def simu_bimodal_meg(snr=6, white=True, seed=None, freqs=[3, 50], n_cycles=[1, 1.5],
phases=[0, 0], offsets=[0, 80]):
tmin = -0.1
sfreq = 1000. # Hz
tstep = 1. / sfreq
n_samples = 600
times = np.linspace(tmin, tmin + n_samples * tstep, n_samples)
# Generate times series from 2 Morlet wavelets
stc_data = np.zeros((len(labels), len(times)))
Ws = morlet(sfreq, freqs, n_cycles=n_cycles)
stc_data[0][:len(Ws[0])] = np.real(Ws[0] * np.exp(1j * phases[0]))
stc_data[1][:len(Ws[1])] = np.real(Ws[1] * np.exp(1j * phases[1]))
stc_data *= 100 * 1e-9 # use nAm as unit
# time translation
stc_data[0] = np.roll(stc_data[0], offsets[0])
stc_data[1] = np.roll(stc_data[1], offsets[1])
stc = generate_sparse_stc(fwd['src'], labels, stc_data, tmin, tstep,
random_state=0)
###############################################################################
# Generate noisy evoked data
picks = mne.pick_types(raw.info, meg=True, exclude='bads')
if white:
iir_filter = None # for white noise
else:
iir_filter = iir_filter_raw(raw, order=5, picks=picks, tmin=60, tmax=180)
evoked = generate_evoked(fwd, stc, evoked_template, cov, snr,
tmin=0.0, tmax=0.2, iir_filter=iir_filter, random_state=seed)
return evoked
def test_simulate_evoked():
""" Test simulation of evoked data """
raw = mne.fiff.Raw(raw_fname)
fwd = read_forward_solution(fwd_fname, force_fixed=True)
fwd = pick_types_forward(fwd, meg=True, eeg=True, exclude=raw.info['bads'])
cov = mne.read_cov(cov_fname)
label_names = ['Aud-lh', 'Aud-rh']
labels = [
read_label(
op.join(data_path, 'MEG', 'sample', 'labels', '%s.label' % label))
for label in label_names
]
evoked_template = mne.fiff.read_evoked(ave_fname, setno=0, baseline=None)
evoked_template = pick_types_evoked(evoked_template,
meg=True,
eeg=True,
exclude=raw.info['bads'])
snr = 6 # dB
tmin = -0.1
sfreq = 1000. # Hz
tstep = 1. / sfreq
n_samples = 600
times = np.linspace(tmin, tmin + n_samples * tstep, n_samples)
# Generate times series from 2 Morlet wavelets
stc_data = np.zeros((len(labels), len(times)))
Ws = morlet(sfreq, [3, 10], n_cycles=[1, 1.5])
stc_data[0][:len(Ws[0])] = np.real(Ws[0])
stc_data[1][:len(Ws[1])] = np.real(Ws[1])
stc_data *= 100 * 1e-9 # use nAm as unit
# time translation
stc_data[1] = np.roll(stc_data[1], 80)
stc = generate_sparse_stc(fwd['src'],
labels,
stc_data,
tmin,
tstep,
random_state=0)
# Generate noisy evoked data
iir_filter = [1, -0.9]
evoked = generate_evoked(fwd,
stc,
evoked_template,
cov,
snr,
tmin=0.0,
tmax=0.2,
iir_filter=iir_filter)
assert_array_almost_equal(evoked.times, stc.times)
assert_true(len(evoked.data) == len(fwd['sol']['data']))
def test_simulate_evoked():
""" Test simulation of evoked data """
raw = Raw(raw_fname)
fwd = read_forward_solution(fwd_fname, force_fixed=True)
fwd = pick_types_forward(fwd, meg=True, eeg=True, exclude=raw.info['bads'])
cov = read_cov(cov_fname)
label_names = ['Aud-lh', 'Aud-rh']
labels = [read_label(op.join(data_path, 'MEG', 'sample', 'labels',
'%s.label' % label)) for label in label_names]
evoked_template = read_evokeds(ave_fname, condition=0, baseline=None)
evoked_template = pick_types_evoked(evoked_template, meg=True, eeg=True,
exclude=raw.info['bads'])
snr = 6 # dB
tmin = -0.1
sfreq = 1000. # Hz
tstep = 1. / sfreq
n_samples = 600
times = np.linspace(tmin, tmin + n_samples * tstep, n_samples)
# Generate times series from 2 Morlet wavelets
stc_data = np.zeros((len(labels), len(times)))
