.svd(ptilde, full_matrices=True)
dtmp = s.max() * svrelcut
s *= (abs(s) >= dtmp)
sinv = [
1. / s[k] if s[k] != 0. else 0.
for k in xrange(ptilde.shape[0])
]
ptildeinv = np.transpose(np.dot(U, np.dot(np.diag(sinv), V)))
for irow in xrange(ptilde.shape[0]):
acoeff[irow] = np.dot(ptildeinv[irow, :], mtilde)
if use_lud is True:
LU, P = scipy.linalg.lu_factor(ptilde)
acoeff = scipy.linalg.lu_solve((LU, P), mtilde)
cdv = np.einsum('ji,i,j->i', lfmag, wdist3, acoeff)
tc1 = time.clock()
tw1 = time.time()
tltotwall += (tw1 - tlw0)
tltotcpu += (tc1 - tlc0)
nltotcall += 1
cdvecs = np.reshape(cdv, (cdv.shape[0] / 3, 3))
cdvnorms = np.sqrt(np.sum(cdvecs**2, axis=1))
# (relerr,rdmerr,mag) = compare_est_exp(ptilde,acoeff,mtilde)
(relerr, rdmerr, mag) = compare_est_exp(pmat, acoeff, slice)
qualdata['relerr'][islice] = relerr
qualdata['rdmerr'][islice] = rdmerr
qualdata['mag'][islice] = mag
tc0 = time.clock()
tw0 = time.time()
tc1 = time.clock()
tw1 = time.time()
tpcdmwall += (tw1 - tw0)
tpcdmcpu += (tc1 - tc0)
npcdmcall += 1
# Write final cdv to file:
for iloc in xrange(n_loc / 3):
#stcdata1[stcinds[iloc][0]][stcinds[iloc][1],islice] = cdvnorms[iloc]
stcdata[iloc, islice] = cdvnorms[iloc]
del wdist
if verbosity >= 2 and islice > 0 and islice % 1000 == 0:
print "\r%6d out of %6d slices done." % (islice, data.shape[1])
if verbosity >= 2 and data.shape[1] > 1000:
print "Done."
vertices = [s['vertno'] for s in fwdmag['src']]
tstep = 1. / indathndl.info['sfreq']
tmin = indathndl.times[0]
if len(vertices) == 1:
stc_mft = mne.VolSourceEstimate(stcdata,
vertices=fwdmag['src'][0]['vertno'],
tmin=tmin,
tstep=tstep,
subject=subject)
else:
vertices = [s['vertno'] for s in fwdmag['src']]
stc_mft = mne.SourceEstimate(stcdata,
vertices=vertices,
tmin=tmin,
tstep=tstep,
subject=subject)
stcdatamft = stc_mft.data
print "##### Results:"
for islice in xrange(data.shape[1]):
print "slice=%4d: relerr=%9.3e rdmerr=%9.3e mag=%5.3f cdvmax=%9.2e" % \
(islice, qualdata['relerr'][islice], qualdata['rdmerr'][islice], qualdata['mag'][islice],\
np.amax(stcdatamft[:, islice]))
# (islice,qualdata[0,islice],qualdata[1,islice],qualdata[2,islice], \
# np.amax([np.amax(sb[:,islice]) for sb in stcdatamft]))
stat = np.allclose(stcdata, stcdatamft, atol=0., rtol=1e-07)
if stat:
print "stcdata from mft-SR and old calc agree."
else:
print "stcdata from mft-SR and old calc DIFFER."
if save_stc:
# save Surface stc.
print "##### Trying to save stc:"
stcmft_fname = os.path.join(
os.path.dirname(datafile),
os.path.basename(datafile).split('-')[0]) + "mft"
print "stcmft basefilename: %s" % stcmft_fname
stc_mft.save(stcmft_fname, verbose=True)
print "##### done."
