agglo_stc.py

import os.path as op
import numpy as np

import mne
from mne import spatial_tris_connectivity, grade_to_tris
from mne.datasets import sample

from sklearn import cluster
from collections import Counter
from surfer import Brain, TimeViewer


###############################################################################
# Set parameters
data_path = sample.data_path()
stc_fname = data_path + '/MEG/sample/sample_audvis-meg'
subjects_dir = data_path + '/subjects'

# Load stc to in common cortical space (fsaverage)
stc = mne.read_source_estimate(stc_fname)
stc.resample(50)

stc = mne.morph_data('sample', 'fsaverage', stc, grade=5, smooth=20,
                     subjects_dir=subjects_dir,n_jobs=-1)
n_vertices_fsave, n_times = stc.data.shape
tstep = stc.tstep

connectivity = spatial_tris_connectivity(grade_to_tris(5))
model = cluster.FeatureAgglomeration(n_clusters=200,connectivity=connectivity
                                     ,linkage='average',affinity='cityblock')
fit = model.fit(stc.data.T)
counts = Counter(fit.labels_)

an_roi = np.zeros(fit.labels_.shape)
an_roi[fit.labels_==1] = 10
cluster_stc = mne.SourceEstimate(an_roi[:,None],vertices=stc.vertices,tmin=1,tstep=1)
# plot movies
brain = cluster_stc.plot(subject='fsaverage',hemi='split', colormap="jet",
              subjects_dir=subjects_dir,views=['lat', 'med'],
              config_opts=dict(width=2500, height=1400))