Ws = morlet(sfreq, [3, 10], n_cycles=[1, 1.5])
stc_data[0][:len(Ws[0])] = np.real(Ws[0])
stc_data[1][:len(Ws[1])] = np.real(Ws[1])
stc_data *= 100 * 1e-9 # use nAm as unit
# time translation
stc_data[1] = np.roll(stc_data[1], 80)
stc = generate_sparse_stc(fwd['src'], labels, stc_data, tmin, tstep,
random_state=0)
# Generate noisy evoked data
iir_filter = [1, -0.9]
with warnings.catch_warnings(record=True):
warnings.simplefilter('always') # positive semidefinite warning
evoked = generate_evoked(fwd, stc, evoked_template, cov, snr,
tmin=0.0, tmax=0.2, iir_filter=iir_filter)
assert_array_almost_equal(evoked.times, stc.times)
assert_true(len(evoked.data) == len(fwd['sol']['data']))
# make a vertex that doesn't exist in fwd, should throw error
stc_bad = stc.copy()
mv = np.max(fwd['src'][0]['vertno'][fwd['src'][0]['inuse']])
stc_bad.vertices[0][0] = mv + 1
assert_raises(RuntimeError, generate_evoked, fwd, stc_bad,
evoked_template, cov, snr, tmin=0.0, tmax=0.2)
def test_simulate_evoked():
""" Test simulation of evoked data """
raw = mne.fiff.Raw(raw_fname)
fwd = read_forward_solution(fwd_fname, force_fixed=True)
fwd = pick_types_forward(fwd, meg=True, eeg=True, exclude=raw.info['bads'])
cov = mne.read_cov(cov_fname)
label_names = ['Aud-lh', 'Aud-rh']
labels = [read_label(op.join(data_path, 'MEG', 'sample', 'labels',
'%s.label' % label)) for label in label_names]
evoked_template = mne.fiff.read_evoked(ave_fname, setno=0, baseline=None)
evoked_template = pick_types_evoked(evoked_template, meg=True, eeg=True,
exclude=raw.info['bads'])
snr = 6 # dB
tmin = -0.1
sfreq = 1000. # Hz
tstep = 1. / sfreq
n_samples = 600
times = np.linspace(tmin, tmin + n_samples * tstep, n_samples)
# Generate times series from 2 Morlet wavelets
stc_data = np.zeros((len(labels), len(times)))
Ws = morlet(sfreq, [3, 10], n_cycles=[1, 1.5])
stc_data[0][:len(Ws[0])] = np.real(Ws[0])
stc_data[1][:len(Ws[1])] = np.real(Ws[1])
stc_data *= 100 * 1e-9 # use nAm as unit
# time translation
stc_data[1] = np.roll(stc_data[1], 80)
stc = generate_sparse_stc(fwd['src'], labels, stc_data, tmin, tstep,
random_state=0)
# Generate noisy evoked data
iir_filter = [1, -0.9]
evoked = generate_evoked(fwd, stc, evoked_template, cov, snr,
tmin=0.0, tmax=0.2, iir_filter=iir_filter)
assert_array_almost_equal(evoked.times, stc.times)
assert_true(len(evoked.data) == len(fwd['sol']['data']))
stc_data[1] = np.roll(stc_data[1], 80)
stc = generate_sparse_stc(fwd['src'],
labels,
stc_data,
tmin,
tstep,
random_state=0)
###############################################################################
# Generate noisy evoked data
picks = pick_types(raw.info, meg=True, exclude='bads')
iir_filter = fit_iir_model_raw(raw, order=5, picks=picks, tmin=60, tmax=180)[1]
evoked = generate_evoked(fwd,
stc,
evoked_template,
cov,
snr,
tmin=0.0,
tmax=0.2,
iir_filter=iir_filter)
###############################################################################
# Plot
plot_sparse_source_estimates(fwd['src'],
stc,
bgcolor=(1, 1, 1),
opacity=0.5,
high_resolution=True)
plt.figure()
plt.psd(evoked.data[0])
def test_dipole_fitting():
"""Test dipole fitting"""
amp = 10e-9
tempdir = _TempDir()
rng = np.random.RandomState(0)
fname_dtemp = op.join(tempdir, 'test.dip')
fname_sim = op.join(tempdir, 'test-ave.fif')
fwd = convert_forward_solution(read_forward_solution(fname_fwd),
