def run(cfg, state=mp.Value('i', 1), queue=None):
"""
Training protocol for Alpha/Theta neurofeedback.
"""
redirect_stdout_to_queue(logger, queue, 'INFO')
# add tdef object
cfg.tdef = trigger_def(cfg.TRIGGER_FILE)
# Extract features
if not state.value:
sys.exit(-1)
raw = mne.io.read_raw_brainvision(cfg.DATA_PATH, preload=True)
outfile = cfg.OUT_MICROSTATES_FILE
ch_names = [
'P3', 'C3', 'F3', 'Fz', 'F4', 'C4', 'P4', 'Cz', 'Pz', 'Fp1', 'Fp2',
'T3', 'T5', 'O1', 'O2', 'F7', 'F8', 'T6', 'T4'
]
raw.pick_channels(ch_names)
raw.set_montage('standard_1005')
raw.set_eeg_reference('average')
raw.filter(1, 30)
maps, segmentation = microstates.segment(raw.get_data(),
n_states=4,
max_n_peaks=10000000,
max_iter=5000,
normalize=True)
np.savetxt(outfile, maps, delimiter=" ")
from mne.datasets import sample
fname = sample.data_path() + '/MEG/sample/sample_audvis_filt-0-40_raw.fif'
raw = mne.io.read_raw_fif(fname, preload=True)
events = mne.find_events(raw)
raw.info['bads'] = ['MEG 2443', 'EEG 053']
raw.set_eeg_reference('average')
raw.pick_types(meg='mag', eeg=True, eog=True, ecg=True)
raw.filter(1, 40)
# Clean EOG with ICA
ica = mne.preprocessing.ICA(0.99).fit(raw)
bads_eog, _ = ica.find_bads_eog(raw)
bads_ecg, _ = ica.find_bads_ecg(raw)
ica.exclude = bads_eog[:2] + bads_ecg[:2]
raw = ica.apply(raw)
# Select sensor type
# raw.pick_types(meg=False, eeg=True)
raw.pick_types(meg='mag', eeg=False)
# Segment the data in 6 microstates
maps, segmentation, _, _ = microstates.segment(raw.get_data(),
n_states=6,
random_state=0)
# Plot the topographic maps of the microstates and the segmentation
microstates.viz.plot_maps(maps, raw.info)
microstates.viz.plot_segmentation(segmentation[:500],
raw.get_data()[:, :500], raw.times[:500])
raw.set_eeg_reference('average')
#raw.plot_psd()
plt.show()
# Highpass filter the data a little bit
raw.filter(0.2, None)
# Selecting the sensor types to use in the analysis. In this example, we
# use only EEG channels
raw.pick_types(meg=False, eeg=True)
# Segment the data into 6 microstates
maps, segmentation = microstates.segment(raw.get_data(),
n_states=nstates,
max_n_peaks=10000000000,
max_iter=5000,
normalize=normalize)
# Plot the topographic maps of the found microstates
#microstates.plot_maps(maps, raw.info)
myplot_maps(maps, raw.info)
# Plot the segmentation of the first 500 samples
cutpoint = 5000
microstates.plot_segmentation(segmentation[cutpoint - 1000:cutpoint],
raw.get_data()[:, cutpoint - 1000:cutpoint],
raw.times[cutpoint - 1000:cutpoint])
plt.show()
np.savetxt(outfile, maps, delimiter=" ")
n_states = 4
n_inits = 10
EGI256 = True
if EGI256 == True:
n_channels = 256 - len(chan_to_drop)
grouped_maps = np.array([], dtype=np.int64).reshape(0, n_channels)
for i, f in enumerate(subj_folder):
fname = HC_RS_path + f + '/' + f +'_clean-epo.fif'
epochs = mne.read_epochs(fname, preload=True)
if EGI256 == True:
epochs.drop_channels(chan_to_drop)
data = epochs.get_data()
# Segment the data in microstates
maps, segmentation, gev, gfp_peaks = mst.segment(data, n_states, n_inits)
grouped_maps = np.concatenate((grouped_maps, maps), axis=0)
# Transpose the maps from maps(n_maps, n_channels) to maps(n_channels, n_maps)
# and treat the n_maps as a sample in time.
grouped_maps_T = grouped_maps.transpose()
# Find the group maps using k-means clustering
group_maps, group_gev = mst.segment(grouped_maps_T, n_states, n_inits, use_peaks=False)
# Plot the maps
mst.viz.plot_maps(group_maps, epochs.info)
# Fitting the maps back to the original epoched data by subject
grouped_segment, all_p = [], []
for i, f in enumerate(subj_folder):
if EGI256 == True:
epochs.drop_channels([
'E67', 'E73', 'E247', 'E251', 'E256', 'E243', 'E246', 'E250', 'E255',
'E82', 'E91', 'E254', 'E249', 'E245', 'E242', 'E253', 'E252', 'E248',
'E244', 'E241', 'E92', 'E102', 'E103', 'E111', 'E112', 'E120', 'E121',
'E133', 'E134', 'E145', 'E146', 'E156', 'E165', 'E166', 'E174', 'E175',
'E187', 'E188', 'E199', 'E200', 'E208', 'E209', 'E216', 'E217', 'E228',
'E229', 'E232', 'E233', 'E236', 'E237', 'E240', 'E218', 'E227', 'E231',
'E235', 'E239', 'E219', 'E225', 'E226', 'E230', 'E234', 'E238'
])
n_epochs, n_chans, n_samples = epochs.get_data().shape
# Segment the data in microstates
maps, segmentation, gev, gfp_peaks = mst.segment(epochs.get_data(),
n_states,
n_inits,
normalize=True,
min_peak_dist=10,
max_n_peaks=10000)
# Smoothen the segmentation
segmentation_smooth = mst.seg_smoothing(data=epochs.get_data(), maps=maps)
# Mark each epoch at a beginning and at an end of an epoch w/ the value 88
seg_w_borders = mst.mark_border_msts(segmentation, n_epochs, n_samples,
n_states)
# Remove the values 88 of the segmentation
seg_wo_borders = seg_w_borders[seg_w_borders != 88]
# Plot the topographic maps of the microstates and the segmentation
mst.viz.plot_maps(maps, epochs.info)
# plot the whole segmentation