In [ ]:
%matplotlib inline

NeuroVault cross-study ICA maps.

This example shows how to download statistical maps from NeuroVault, label them with NeuroSynth terms, and compute ICA components across all the maps.

See :func:nilearn.datasets.fetch_neurovault documentation for more details.

In [ ]:
# Author: Ben Cipollini
# License: BSD
# Ported from code authored by Chris Filo Gorgolewski, Gael Varoquaux
import warnings

import numpy as np
from scipy import stats
from sklearn.decomposition import FastICA

from nilearn.datasets import fetch_neurovault
from nilearn.image import smooth_img

from nilearn.datasets import load_mni152_brain_mask
from nilearn.input_data import NiftiMasker

from nilearn import plotting

Get image and term data

In [ ]:
# Download images
# Here by default we only download 80 images to save time,
# but for better results I recommend using at least 200.
print("Fetching Neurovault images; "
      "if you haven't downloaded any Neurovault data before "
      "this will take several minutes.")
nv_data = fetch_neurovault(max_images=30, fetch_neurosynth_words=True)

images = nv_data['images']
term_weights = nv_data['word_frequencies']
vocabulary = nv_data['vocabulary']
if term_weights is None:
    term_weights = np.ones((len(images), 2))
    vocabulary = np.asarray(
        ["Neurosynth is down", "Please try again later"])

# Clean and report term scores
term_weights[term_weights < 0] = 0
total_scores = np.mean(term_weights, axis=0)

print("\nTop 10 neurosynth terms from downloaded images:\n")

for term_idx in np.argsort(total_scores)[-10:][::-1]:

Reshape and mask images

In [ ]:
print("\nReshaping and masking images.\n")

with warnings.catch_warnings():
    warnings.simplefilter('ignore', UserWarning)
    warnings.simplefilter('ignore', DeprecationWarning)

    mask_img = load_mni152_brain_mask()
    masker = NiftiMasker(
        mask_img=mask_img, memory='nilearn_cache', memory_level=1)
    masker =

    # Images may fail to be transformed, and are of different shapes,
    # so we need to transform one-by-one and keep track of failures.
    X = []
    is_usable = np.ones((len(images),), dtype=bool)

    for index, image_path in enumerate(images):
        # load image and remove nan and inf values.
        # applying smooth_img to an image with fwhm=None simply cleans up
        # non-finite values but otherwise doesn't modify the image.
        image = smooth_img(image_path, fwhm=None)
        except Exception as e:
            meta = nv_data['images_meta'][index]
            print("Failed to mask/reshape image: id: {0}; "
                  "name: '{1}'; collection: {2}; error: {3}".format(
                      meta.get('id'), meta.get('name'),
                      meta.get('collection_id'), e))
            is_usable[index] = False

# Now reshape list into 2D matrix, and remove failed images from terms
X = np.vstack(X)
term_weights = term_weights[is_usable, :]

Run ICA and map components to terms

In [ ]:
print("Running ICA; may take time...")
# We use a very small number of components as we have downloaded only 80
# images. For better results, increase the number of images downloaded
# and the number of components
n_components = 8
fast_ica = FastICA(n_components=n_components, random_state=0)
ica_maps = fast_ica.fit_transform(X.T).T

term_weights_for_components =, term_weights)
print('Done, plotting results.')

Generate figures

In [ ]:
with warnings.catch_warnings():
    warnings.simplefilter('ignore', DeprecationWarning)

    for index, (ic_map, ic_terms) in enumerate(
            zip(ica_maps, term_weights_for_components)):
        if -ic_map.min() > ic_map.max():
            # Flip the map's sign for prettiness
            ic_map = - ic_map
            ic_terms = - ic_terms

        ic_threshold = stats.scoreatpercentile(np.abs(ic_map), 90)
        ic_img = masker.inverse_transform(ic_map)
        important_terms = vocabulary[np.argsort(ic_terms)[-3:]]
        title = 'IC%i  %s' % (index, ', '.join(important_terms[::-1]))

            ic_img, threshold=ic_threshold, colorbar=False,

As we can see, some of the components capture cognitive or neurological maps, while other capture noise in the database. More data, better filtering, and better cognitive labels would give better maps

In [ ]:
# Done.