surf_ori=False, force_fixed=True)
evoked = read_evokeds(fname_evo)[0]
cov = read_cov(fname_cov)
n_per_hemi = 5
vertices = [np.sort(rng.permutation(s['vertno'])[:n_per_hemi])
for s in fwd['src']]
nv = sum(len(v) for v in vertices)
stc = SourceEstimate(amp * np.eye(nv), vertices, 0, 0.001)
with warnings.catch_warnings(record=True): # semi-def cov
evoked = generate_evoked(fwd, stc, evoked, cov, snr=20,
random_state=rng)
# For speed, let's use a subset of channels (strange but works)
picks = np.sort(np.concatenate([
pick_types(evoked.info, meg=True, eeg=False)[::2],
pick_types(evoked.info, meg=False, eeg=True)[::2]]))
evoked.pick_channels([evoked.ch_names[p] for p in picks])
evoked.add_proj(make_eeg_average_ref_proj(evoked.info))
write_evokeds(fname_sim, evoked)
# Run MNE-C version
run_subprocess([
'mne_dipole_fit', '--meas', fname_sim, '--meg', '--eeg',
'--noise', fname_cov, '--dip', fname_dtemp,
'--mri', fname_fwd, '--reg', '0', '--tmin', '0',
])
dip_c = read_dipole(fname_dtemp)
# Run mne-python version
sphere = make_sphere_model(head_radius=0.1)
dip, residuals = fit_dipole(evoked, fname_cov, sphere, fname_fwd)
# Sanity check: do our residuals have less power than orig data?
data_rms = np.sqrt(np.sum(evoked.data ** 2, axis=0))
resi_rms = np.sqrt(np.sum(residuals ** 2, axis=0))
assert_true((data_rms > resi_rms).all())
# Compare to original points
transform_surface_to(fwd['src'][0], 'head', fwd['mri_head_t'])
transform_surface_to(fwd['src'][1], 'head', fwd['mri_head_t'])
src_rr = np.concatenate([s['rr'][v] for s, v in zip(fwd['src'], vertices)],
axis=0)
src_nn = np.concatenate([s['nn'][v] for s, v in zip(fwd['src'], vertices)],
axis=0)
# MNE-C skips the last "time" point :(
dip.crop(dip_c.times[0], dip_c.times[-1])
src_rr, src_nn = src_rr[:-1], src_nn[:-1]
# check that we did at least as well
corrs, dists, gc_dists, amp_errs, gofs = [], [], [], [], []
for d in (dip_c, dip):
new = d.pos
diffs = new - src_rr
corrs += [np.corrcoef(src_rr.ravel(), new.ravel())[0, 1]]
dists += [np.sqrt(np.mean(np.sum(diffs * diffs, axis=1)))]
gc_dists += [180 / np.pi * np.mean(np.arccos(np.sum(src_nn * d.ori,
axis=1)))]
amp_errs += [np.sqrt(np.mean((amp - d.amplitude) ** 2))]
gofs += [np.mean(d.gof)]
assert_true(dists[0] >= dists[1], 'dists: %s' % dists)
assert_true(corrs[0] <= corrs[1], 'corrs: %s' % corrs)
assert_true(gc_dists[0] >= gc_dists[1], 'gc-dists (ori): %s' % gc_dists)
assert_true(amp_errs[0] >= amp_errs[1], 'amplitude errors: %s' % amp_errs)
times = np.linspace(tmin, tmin + n_samples * tstep, n_samples)
# Generate times series from 2 Morlet wavelets
stc_data = np.zeros((len(labels), len(times)))
Ws = morlet(sfreq, [3, 10], n_cycles=[1, 1.5])
stc_data[0][:len(Ws[0])] = np.real(Ws[0])
stc_data[1][:len(Ws[1])] = np.real(Ws[1])
stc_data *= 100 * 1e-9 # use nAm as unit
# time translation
stc_data[1] = np.roll(stc_data[1], 80)
stc = generate_sparse_stc(fwd['src'], labels, stc_data, tmin, tstep,
random_state=0)
###############################################################################
# Generate noisy evoked data
picks = mne.fiff.pick_types(raw.info, meg=True, exclude='bads')
iir_filter = iir_filter_raw(raw, order=5, picks=picks, tmin=60, tmax=180)
evoked = generate_evoked(fwd, stc, evoked_template, cov, snr,
tmin=0.0, tmax=0.2, iir_filter=iir_filter)
