Single-subject data (two runs) in native space

The example shows the analysis of an SPM dataset, with two conditions: viewing a face image or a scrambled face image.

This example takes a lot of time because the input are lists of 3D images sampled in different positions (encoded by different affine functions).

See also

For more information see the dataset description.

Fetch and inspect the data

Fetch the SPM multimodal_faces data.

from nilearn.datasets import fetch_spm_multimodal_fmri

subject_data = fetch_spm_multimodal_fmri()

[fetch_spm_multimodal_fmri] Dataset created in
/home/runner/nilearn_data/spm_multimodal_fmri
[fetch_spm_multimodal_fmri] Missing 390 functional scans for session 1.
[fetch_spm_multimodal_fmri] Data absent, downloading...
[fetch_spm_multimodal_fmri] Downloading data from
https://www.fil.ion.ucl.ac.uk/spm/download/data/mmfaces/multimodal_fmri.zip ...
[fetch_spm_multimodal_fmri] Downloaded 892928 of 134263085 bytes (0.7%%,  2.7min
remaining)
[fetch_spm_multimodal_fmri] Downloaded 5562368 of 134263085 bytes (4.1%%,
48.3s remaining)
[fetch_spm_multimodal_fmri] Downloaded 10190848 of 134263085 bytes (7.6%%,
38.1s remaining)
[fetch_spm_multimodal_fmri] Downloaded 15638528 of 134263085 bytes (11.6%%,
31.3s remaining)
[fetch_spm_multimodal_fmri] Downloaded 20717568 of 134263085 bytes (15.4%%,
28.3s remaining)
[fetch_spm_multimodal_fmri] Downloaded 26468352 of 134263085 bytes (19.7%%,
25.2s remaining)
[fetch_spm_multimodal_fmri] Downloaded 31694848 of 134263085 bytes (23.6%%,
23.2s remaining)
[fetch_spm_multimodal_fmri] Downloaded 36864000 of 134263085 bytes (27.5%%,
21.6s remaining)
[fetch_spm_multimodal_fmri] Downloaded 42696704 of 134263085 bytes (31.8%%,
19.7s remaining)
[fetch_spm_multimodal_fmri] Downloaded 47939584 of 134263085 bytes (35.7%%,
18.4s remaining)
[fetch_spm_multimodal_fmri] Downloaded 53837824 of 134263085 bytes (40.1%%,
16.8s remaining)
[fetch_spm_multimodal_fmri] Downloaded 59187200 of 134263085 bytes (44.1%%,
15.6s remaining)
[fetch_spm_multimodal_fmri] Downloaded 65363968 of 134263085 bytes (48.7%%,
14.0s remaining)
[fetch_spm_multimodal_fmri] Downloaded 71163904 of 134263085 bytes (53.0%%,
12.7s remaining)
[fetch_spm_multimodal_fmri] Downloaded 78036992 of 134263085 bytes (58.1%%,
11.0s remaining)
[fetch_spm_multimodal_fmri] Downloaded 84729856 of 134263085 bytes (63.1%%,
9.6s remaining)
[fetch_spm_multimodal_fmri] Downloaded 92921856 of 134263085 bytes (69.2%%,
7.7s remaining)
[fetch_spm_multimodal_fmri] Downloaded 99221504 of 134263085 bytes (73.9%%,
6.5s remaining)
[fetch_spm_multimodal_fmri] Downloaded 106348544 of 134263085 bytes (79.2%%,
5.1s remaining)
[fetch_spm_multimodal_fmri] Downloaded 113319936 of 134263085 bytes (84.4%%,
3.8s remaining)
[fetch_spm_multimodal_fmri] Downloaded 121479168 of 134263085 bytes (90.5%%,
2.3s remaining)
[fetch_spm_multimodal_fmri] Downloaded 129024000 of 134263085 bytes (96.1%%,
0.9s remaining)
[fetch_spm_multimodal_fmri]  ...done. (25 seconds, 0 min)

[fetch_spm_multimodal_fmri] Extracting data from
/home/runner/nilearn_data/spm_multimodal_fmri/sub001/multimodal_fmri.zip...
[fetch_spm_multimodal_fmri] .. done.

[fetch_spm_multimodal_fmri] Downloading data from
https://www.fil.ion.ucl.ac.uk/spm/download/data/mmfaces/multimodal_smri.zip ...
[fetch_spm_multimodal_fmri] Downloaded 901120 of 6852766 bytes (13.1%%,    6.7s
remaining)
[fetch_spm_multimodal_fmri] Downloaded 4653056 of 6852766 bytes (67.9%%,    1.0s
remaining)
[fetch_spm_multimodal_fmri]  ...done. (4 seconds, 0 min)

[fetch_spm_multimodal_fmri] Extracting data from
/home/runner/nilearn_data/spm_multimodal_fmri/sub001/multimodal_smri.zip...
[fetch_spm_multimodal_fmri] .. done.

Let’s inspect one of the event files before using them.

import pandas as pd

events = [subject_data.events1, subject_data.events2]

events_dataframe = pd.read_csv(events[0], sep="\t")
events_dataframe["trial_type"].value_counts()

trial_type
scrambled    86
faces        64
Name: count, dtype: int64

We can confirm there are only 2 conditions in the dataset.

from nilearn.plotting import plot_event, show

plot_event(events)

show()

Resample the images: this is achieved by the concat_imgs function of Nilearn.

import warnings

from nilearn.image import concat_imgs, mean_img, resample_img

# Avoid getting too many warnings due to resampling
with warnings.catch_warnings():
    warnings.simplefilter("ignore")
    fmri_img = [
        concat_imgs(subject_data.func1, auto_resample=True),
        concat_imgs(subject_data.func2, auto_resample=True),
    ]
affine, shape = fmri_img[0].affine, fmri_img[0].shape
print("Resampling the second image (this takes time)...")
fmri_img[1] = resample_img(fmri_img[1], affine, shape[:3])

Resampling the second image (this takes time)...

