1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
|
import numpy as np
from nose.tools import assert_raises, assert_equal, assert_true
from numpy.testing import assert_array_equal
from os import path as op
import warnings
import mne
from mne.io import read_raw_fif
from mne.utils import sum_squared
from mne.time_frequency import csd_epochs, csd_array, tfr_morlet
warnings.simplefilter('always')
base_dir = op.join(op.dirname(__file__), '..', '..', 'io', 'tests', 'data')
raw_fname = op.join(base_dir, 'test_raw.fif')
event_fname = op.join(base_dir, 'test-eve.fif')
def _get_data(mode='real'):
"""Get data."""
raw = read_raw_fif(raw_fname, add_eeg_ref=False)
events = mne.read_events(event_fname)[0:100]
if mode == 'real':
# Read raw data
raw.info['bads'] = ['MEG 2443', 'EEG 053'] # 2 bads channels
# Set picks
picks = mne.pick_types(raw.info, meg=True, eeg=False, eog=False,
stim=False, exclude='bads')
# Read several epochs
event_id, tmin, tmax = 1, -0.2, 0.5
epochs = mne.Epochs(raw, events, event_id, tmin, tmax, proj=True,
picks=picks, baseline=(None, 0), preload=True,
reject=dict(grad=4000e-13, mag=4e-12),
add_eeg_ref=False)
elif mode == 'sin':
# Create an epochs object with one epoch and one channel of artificial
# data
event_id, tmin, tmax = 1, 0.0, 1.0
epochs = mne.Epochs(raw, events[0:5], event_id, tmin, tmax, proj=True,
picks=[0], baseline=(None, 0), preload=True,
reject=dict(grad=4000e-13), add_eeg_ref=False)
freq = 10
epochs._data = np.sin(2 * np.pi * freq *
epochs.times)[None, None, :]
return epochs
def test_csd_epochs():
"""Test computing cross-spectral density from epochs."""
epochs = _get_data(mode='real')
# Check that wrong parameters are recognized
assert_raises(ValueError, csd_epochs, epochs, mode='notamode')
assert_raises(ValueError, csd_epochs, epochs, fmin=20, fmax=10)
assert_raises(ValueError, csd_epochs, epochs, fmin=20, fmax=20.1)
assert_raises(ValueError, csd_epochs, epochs, tmin=0.15, tmax=0.1)
assert_raises(ValueError, csd_epochs, epochs, tmin=0, tmax=10)
assert_raises(ValueError, csd_epochs, epochs, tmin=10, tmax=11)
data_csd_mt = csd_epochs(epochs, mode='multitaper', fmin=8, fmax=12,
tmin=0.04, tmax=0.15)
data_csd_fourier = csd_epochs(epochs, mode='fourier', fmin=8, fmax=12,
tmin=0.04, tmax=0.15)
# Check shape of the CSD matrix
n_chan = len(data_csd_mt.ch_names)
assert_equal(data_csd_mt.data.shape, (n_chan, n_chan))
assert_equal(data_csd_fourier.data.shape, (n_chan, n_chan))
# Check if the CSD matrix is hermitian
assert_array_equal(np.tril(data_csd_mt.data).T.conj(),
np.triu(data_csd_mt.data))
assert_array_equal(np.tril(data_csd_fourier.data).T.conj(),
np.triu(data_csd_fourier.data))
# Computing induced power for comparison
epochs.crop(tmin=0.04, tmax=0.15)
tfr = tfr_morlet(epochs, freqs=[10], n_cycles=0.6, return_itc=False)
power = np.mean(tfr.data, 2)
# Maximum PSD should occur for specific channel
max_ch_power = power.argmax()
max_ch_mt = data_csd_mt.data.diagonal().argmax()
max_ch_fourier = data_csd_fourier.data.diagonal().argmax()
assert_equal(max_ch_mt, max_ch_power)
assert_equal(max_ch_fourier, max_ch_power)
# Maximum CSD should occur for specific channel
ch_csd_mt = np.abs(data_csd_mt.data[max_ch_power])
ch_csd_mt[max_ch_power] = 0.
max_ch_csd_mt = np.argmax(ch_csd_mt)
ch_csd_fourier = np.abs(data_csd_fourier.data[max_ch_power])
ch_csd_fourier[max_ch_power] = 0.