###############################################################################
# Plot
plot_sparse_source_estimates(fwd['src'], stc, bgcolor=(1, 1, 1),
opacity=0.5, high_resolution=True)
plt.figure()
plt.psd(evoked.data[0])
plot_evoked(evoked)
def test_simulate_evoked():
""" Test simulation of evoked data """
raw = Raw(raw_fname)
fwd = read_forward_solution(fwd_fname, force_fixed=True)
fwd = pick_types_forward(fwd, meg=True, eeg=True, exclude=raw.info['bads'])
cov = read_cov(cov_fname)
label_names = ['Aud-lh', 'Aud-rh']
labels = [
read_label(
op.join(data_path, 'MEG', 'sample', 'labels', '%s.label' % label))
for label in label_names
]
evoked_template = read_evokeds(ave_fname, condition=0, baseline=None)
evoked_template.pick_types(meg=True, eeg=True, exclude=raw.info['bads'])
snr = 6 # dB
tmin = -0.1
sfreq = 1000. # Hz
tstep = 1. / sfreq
n_samples = 600
times = np.linspace(tmin, tmin + n_samples * tstep, n_samples)
# Generate times series from 2 Morlet wavelets
stc_data = np.zeros((len(labels), len(times)))
Ws = morlet(sfreq, [3, 10], n_cycles=[1, 1.5])
stc_data[0][:len(Ws[0])] = np.real(Ws[0])
stc_data[1][:len(Ws[1])] = np.real(Ws[1])
stc_data *= 100 * 1e-9 # use nAm as unit
# time translation
stc_data[1] = np.roll(stc_data[1], 80)
stc = generate_sparse_stc(fwd['src'],
labels,
stc_data,
tmin,
tstep,
random_state=0)
# Generate noisy evoked data
iir_filter = [1, -0.9]
with warnings.catch_warnings(record=True):
warnings.simplefilter('always') # positive semidefinite warning
evoked = generate_evoked(fwd,
stc,
evoked_template,
cov,
snr,
tmin=0.0,
tmax=0.2,
iir_filter=iir_filter)
assert_array_almost_equal(evoked.times, stc.times)
assert_true(len(evoked.data) == len(fwd['sol']['data']))
# make a vertex that doesn't exist in fwd, should throw error
stc_bad = stc.copy()
mv = np.max(fwd['src'][0]['vertno'][fwd['src'][0]['inuse']])
stc_bad.vertices[0][0] = mv + 1
assert_raises(RuntimeError,
generate_evoked,
fwd,
stc_bad,
evoked_template,
cov,
snr,
tmin=0.0,
tmax=0.2)
def test_dipole_fitting():
"""Test dipole fitting"""
amp = 10e-9
tempdir = _TempDir()
rng = np.random.RandomState(0)
fname_dtemp = op.join(tempdir, 'test.dip')
fname_sim = op.join(tempdir, 'test-ave.fif')
fwd = convert_forward_solution(read_forward_solution(fname_fwd),
surf_ori=False,
force_fixed=True)
evoked = read_evokeds(fname_evo)[0]
cov = read_cov(fname_cov)
n_per_hemi = 5
vertices = [
np.sort(rng.permutation(s['vertno'])[:n_per_hemi]) for s in fwd['src']
]
nv = sum(len(v) for v in vertices)
stc = SourceEstimate(amp * np.eye(nv), vertices, 0, 0.001)
with warnings.catch_warnings(record=True): # semi-def cov
evoked = generate_evoked(fwd,
stc,
evoked,
cov,
snr=20,
random_state=rng)
# For speed, let's use a subset of channels (strange but works)
picks = np.sort(
np.concatenate([
pick_types(evoked.info, meg=True, eeg=False)[::2],
pick_types(evoked.info, meg=False, eeg=True)[::2]
]))
evoked.pick_channels([evoked.ch_names[p] for p in picks])
evoked.add_proj(make_eeg_average_ref_proj(evoked.info))
write_evokeds(fname_sim, evoked)
# Run MNE-C version
run_subprocess([
'mne_dipole_fit',
'--meas',
fname_sim,
'--meg',
'--eeg',
'--noise',
fname_cov,
'--dip',
fname_dtemp,
'--mri',
fname_fwd,
'--reg',
'0',
'--tmin',
'0',
])
dip_c = read_dipole(fname_dtemp)
# Run mne-python version
sphere = make_sphere_model(head_radius=0.1)
dip, residuals = fit_dipole(evoked, fname_cov, sphere, fname_fwd)
