from matplotlib.lines import Line2D
def plot_lines(mask, linewidth=3, color='b'):
for i, j in np.ndindex(mask.shape):
if i + 1 < mask.shape[0] and mask[i, j] != mask[i + 1, j]:
pl.gca().add_line(Line2D([j - .5, j + .5], [i + .5, i + .5],
color=color, linewidth=linewidth))
if j + 1 < mask.shape[1] and mask[i, j] != mask[i, j + 1]:
pl.gca().add_line(Line2D([j + .5, j + .5], [i - .5, i + .5],
color=color, linewidth=linewidth))
sbrain = masking.unmask(np.array(scores).mean(0), dataset.mask)
bg = nibabel.load(os.path.join('utils', 'bg.nii.gz'))
pl.figure(figsize=(8, 8))
ax1 = pl.axes([0., 0., 1., 1.])
pl.imshow(bg.get_data()[:, :, 10].T, interpolation="nearest", cmap='gray',
origin='lower')
pl.imshow(np.ma.masked_less(sbrain[:, :, 10].T, 1e-6),
interpolation="nearest", cmap='hot', origin="lower")
plot_lines(contour[:, :, 10].T)
pl.axis('off')
ax2 = pl.axes([.08, .5, .05, .47])
cb = pl.colorbar(cax=ax2, ax=ax1)
cb.ax.yaxis.set_ticks_position('left')
cb.ax.yaxis.set_tick_params(labelcolor='white')
def plot_lines(mask, linewidth=3, color='b'):
for i, j in np.ndindex(mask.shape):
if i + 1 < mask.shape[0] and mask[i, j] != mask[i + 1, j]:
pl.gca().add_line(
Line2D([j - .5, j + .5], [i + .5, i + .5],
color=color,
linewidth=linewidth))
if j + 1 < mask.shape[1] and mask[i, j] != mask[i, j + 1]:
pl.gca().add_line(
Line2D([j + .5, j + .5], [i - .5, i + .5],
color=color,
linewidth=linewidth))
sbrain = masking.unmask(np.array(scores).mean(0), dataset.mask)
bg = nibabel.load(os.path.join('utils', 'bg.nii.gz'))
pl.figure(figsize=(8, 8))
ax1 = pl.axes([0., 0., 1., 1.])
pl.imshow(bg.get_data()[:, :, 10].T,
interpolation="nearest",
cmap='gray',
origin='lower')
pl.imshow(np.ma.masked_less(sbrain[:, :, 10].T, 1e-6),
interpolation="nearest",
cmap='hot',
origin="lower")
plot_lines(contour[:, :, 10].T)
pl.axis('off')
X = X_
### CanICA ####################################################################
if not exists(join(path, 'canica.nii.gz')):
try:
from nilearn.decomposition.canica import CanICA
t0 = time.time()
canica = CanICA(n_components=n_components, mask=mask_img,
smoothing_fwhm=6.,
memory="nilearn_cache", memory_level=1,
threshold=None,
random_state=1, n_jobs=-1)
canica.fit(dataset.func)
print('Canica: %f' % (time.time() - t0))
canica_components = masking.unmask(canica.components_, mask_img)
nibabel.save(nibabel.Nifti1Image(canica_components,
mask_img.get_affine()), join(path, 'canica.nii.gz'))
except ImportError:
import warnings
warnings.warn('nilearn must be installed to run CanICA')
canica_dmn = nibabel.load(join(path, 'canica.nii.gz')).get_data()[..., 4]
### Melodic ICA ############################################################
# To have MELODIC results, please use my melodic branch of nilearn
melodic_dmn = nibabel.load(join(path, 'melodic.nii.gz')).get_data()[..., 3]
for i, j in np.ndindex(mask.shape):
if i + 1 < mask.shape[0] and mask[i, j] != mask[i + 1, j]:
pl.gca().add_line(
Line2D([j - .5, j + .5], [i + .5, i + .5],
color=color,
linewidth=linewidth))
if j + 1 < mask.shape[1] and mask[i, j] != mask[i, j + 1]:
pl.gca().add_line(
Line2D([j + .5, j + .5], [i - .5, i + .5],
color=color,
linewidth=linewidth))
fig = pl.figure(figsize=(8, 8))
ax1 = pl.axes([0., 0., 1., 1.])