max_ch_csd_fourier = np.argmax(ch_csd_fourier)
assert_equal(max_ch_csd_mt, max_ch_csd_fourier)
# Check a list of CSD matrices is returned for multiple frequencies within
# a given range when fsum=False
csd_fsum = csd_epochs(epochs, mode='fourier', fmin=8, fmax=20, fsum=True)
csds = csd_epochs(epochs, mode='fourier', fmin=8, fmax=20, fsum=False)
freqs = [csd.frequencies[0] for csd in csds]
csd_sum = np.zeros_like(csd_fsum.data)
for csd in csds:
csd_sum += csd.data
assert_equal(len(csds), 2)
assert_equal(len(csd_fsum.frequencies), 2)
assert_array_equal(csd_fsum.frequencies, freqs)
assert_array_equal(csd_fsum.data, csd_sum)
def test_csd_epochs_on_artificial_data():
"""Test computing CSD on artificial data."""
epochs = _get_data(mode='sin')
sfreq = epochs.info['sfreq']
# Computing signal power in the time domain
signal_power = sum_squared(epochs._data)
signal_power_per_sample = signal_power / len(epochs.times)
# Computing signal power in the frequency domain
data_csd_fourier = csd_epochs(epochs, mode='fourier')
data_csd_mt = csd_epochs(epochs, mode='multitaper')
fourier_power = np.abs(data_csd_fourier.data[0, 0]) * sfreq
mt_power = np.abs(data_csd_mt.data[0, 0]) * sfreq
assert_true(abs(fourier_power - signal_power) <= 0.5)
assert_true(abs(mt_power - signal_power) <= 1)
# Power per sample should not depend on time window length
for tmax in [0.2, 0.8]:
for add_n_fft in [0, 30]:
t_mask = (epochs.times >= 0) & (epochs.times <= tmax)
n_samples = sum(t_mask)
n_fft = n_samples + add_n_fft
data_csd_fourier = csd_epochs(epochs, mode='fourier',
tmin=None, tmax=tmax, fmin=0,
fmax=np.inf, n_fft=n_fft)
first_samp = data_csd_fourier.data[0, 0]
fourier_power_per_sample = np.abs(first_samp) * sfreq / n_fft
assert_true(abs(signal_power_per_sample -
fourier_power_per_sample) < 0.003)
# Power per sample should not depend on number of tapers
for n_tapers in [1, 2, 5]:
for add_n_fft in [0, 30]:
mt_bandwidth = sfreq / float(n_samples) * (n_tapers + 1)
data_csd_mt = csd_epochs(epochs, mode='multitaper',
tmin=None, tmax=tmax, fmin=0,
fmax=np.inf,
mt_bandwidth=mt_bandwidth,
n_fft=n_fft)
mt_power_per_sample = np.abs(data_csd_mt.data[0, 0]) *\
sfreq / data_csd_mt.n_fft
# The estimate of power gets worse for small time windows when
# more tapers are used
if n_tapers == 5 and tmax == 0.2:
delta = 0.05
else:
delta = 0.004
assert_true(abs(signal_power_per_sample -
mt_power_per_sample) < delta)
def test_compute_csd():
"""Test computing cross-spectral density from ndarray."""
epochs = _get_data(mode='real')
tmin = 0.04
tmax = 0.15
tmp = np.where(np.logical_and(epochs.times >= tmin,
epochs.times <= tmax))[0]
picks_meeg = mne.pick_types(epochs[0].info, meg=True, eeg=True, eog=False,
ref_meg=False, exclude='bads')
epochs_data = [e[picks_meeg][:, tmp].copy() for e in epochs]
n_trials = len(epochs)
n_series = len(picks_meeg)
X = np.concatenate(epochs_data, axis=0)
X = np.reshape(X, (n_trials, n_series, -1))
X_list = epochs_data
sfreq = epochs.info['sfreq']
# Check data types and sizes are checked
diff_types = [np.random.randn(3, 5), "error"]
err_data = [np.random.randn(3, 5), np.random.randn(2, 4)]
assert_raises(ValueError, csd_array, err_data, sfreq)
assert_raises(ValueError, csd_array, diff_types, sfreq)
assert_raises(ValueError, csd_array, np.random.randn(3), sfreq)
# Check that wrong parameters are recognized
assert_raises(ValueError, csd_array, X, sfreq, mode='notamode')
assert_raises(ValueError, csd_array, X, sfreq, fmin=20, fmax=10)
assert_raises(ValueError, csd_array, X, sfreq, fmin=20, fmax=20.1)
data_csd_mt, freqs_mt = csd_array(X, sfreq, mode='multitaper',
fmin=8, fmax=12)
data_csd_fourier, freqs_fft = csd_array(X, sfreq, mode='fourier',
fmin=8, fmax=12)
# Test as list too
data_csd_mt_list, freqs_mt_list = csd_array(X_list, sfreq,