write_tab_files = True
if write_tab_files:
time_idx = np.argmax(np.max(stcdata, axis=0))
tabfilenam = 'testtab.dat'
print "##### Creating %s with |cdv(time_idx=%d)|" % (tabfilenam,
time_idx)
tabfile = open(tabfilenam, mode='w')
cdvnmax = np.max(stcdata[:, time_idx])
tabfile.write("# time_idx = %d\n" % time_idx)
tabfile.write("# max amplitude = %11.4e\n" % cdvnmax)
tabfile.write("# x/mm y/mm z/mm |cdv| index\n")
for ipnt in xrange(n_loc / 3):
copnt = 1000. * fwdmag['source_rr'][ipnt]
tabfile.write(" %7.2f %7.2f %7.2f %11.4e %5d\n" % \
(copnt[0], copnt[1], copnt[2], stcdata[ipnt, time_idx], ipnt))
tabfile.close()
tabfilenam = 'testwtab.dat'
print "##### Creating %s with wdist0" % tabfilenam
tabfile = open(tabfilenam, mode='w')
tabfile.write("# time_idx = %d\n" % time_idx)
for icnt in xrange(prbcnt.shape[0]):
cocnt = 1000. * prbcnt[icnt, :]
tabfile.write("# center %7.2f %7.2f %7.2f\n" %
(cocnt[0], cocnt[1], cocnt[2]))
tabfile.write("# max value = %11.4e\n" % np.max(wdist0))
tabfile.write("# x/mm y/mm z/mm wdist0 index")
for icnt in xrange(prbcnt.shape[0]):
tabfile.write(" d_%d/mm" % (icnt + 1))
tabfile.write("\n")
for ipnt in xrange(n_loc / 3):
copnt = 1000. * fwdmag['source_rr'][ipnt]
tabfile.write(" %7.2f %7.2f %7.2f %11.4e %5d" %\
(copnt[0],copnt[1],copnt[2],wdist0[ipnt],ipnt))
for icnt in xrange(prbcnt.shape[0]):
cocnt = 1000. * prbcnt[icnt, :]
dist = np.sqrt(np.dot((copnt - cocnt), (copnt - cocnt)))
tabfile.write(" %7.2f" % dist)
tabfile.write("\n")
tabfile.close()
twgbl1 = time.time()
tcgbl1 = time.clock()
if verbosity >= 1:
print "calc(lf*w*lf.T) took total %9.2f s CPU-time (%9.2f s walltime)" % (
tptotcpu, tptotwall)
print "calc(lf*w*lf.T) took per call %9.2fms CPU-time (%9.2fms walltime)" % \
(1000.*tptotcpu/float(nptotcall),1000.*tptotwall/float(nptotcall))
print "iteration-loops took total %9.2f s CPU-time (%9.2f s walltime)" % (
tltotcpu, tltotwall)
print "iteration-loops took per call %9.2fms CPU-time (%9.2fms walltime)" % \
(1000.*tltotcpu/float(nltotcall),1000.*tltotwall/float(nltotcall))
print "Total mft-call took total %9.2f s CPU-time (%9.2f s walltime)" % (
(tcgbl1 - tcgbl0), (twgbl1 - twgbl0))
return (fwdmag, qualdata, stc_mft)
src = mne.setup_volume_source_space(
pos=7.5,
subject='spm',
bem=bem,
subjects_dir='/home/robbis/mne_data/MNE-spm-face/subjects/')
voldata = np.zeros((src[0]['nuse'], data.shape[0]))
vertno = src[0]['vertno']
vertpos = src[0]['rr'][vertno]
dist = cdist(vertpos, inside)
minvert = np.argmin(dist, axis=0)
voldata[minvert] = data.T
stc = mne.VolSourceEstimate(voldata, [src[0]['vertno']], 0, 1, subject='spm')
brain = stc.plot(
src,
colormap='gnuplot',
subjects_dir='/home/robbis/mne_data/MNE-spm-face/subjects/',
mode='glass_brain',
clim=dict(kind='percent', lims=[75, 85, 95]),
initial_pos=np.array([-36, -25, 60]) / 1000.,
)
morph = mne.compute_source_morph(src,
subject_from=None,
subject_to='fsaverage',
subjects_dir=None,
zooms='auto',
niter_affine=(100, 100, 10),