示例#1
0
def main():

    mne.utils.set_log_level('ERROR')

    layout = load_layout(MEG)

    print "Reading and processing data from files.."

    filenames = sys.argv[1:]

    new_components = []

    for idx, fname in enumerate(filenames):
        print "Handling " + str(idx+1) + ". subject"
        raw = mne.io.read_raw_fif(fname, preload=True)

        picks = mne.pick_types(raw.info, eeg=True, meg=True)
        raw.drop_channels([ch_name for ix, ch_name in 
            enumerate(raw.info['ch_names']) if ix not in picks])
        raw.drop_channels(raw.info['bads'])

        fica = FourierICA(wsize=WSIZE, n_components=COMPONENTS, maxiter=7000,
                          sfreq=raw.info['sfreq'], hpass=BAND[0], lpass=BAND[1])
        fica.fit(raw._data)

        components = get_components(fica, raw.info, len(raw.times), layout)

        handle = plot_components(components, layout)
        input_ = int(raw_input("Which component to use: "))
        if input_ == -1:
            continue
        new_components.append(components[input_ - 1])

    import pdb; pdb.set_trace()
    print "kissa"
示例#2
0
def main():

    mne.utils.set_log_level('ERROR')

    if MEG:
        layout = None
    else:
        layout = load_layout()

    filenames = sys.argv[1:]

    components = []
    for fname in filenames:
        print "Opening " + fname
        part = pickle.load(open(fname, "rb"))
        components.extend(part)

    plot_components(components, layout)

    import pdb; pdb.set_trace()
    print "kissa"
def main():

    mne.utils.set_log_level('ERROR')

    layout = load_layout(MEG)

    filenames = sys.argv[1:]

    components = []
    for fname in filenames:
        print "Opening " + fname
        part = pickle.load(open(fname, "rb"))
        components.extend(part)

    tses = []
    for component in components:
        tses.append(_get_tse(component))

    epochs = []
    for i in range(len(components)):
        component_epochs = _get_epochs(components[i], tses[i])
        # normalize
        for j, epoch in enumerate(component_epochs):
            component_epochs[j] = epoch / np.mean(tses[i])

        epochs.extend(component_epochs)

    print "Total: " + str(len(epochs)) + " epochs."

    average = np.mean(epochs, axis=0)

    # smoothen?
    # average = np.convolve(average, np.ones((3,))/3, mode='valid')

    step = (LIMITS[0]+LIMITS[1]) / float(len(average))
    times = np.array(range(len(average)))
    times = times * step - LIMITS[0] + step/2

    plt.plot(times, average)
    plt.show()
    # crop away eyes closed resting
    raws[0].crop(tmin=0, tmax=90)

# drop bad and non-data channels
for raw in raws:
    picks = mne.pick_types(raw.info, eeg=True, meg=True)
    raw.drop_channels([name for idx, name in enumerate(raw.info['ch_names'])
                       if idx not in picks])
    raw.drop_channels(raw.info['bads'])

raw = mne.concatenate_raws(raws)

if MEG:
    layout = None
else:
    layout = load_layout()

wsize = 4096
sfreq = raw.info['sfreq']
states = [(0, 85), (95, len(raw.times)/sfreq)]

# calculate fourier-ica
fica = FourierICA(wsize=wsize, n_components=N_COMPONENTS,
                  sfreq=sfreq, hpass=4, lpass=30,
                  maxiter=7000)
fica.fit(raw._data[:, raw.first_samp:raw.last_samp])

source_stft = fica.source_stft
freqs = fica.freqs

# plot components in sensor space on head topographies
    mne.io.Raw(sys.argv[-1], preload=True, add_eeg_ref=False),
]

for raw in raws:
    raw.add_proj([], remove_existing=True)

# drop bad and non-data channels
for raw in raws:
    picks = mne.pick_types(raw.info, eeg=True, meg=True)
    raw.drop_channels([name for idx, name in enumerate(raw.info['ch_names'])
                       if idx not in picks])
    raw.drop_channels(raw.info['bads'])

raw = mne.concatenate_raws(raws)

layout = load_layout(MEG=True)

wsize = 4096
sfreq = raw.info['sfreq']

states = []

events = mne.find_events(raw)

# calculate fourier-ica
fica = FourierICA(wsize=wsize, n_components=N_COMPONENTS,
                  sfreq=sfreq, hpass=4, lpass=30,
                  maxiter=7000)
fica.fit(raw._data)

source_stft = fica.source_stft