mode='multitaper',
fmin=8, fmax=12)
data_csd_fourier_list, freqs_fft_list = csd_array(X_list, sfreq,
mode='fourier',
fmin=8, fmax=12)
assert_array_equal(data_csd_mt, data_csd_mt_list)
assert_array_equal(data_csd_fourier, data_csd_fourier_list)
assert_array_equal(freqs_mt, freqs_mt_list)
assert_array_equal(freqs_fft, freqs_fft_list)
# Check shape of the CSD matrix
n_chan = len(epochs.ch_names)
assert_equal(data_csd_mt.shape, (n_chan, n_chan))
assert_equal(data_csd_fourier.shape, (n_chan, n_chan))
# Check if the CSD matrix is hermitian
assert_array_equal(np.tril(data_csd_mt).T.conj(),
np.triu(data_csd_mt))
assert_array_equal(np.tril(data_csd_fourier).T.conj(),
np.triu(data_csd_fourier))
# Computing induced power for comparison
epochs.crop(tmin=0.04, tmax=0.15)
tfr = tfr_morlet(epochs, freqs=[10], n_cycles=0.6, return_itc=False)
power = np.mean(tfr.data, 2)
# Maximum PSD should occur for specific channel
max_ch_power = power.argmax()
max_ch_mt = data_csd_mt.diagonal().argmax()
max_ch_fourier = data_csd_fourier.diagonal().argmax()
assert_equal(max_ch_mt, max_ch_power)
assert_equal(max_ch_fourier, max_ch_power)
# Maximum CSD should occur for specific channel
ch_csd_mt = np.abs(data_csd_mt[max_ch_power])
ch_csd_mt[max_ch_power] = 0.
max_ch_csd_mt = np.argmax(ch_csd_mt)
ch_csd_fourier = np.abs(data_csd_fourier[max_ch_power])
ch_csd_fourier[max_ch_power] = 0.
max_ch_csd_fourier = np.argmax(ch_csd_fourier)
assert_equal(max_ch_csd_mt, max_ch_csd_fourier)
# Check a list of CSD matrices is returned for multiple frequencies within
# a given range when fsum=False
csd_fsum, freqs_fsum = csd_array(X, sfreq, mode='fourier', fmin=8,
fmax=20, fsum=True)
csds, freqs = csd_array(X, sfreq, mode='fourier', fmin=8, fmax=20,
fsum=False)
csd_sum = np.sum(csds, axis=2)
assert_equal(csds.shape[2], 2)
assert_equal(len(freqs), 2)
assert_array_equal(freqs_fsum, freqs)
assert_array_equal(csd_fsum, csd_sum)
def test_csd_on_artificial_data():
"""Test computing CSD on artificial data. """
epochs = _get_data(mode='sin')
sfreq = epochs.info['sfreq']
# Computing signal power in the time domain
signal_power = sum_squared(epochs._data)
signal_power_per_sample = signal_power / len(epochs.times)
# Computing signal power in the frequency domain
data_csd_mt, freqs_mt = csd_array(epochs._data, sfreq,
mode='multitaper')
data_csd_fourier, freqs_fft = csd_array(epochs._data, sfreq,
mode='fourier')
fourier_power = np.abs(data_csd_fourier[0, 0]) * sfreq
mt_power = np.abs(data_csd_mt[0, 0]) * sfreq
assert_true(abs(fourier_power - signal_power) <= 0.5)
assert_true(abs(mt_power - signal_power) <= 1)
# Power per sample should not depend on time window length
for tmax in [0.2, 0.8]:
tslice = np.where(epochs.times <= tmax)[0]
for add_n_fft in [0, 30]:
t_mask = (epochs.times >= 0) & (epochs.times <= tmax)
n_samples = sum(t_mask)
n_fft = n_samples + add_n_fft
data_csd_fourier, _ = csd_array(epochs._data[:, :, tslice],
sfreq, mode='fourier',
fmin=0, fmax=np.inf, n_fft=n_fft)
first_samp = data_csd_fourier[0, 0]
fourier_power_per_sample = np.abs(first_samp) * sfreq / n_fft
assert_true(abs(signal_power_per_sample -
fourier_power_per_sample) < 0.003)
# Power per sample should not depend on number of tapers
for n_tapers in [1, 2, 5]:
for add_n_fft in [0, 30]:
mt_bandwidth = sfreq / float(n_samples) * (n_tapers + 1)
data_csd_mt, _ = csd_array(epochs._data[:, :, tslice],
sfreq, mt_bandwidth=mt_bandwidth,
n_fft=n_fft)
mt_power_per_sample = np.abs(data_csd_mt[0, 0]) *\
sfreq / n_fft
# The estimate of power gets worse for small time windows when
# more tapers are used
if n_tapers == 5 and tmax == 0.2:
delta = 0.05
else:
delta = 0.004
assert_true(abs(signal_power_per_sample -
mt_power_per_sample) < delta)
|
