def runica(x, fs, channels, mode='ica'):
from PyQt5.QtWidgets import QApplication
from pynfb.protocols.ssd.topomap_selector_ica import ICADialog
a = QApplication([])
ica = ICADialog(x, channels, fs, mode=mode)
ica.exec_()
a.exit()
return ica.spatial, ica.topography
def run_ica_and_select_band(data, channels, fs):
app = QApplication([])
ica = ICADialog(data, channels, fs)
ica.exec_()
band = select_band(data.dot(ica.spatial), fs)
results = {
'spatial_filter': ica.spatial,
'spatial_filter_ind': ica.table.get_checked_rows()[0],
'filters': ica.decomposition.filters,
'topographies': ica.decomposition.topographies,
'band': band
}
return results
def runica2(x, fs, channels, names=('Right', 'Left'), mode='ica'):
from PyQt5.QtWidgets import QApplication
from pynfb.protocols.ssd.topomap_selector_ica import ICADialog
a = QApplication([])
res = []
decomposition = None
for n in names:
print('*** Select component for condition: ' + n)
ica = ICADialog(x, channels, fs, decomposition=decomposition, mode=mode)
ica.exec_()
res.append(np.array((ica.spatial, ica.topography)))
decomposition = ica.decomposition
a.exit()
return res
dir_ = 'D:\\vnd_spbu\\pilot\\mu5days'
experiment = 'pilot5days_Rakhmankulov_Day3_03-01_12-51-41'
with h5py.File('{}\\{}\\{}'.format(dir_, experiment,
'experiment_data.h5')) as f:
ica = f['protocol1/signals_stats/left/rejections/rejection1'][:]
x_filters = dc_blocker(np.dot(f['protocol1/raw_data'][:], ica))
x_rotation = dc_blocker(np.dot(f['protocol2/raw_data'][:], ica))
x_dict = {
'closed': x_filters[:x_filters.shape[0] // 2],
'opened': x_filters[x_filters.shape[0] // 2:],
'rotate': x_rotation
}
drop_channels = ['AUX', 'A1', 'A2']
labels, fs = get_lsl_info_from_xml(f['stream_info.xml'][0])
print('fs: {}\nall labels {}: {}'.format(fs, len(labels), labels))
channels = [label for label in labels if label not in drop_channels]
scores, unmixing_matrix, topographies = csp3(x_dict,
250, (9, 14),
lambda_=0.5,
regularization_coef=0.01)
app = QtWidgets.QApplication([])
selector = ICADialog(np.vstack((x_filters, x_rotation)),
channels,
fs,
unmixing_matrix=unmixing_matrix,
mode='csp',
scores=scores)
selector.exec_()
return df.loc[good_samples_mask].values, good_samples_mask
N_EEG_CHANNELS = 30
raw = mne.io.read_raw_fif('/media/kolai/prispasobo/Kondrashin_raw.fif')
channels = raw.info['ch_names'][:N_EEG_CHANNELS]
fs = int(raw.info['sfreq'])
data = raw.load_data().notch_filter([50, 100]).filter(
0.5, 40.).get_data()[:N_EEG_CHANNELS].T
# plt.plot(data[:fs*100])
ica_data, good_samples_mask = remove_bad_samples(data[:fs * 60 * 10],
window_size=fs * 5,
threshold=2000)
ica = ICADialog(ica_data, channels, fs)
ica.exec_()
# ica.table.get_checked_rows()
plt.plot(data[:, :] + np.arange(30) * 2000, 'r')
plt.plot(ica.rejection.apply(data)[:, :] + np.arange(30) * 2000, 'k')
plt.fill_between(np.arange(data.shape[0]),
(~good_samples_mask).astype(int) * 30 * 2000,
color='r',
alpha=0.4)
plt.gca().set_yticks(np.arange(30) * 2000)
plt.gca().set_yticklabels(channels)
plt.xlim(0, fs * 30)
