Meta-analysis: Single predictor models
Contents
Meta-analysis: Single predictor models¶
Meta-analyses of 8 different features: 4 visual categories, 2 low level visual features and 2 auditory features
Each model contains only the target feature plus standard nuisance covariates.
Meta-analyses are run using analyze_collection, a helper function which pulls Neuroscout result maps from NeuroVault using analysis ids, creates a NiMare Dataset, and estimates meta-analytic maps.
from meta import analyze_collection
from pyns import Neuroscout
import json
from viz import plot_surfaces
from IPython.display import Image
%matplotlib inline
single_predictors = json.load(open('../analyses/models/single_predictor.json'))
ns = Neuroscout()
Meta-Analysis¶
low_visual = []
for label in ['brightness', 'shot_change']:
eff = analyze_collection(
single_predictors[label],
collection_kwargs={'estimator': 'afni'},
plot_kwargs=dict(vmax=18, cut_coords=range(-25, 55, 16), threshold=3.29)
)
low_visual.append(eff)
INFO:nimare.utils:Shared path detected: '/Users/rr48396/repos/neuroscout-paper/meta/images/brightness/'
WARNING:nimare.base:Masking out 9198 additional voxels. The updated masker is available in the Estimator.masker attribute.
meta-analyzing 19 for brightness
INFO:nimare.utils:Shared path detected: '/Users/rr48396/repos/neuroscout-paper/meta/images/shotChange/'
WARNING:nimare.base:Masking out 8768 additional voxels. The updated masker is available in the Estimator.masker attribute.
meta-analyzing 9 for shotChange
We now generate surface plots for the same images (displayed in Figure 3) and save them in a png file in figures.
We suppress the cell output and not display the plots to keep this notebook lightweight.
Note that it can take a few seconds to generate these plots.
%%capture
plot_surfaces(low_visual,
img_names=['brightness', 'shot_change'],
figname='low_level_visual')
Image(filename='figures/low_level_visual.png')
audio_outs = []
auditory = ['speech', 'rms']
for label in auditory:
eff = analyze_collection(
single_predictors[label],
collection_kwargs={'estimator': 'afni'},
plot_kwargs=dict(vmax=18, cut_coords=range(-25, 55, 16), threshold=3.29)
)
audio_outs.append(eff)
INFO:nimare.utils:Shared path detected: '/Users/rr48396/repos/neuroscout-paper/meta/images/speech/'
WARNING:nimare.base:Masking out 9202 additional voxels. The updated masker is available in the Estimator.masker attribute.
meta-analyzing 19 for speech
INFO:nimare.utils:Shared path detected: '/Users/rr48396/repos/neuroscout-paper/meta/images/rms/'
WARNING:nimare.base:Masking out 9202 additional voxels. The updated masker is available in the Estimator.masker attribute.
meta-analyzing 20 for rms
%%capture
plot_surfaces(audio_outs, img_names=auditory, figname='auditory')
Image(filename='figures/auditory.png')
high_visual = []
clarifai = ['text', 'landscape', 'building', 'tool']
for label in clarifai:
eff = analyze_collection(
single_predictors[label],
collection_kwargs={'estimator': 'afni'},
plot_kwargs=dict(vmax=18, cut_coords=range(-25, 55, 16), threshold=3.29)
)
high_visual.append(eff)
INFO:nimare.utils:Shared path detected: '/Users/rr48396/repos/neuroscout-paper/meta/images/text/'
WARNING:nimare.base:Masking out 9198 additional voxels. The updated masker is available in the Estimator.masker attribute.
meta-analyzing 19 for text
INFO:nimare.utils:Shared path detected: '/Users/rr48396/repos/neuroscout-paper/meta/images/landscape/'
WARNING:nimare.base:Masking out 9198 additional voxels. The updated masker is available in the Estimator.masker attribute.
meta-analyzing 19 for landscape
INFO:nimare.extract.utils:Dataset created in /Users/rr48396/repos/neuroscout-paper/meta/images/building
INFO:numexpr.utils:NumExpr defaulting to 8 threads.
INFO:nimare.utils:Shared path detected: '/Users/rr48396/repos/neuroscout-paper/meta/images/building/'
INFO:nimare.utils:Shared path detected: '/Users/rr48396/repos/neuroscout-paper/meta/images/building/'
WARNING:nimare.base:Masking out 9198 additional voxels. The updated masker is available in the Estimator.masker attribute.
meta-analyzing 19 for building
INFO:nimare.utils:Shared path detected: '/Users/rr48396/repos/neuroscout-paper/meta/images/tool/'
WARNING:nimare.base:Masking out 9198 additional voxels. The updated masker is available in the Estimator.masker attribute.
meta-analyzing 19 for tool
%%capture
plot_surfaces(high_visual, img_names=clarifai, figname='high_level_visual')
Image(filename='figures/high_level_visual.png')
Figure 4 meta-analytic maps¶
text_img = analyze_collection(
single_predictors['text'],
collection_kwargs={'estimator': 'afni'},
plot_kwargs=dict(vmax=18, cut_coords=[-10], threshold=3.29)
)[0]
INFO:nimare.utils:Shared path detected: '/home/alejandro/repos/neuroscout-paper/meta/images/text/'
WARNING:nimare.base:Masking out 7754 additional voxels. The updated masker is available in the Estimator.masker attribute.
meta-analyzing 20 for text
building_img = analyze_collection(
single_predictors['building'],
collection_kwargs={'estimator': 'afni'},
plot_kwargs=dict(vmax=18, threshold=3.29, cut_coords=[-12])
)[0]
INFO:nimare.extract.utils:Dataset created in /home/alejandro/repos/neuroscout-paper/meta/images/building
INFO:nimare.utils:Shared path detected: '/home/alejandro/repos/neuroscout-paper/meta/images/building/'
INFO:nimare.utils:Shared path detected: '/home/alejandro/repos/neuroscout-paper/meta/images/building/'
WARNING:nimare.base:Masking out 7754 additional voxels. The updated masker is available in the Estimator.masker attribute.
meta-analyzing 20 for building
rms_img = analyze_collection(
single_predictors['rms'],
collection_kwargs={'estimator': 'afni'},
plot_kwargs=dict(vmax=18, cut_coords=[6], threshold=3.29)
)[0]
INFO:nimare.utils:Shared path detected: '/home/alejandro/repos/neuroscout-paper/meta/images/rms/'
WARNING:nimare.base:Masking out 7754 additional voxels. The updated masker is available in the Estimator.masker attribute.
meta-analyzing 21 for rms