sbrain = masking.unmask(lr_coef[0], dataset.mask)
bg = nibabel.load(os.path.join('utils', 'bg.nii.gz'))
pl.imshow(bg.get_data()[:, :, 10].T,
interpolation="nearest",
cmap='gray',
origin='lower')
pl.imshow(np.ma.masked_equal(sbrain[:, :, 10].T, 0.),
interpolation="nearest",
cmap=cm.bluegreen,
origin='lower',
vmin=0.,
vmax=2.6)
plot_lines(contour[:, :, 10].T, color='r')
pl.axis('off')
ax2 = pl.axes([.1, .5, .05, .45])
cb = pl.colorbar(cax=ax2, ax=ax1)
connectivity = image.grid_to_graph(n_x=shape[0],
n_y=shape[1],
n_z=shape[2],
mask=mask)
for n_clusters in 100, 1000:
# Compute Ward clustering
from sklearn.cluster import WardAgglomeration
ward = WardAgglomeration(n_clusters=n_clusters,
connectivity=connectivity,
memory='nilearn_cache',
compute_full_tree=True)
ward.fit(X)
labels = ward.labels_ + 1
labels = masking.unmask(labels, adhd_mask)
# 0 is the background, putting it to -1
labels = labels - 1
# Display the labels
plot_labels(labels, 8)
pl.savefig(join('clustering', 'ward_%i.eps' % n_clusters))
pl.savefig(join('clustering', 'ward_%i.pdf' % n_clusters))
# Compute Kmeans clustering
from sklearn.cluster import MiniBatchKMeans
kmeans = MiniBatchKMeans(n_clusters=n_clusters, random_state=1)
kmeans.fit(X_smoothed.T)
labels = kmeans.labels_ + 1
# Compute connectivity matrix: which voxel is connected to which
from sklearn.feature_extraction import image
shape = mask.shape
connectivity = image.grid_to_graph(n_x=shape[0], n_y=shape[1],
n_z=shape[2], mask=mask)
for n_clusters in 100, 1000:
# Compute Ward clustering
from sklearn.cluster import WardAgglomeration
ward = WardAgglomeration(n_clusters=n_clusters, connectivity=connectivity,
memory='nilearn_cache', compute_full_tree=True)
ward.fit(X)
labels = ward.labels_ + 1
labels = masking.unmask(labels, adhd_mask)
# 0 is the background, putting it to -1
labels = labels - 1
# Display the labels
plot_labels(labels, 8)
pl.savefig(join('clustering', 'ward_%i.eps' % n_clusters))
pl.savefig(join('clustering', 'ward_%i.pdf' % n_clusters))
# Compute Kmeans clustering
from sklearn.cluster import MiniBatchKMeans
kmeans = MiniBatchKMeans(n_clusters=n_clusters, random_state=1)
kmeans.fit(X_smoothed.T)
labels = kmeans.labels_ + 1
pixmask[p] = 1
def plot_lines(mask, linewidth=3, color='b'):
for i, j in np.ndindex(mask.shape):
if i + 1 < mask.shape[0] and mask[i, j] != mask[i + 1, j]:
pl.gca().add_line(Line2D([j - .5, j + .5], [i + .5, i + .5],
color=color, linewidth=linewidth))
if j + 1 < mask.shape[1] and mask[i, j] != mask[i, j + 1]:
pl.gca().add_line(Line2D([j + .5, j + .5], [i - .5, i + .5],
color=color, linewidth=linewidth))
fig = pl.figure(figsize=(8, 8))
ax1 = pl.axes([0., 0., 1., 1.])
sbrain = masking.unmask(lr_coef[0], dataset.mask)
bg = nibabel.load(os.path.join('utils', 'bg.nii.gz'))
pl.imshow(bg.get_data()[:, :, 10].T, interpolation="nearest", cmap='gray',
origin='lower')
pl.imshow(np.ma.masked_equal(sbrain[:, :, 10].T, 0.), interpolation="nearest",
cmap=bluegreen, origin='lower', vmin=0., vmax=2.6)
plot_lines(contour[:, :, 10].T, color='r')
pl.axis('off')
ax2 = pl.axes([.1, .5, .05, .45])
cb = pl.colorbar(cax=ax2, ax=ax1)
cb.ax.yaxis.set_ticks_position('left')
cb.ax.yaxis.set_tick_params(labelcolor='white')
cb.ax.yaxis.set_tick_params(labelsize=32)
cb.set_ticks([0., 1.3, 2.6])
pl.savefig(os.path.join('miyawaki', 'pixel_logistic.pdf'))
pl.savefig(os.path.join('miyawaki', 'pixel_logistic.png'))
if not exists(join(path, 'canica.nii.gz')):
try:
from nilearn.decomposition.canica import CanICA
t0 = time.time()
canica = CanICA(n_components=n_components,
mask=mask_img,
smoothing_fwhm=6.,
memory="nilearn_cache",
memory_level=1,
threshold=None,
random_state=1,
n_jobs=-1)
canica.fit(dataset.func)
print('Canica: %f' % (time.time() - t0))
canica_components = masking.unmask(canica.components_, mask_img)
nibabel.save(
nibabel.Nifti1Image(canica_components, mask_img.get_affine()),
join(path, 'canica.nii.gz'))
except ImportError:
import warnings
warnings.warn('nilearn must be installed to run CanICA')
canica_dmn = nibabel.load(join(path, 'canica.nii.gz')).get_data()[..., 4]
### Melodic ICA ############################################################
# To have MELODIC results, please use my melodic branch of nilearn
melodic_dmn = nibabel.load(join(path, 'melodic.nii.gz')).get_data()[..., 3]
### FastICA ##################################################################
X_img = Nifti1Image(X, affine)
X = masking.apply_mask(X_img, mask, smoothing_fwhm=4)
X = signal.clean(X, standardize=True, detrend=False)
###############################################################################
# #
# F-score #
# #
###############################################################################
from sklearn.feature_selection import f_classif
f_values, p_values = f_classif(X, y)
p_values = -np.log10(p_values)
p_values[np.isnan(p_values)] = 0
p_values[p_values > 10] = 10
p_unmasked = masking.unmask(p_values, mask)
plot_haxby(p_unmasked, 'F-score')
pl.savefig('haxby/haxby_fscore.pdf')
pl.savefig('haxby/haxby_fscore.eps')
###############################################################################
# #
# SVC #
# #
###############################################################################
### Define the estimator
from sklearn.svm import SVC
clf = SVC(kernel='linear', C=0.01)
### Dimension reduction #######################################################
