"""

Computation of covariance matrix between brain regions

======================================================

The following things are performed:

- parcellation loading, and signals extraction

- improvement of fMRI data SNR (confound removal, filtering, etc.)

- covariance/precision matrices computation

- display of matrices

"""

# Running this script may require to set the dirname parameter in the

# load_harvard_oxford() function below

import numpy as np

import pylab as pl

import matplotlib

from sklearn import covariance

import nisl.datasets

import nisl.image

import nisl.io

# Copied from matplotlib 1.2.0 for matplotlib 0.99

_bwr_data = ((0.0, 0.0, 1.0), (1.0, 1.0, 1.0), (1.0, 0.0, 0.0))

pl.cm.register_cmap(cmap=matplotlib.colors.LinearSegmentedColormap.from_list(

"bwr", _bwr_data))

def plot_matrices(cov, prec, title, subject_n=0):

pl.title(title + " / precision")

subject_n = 1

dataset = nisl.datasets.fetch_adhd()

filename = dataset["func"][subject_n]

confound_file = dataset["confounds"][subject_n]

print("-- Loading raw data ({0:d}) and masking ...".format(subject_n))

labels_img = nisl.datasets.load_harvard_oxford(

"cort-maxprob-thr25-2mm", symmetric_split=True)

print("-- Computing confounds ...")

hv_confounds = nisl.image.high_variance_confounds(filename)

mvt_confounds = np.loadtxt(confound_file, skiprows=1)

confounds = np.hstack((hv_confounds, mvt_confounds))

print("-- Computing region signals ...")

nifti_regions = nisl.io.NiftiLabelsMasker(labels_img=labels_img,

)

region_ts = nifti_regions.fit_transform(filename, confounds=confounds)

print("-- Computing covariance matrices ...")

estimator = covariance.GraphLassoCV()

estimator.fit(region_ts)

plot_matrices(estimator.covariance_, -estimator.precision_,

title="Graph Lasso CV ({0:.3f})".format(estimator.alpha_),

subject_n=subject_n)

pl.show()