# Sanity check: do our residuals have less power than orig data?
data_rms = np.sqrt(np.sum(evoked.data**2, axis=0))
resi_rms = np.sqrt(np.sum(residuals**2, axis=0))
assert_true((data_rms > resi_rms).all())
# Compare to original points
transform_surface_to(fwd['src'][0], 'head', fwd['mri_head_t'])
transform_surface_to(fwd['src'][1], 'head', fwd['mri_head_t'])
src_rr = np.concatenate([s['rr'][v] for s, v in zip(fwd['src'], vertices)],
axis=0)
src_nn = np.concatenate([s['nn'][v] for s, v in zip(fwd['src'], vertices)],
axis=0)
# MNE-C skips the last "time" point :(
dip.crop(dip_c.times[0], dip_c.times[-1])
src_rr, src_nn = src_rr[:-1], src_nn[:-1]
# check that we did at least as well
corrs, dists, gc_dists, amp_errs, gofs = [], [], [], [], []
for d in (dip_c, dip):
new = d.pos
diffs = new - src_rr
corrs += [np.corrcoef(src_rr.ravel(), new.ravel())[0, 1]]
dists += [np.sqrt(np.mean(np.sum(diffs * diffs, axis=1)))]
gc_dists += [
180 / np.pi * np.mean(np.arccos(np.sum(src_nn * d.ori, axis=1)))
]
amp_errs += [np.sqrt(np.mean((amp - d.amplitude)**2))]
gofs += [np.mean(d.gof)]
assert_true(dists[0] >= dists[1], 'dists: %s' % dists)
assert_true(corrs[0] <= corrs[1], 'corrs: %s' % corrs)
assert_true(gc_dists[0] >= gc_dists[1], 'gc-dists (ori): %s' % gc_dists)
assert_true(amp_errs[0] >= amp_errs[1], 'amplitude errors: %s' % amp_errs)
stc_data[1][:len(Ws[1])] = np.real(Ws[1])
stc_data[2][:len(Ws[2])] = np.real(Ws[2])
stc_data[3][:len(Ws[3])] = np.real(Ws[3])
stc_data += 100 * 1e-9 # use nAm as unit
# time translation
stc_data[1] = np.roll(stc_data[1], 80)
"""
stc = generate_sparse_stc(fwd["src"], labels, stc_data, tmin, tstep)
###############################################################################
# Generate noisy evoked data
picks = mne.fiff.pick_types(raw.info, meg=True)
iir_filter = iir_filter_raw(raw, order=5, picks=picks, tmin=60, tmax=180)
evoked = generate_evoked(fwd, stc, evoked_template, cov, snr) # , iir_filter=iir_filter)
raw[:306, :ntimes] = evoked.data
raw = raw.crop(tmax=evoked.times[-1])
raw.save("t2-raw.fif")
raw2 = mne.fiff.Raw("t2-raw.fif")
cov = mne.compute_raw_data_covariance(raw2)
dsim = stc.in_label(labels[0])
pl.plot(dsim.data.T)
weights = ve.calculate_weights(fwd, cov, reg=0)
d, _ = raw2[:306, :]
sol = np.dot(weights, d)
src = mne.SourceEstimate(sol, (fwd["src"][0]["vertno"], fwd["src"][1]["vertno"]), tmin, tstep)
stc_data[1][:len(Ws[1])] = np.real(Ws[1])
stc_data[2][:len(Ws[2])] = np.real(Ws[2])
stc_data[3][:len(Ws[3])] = np.real(Ws[3])
stc_data += 100 * 1e-9 # use nAm as unit
# time translation
stc_data[1] = np.roll(stc_data[1], 80)
'''
stc = generate_sparse_stc(fwd['src'], labels, stc_data, tmin, tstep)
###############################################################################
# Generate noisy evoked data
picks = mne.fiff.pick_types(raw.info, meg=True)
iir_filter = iir_filter_raw(raw, order=5, picks=picks, tmin=60, tmax=180)
evoked = generate_evoked(fwd, stc, evoked_template, cov, snr) #, iir_filter=iir_filter)
raw[:306,:ntimes] = evoked.data
raw = raw.crop(tmax=evoked.times[-1])
raw.save('t2-raw.fif')
raw2 = mne.fiff.Raw('t2-raw.fif')
cov = mne.compute_raw_data_covariance(raw2)
dsim = stc.in_label(labels[0])
pl.plot(dsim.data.T)
weights = ve.calculate_weights(fwd, cov, reg=0)
d, _ = raw2[:306,:]
sol = np.dot(weights, d)
src = mne.SourceEstimate(sol, (fwd['src'][0]['vertno'], fwd['src'][1]['vertno']), tmin, tstep)