Let’s create mean image for display purposes.

mean_image = mean_img(fmri_img)

Fit the model

Fit the GLM for the 2 runs by specifying a FirstLevelModel and then fitting it.

# Sample at the beginning of each acquisition.
slice_time_ref = 0.0
# We use a discrete cosine transform to model signal drifts.
drift_model = "cosine"
# The cutoff for the drift model is 0.01 Hz.
high_pass = 0.01
# The hemodynamic response function
hrf_model = "spm + derivative"

from nilearn.glm.first_level import FirstLevelModel

print("Fitting a GLM")
fmri_glm = FirstLevelModel(
    smoothing_fwhm=None,
    t_r=subject_data.t_r,
    hrf_model=hrf_model,
    drift_model=drift_model,
    high_pass=high_pass,
    verbose=1,
)


fmri_glm = fmri_glm.fit(fmri_img, events=events)

Fitting a GLM
\[FirstLevelModel.fit] Computing run 1 out of 2 runs (go take a coffee, a big
one).
\[FirstLevelModel.fit] Performing mask computation.
\[FirstLevelModel.fit] Masking took 0 seconds.
\[FirstLevelModel.fit] Performing GLM computation.
\[FirstLevelModel.fit] GLM took 1 seconds.
\[FirstLevelModel.fit] Computing run 2 out of 2 runs (2 seconds remaining).
\[FirstLevelModel.fit] Performing mask computation.
\[FirstLevelModel.fit] Masking took 0 seconds.
\[FirstLevelModel.fit] Performing GLM computation.
\[FirstLevelModel.fit] GLM took 1 seconds.
\[FirstLevelModel.fit] Computation of 2 runs done in 4 seconds.

View the results

Now we can compute contrast-related statistical maps (in z-scale), and plot them.

from nilearn.plotting import plot_stat_map

print("Computing contrasts")

Computing contrasts

We actually want more interesting contrasts. The simplest contrast just makes the difference between the two main conditions. We define the two opposite versions to run one-tailed t-tests.

contrasts = ["faces - scrambled", "scrambled - faces"]

Let’s store common parameters for all plots.

We plot the contrasts values overlaid on the mean fMRI image and we will use the z-score values as transparency, with any voxel with | Z-score | > 3 being fully opaque and any voxel with 0 < | Z-score | < 1.96 being partly transparent.

plot_param = {
    "vmin": 0,
    "display_mode": "z",
    "cut_coords": 3,
    "black_bg": True,
    "bg_img": mean_image,
    "cmap": "inferno",
    "transparency_range": [0, 3],
}

# Iterate on contrasts to compute and plot them.
for contrast_id in contrasts:
    print(f"\tcontrast id: {contrast_id}")

    results = fmri_glm.compute_contrast(contrast_id, output_type="all")

    plot_stat_map(
        results["stat"],
        title=contrast_id,
        transparency=results["z_score"],
        **plot_param,
    )

• 
• 

        contrast id: faces - scrambled
/home/runner/work/nilearn/nilearn/examples/04_glm_first_level/plot_spm_multimodal_faces.py:134: RuntimeWarning:

The same contrast will be used for all 2 runs. If the design matrices are not the same for all runs, (for example with different column names or column order across runs) you should pass contrast as an expression using the name of the conditions as they appear in the design matrices.

        contrast id: scrambled - faces
/home/runner/work/nilearn/nilearn/examples/04_glm_first_level/plot_spm_multimodal_faces.py:134: RuntimeWarning:

The same contrast will be used for all 2 runs. If the design matrices are not the same for all runs, (for example with different column names or column order across runs) you should pass contrast as an expression using the name of the conditions as they appear in the design matrices.

We also define the effects of interest contrast, a 2-dimensional contrasts spanning the two conditions.

import numpy as np

contrasts = np.eye(2)

results = fmri_glm.compute_contrast(contrasts, output_type="all")

plot_stat_map(
    results["stat"],
    title="effects of interest",
    transparency=results["z_score"],
    **plot_param,
)

show()

/home/runner/work/nilearn/nilearn/examples/04_glm_first_level/plot_spm_multimodal_faces.py:151: RuntimeWarning:

The same contrast will be used for all 2 runs. If the design matrices are not the same for all runs, (for example with different column names or column order across runs) you should pass contrast as an expression using the name of the conditions as they appear in the design matrices.

/home/runner/work/nilearn/nilearn/examples/04_glm_first_level/plot_spm_multimodal_faces.py:151: UserWarning:

F contrasts should have 20 columns, but it has only 2. The rest of the contrast was padded with zeros.

/home/runner/work/nilearn/nilearn/examples/04_glm_first_level/plot_spm_multimodal_faces.py:151: UserWarning:

Running approximate fixed effects on F statistics.

Based on the resulting maps we observe that the analysis results in wide activity for the ‘effects of interest’ contrast, showing the implications of large portions of the visual cortex in the conditions. By contrast, the differential effect between “faces” and “scrambled” involves sparser, more anterior and lateral regions. It also displays some responses in the frontal lobe.

Estimated memory usage: 1037 MB