Note
Go to the end to download the full example code.
Extracting Regionwise Signals From Activity Recording
siibra integrates nilearn to allow seemles extraction of signals given recordings such as fMRI, PET, and others. This notebook downloads two fMRI images from AOMIC-PIOP2 dataset (https://openneuro.org/datasets/ds002790/versions/2.0.0) and compares the extraction results for different tasks.
We start by importing the Python packages used in this notebook. siibra provides access to brain atlases, spaces, maps, and datasets. nilearn.plotting is used for interactive neuroimaging visualizations. plotly.express.imshow helps us display matrix-like signal data, while matplotlib and pandas are used for plotting and tabular data analysis.
import siibra
from nilearn import plotting
from plotly.express import imshow
import matplotlib.pyplot as plt
import pandas as pd
import json
import pathlib
import requests
Note
The surface time-series analysis at the end of this notebook requires a FreeSurfer fsaverage5 template and a parcellation map defined in the same space. The Julich-Brain fsaverage5 map is not available out of the box, however, fsaverage7 version can be downsampled using Workbench. This notebook loads the map from a dedicated configuration branch where this map is precomputred.
siibra.use_configuration(
siibra.retrieval.repositories.GithubConnector(
owner="FZJ-INM1-BDA",
repo="siibra-configurations",
reftag="feat_timeseries_volume",
)
)
The fMRI data used in this example comes from the AOMIC-PIOP2 dataset hosted on OpenNeuro. Here we define the subject identifier and construct the base URL that points to the preprocessed fMRI files generated by fMRIPrep. This allows us to build the full download URLs for different tasks by changing only the file name.
subject = "sub-0001"
dataset_base_url = "https://s3.amazonaws.com/openneuro.org/ds002790/"
folder = f"derivatives/fmriprep/{subject}/func/"
url_template = dataset_base_url + folder + "{file}"
We first load the preprocessed working-memory fMRI image. The file is in MNI152 space, which is a commonly used standard anatomical reference space. The image is wrapped as a siibra volume using siibra.volumes.from_url, which allows the data to be passed directly to siibra-based workflows. The time=[] argument indicates that the full 4D time series should be used.
file = f"{subject}_task-workingmemory_acq-seq_space-MNI152NLin2009cAsym_desc-preproc_bold.nii.gz"
workingmemory_fmri = siibra.volumes.from_url(
url_template.format(file=file), space="mni152", time=[]
)
print(type(workingmemory_fmri))
<class 'siibra.volumes.volume.TimeSeriesVolume'>
Next, we load the resting-state fMRI recording from the same subject. This image has the same spatial reference space as the working-memory image, which makes it possible to compare the extracted region-wise signals across both recordings using the same brain atlas.
file = f"{subject}_task-restingstate_acq-seq_space-MNI152NLin2009cAsym_desc-preproc_bold.nii.gz"
restingstate_fmri = siibra.volumes.from_url(
url_template.format(file=file), space="mni152", time=[]
)
Functional MRI recordings often contain noise caused by motion, physiology, scanner drift, or other non-neural sources. The fMRIPrep outputs include confound regressors that can be used during signal extraction. Here we load the confound files for both the working-memory and resting-state recordings into pandas DataFrames. Later, we will use the CSF signal as an example confound.
filename = f"{subject}_task-workingmemory_acq-seq_desc-confounds_regressors.tsv"
workingmemory_confounds = pd.read_csv(url_template.format(file=filename), sep="\t")
filename = f"{subject}_task-restingstate_acq-seq_desc-confounds_regressors.tsv"
restingstate_confounds = pd.read_csv(url_template.format(file=filename), sep="\t")
We now select the brain parcellation that will define the regions from which signals are extracted. In this example, we use the Julich-Brain 3.1 in MNI152 space. Since the fMRI data is also in MNI152 space, this atlas is suitable for extracting region-wise signals from the volume data.
julichbrain = siibra.get_map("julich 3.1", "mni152")
The extract_signals_with_nilearn` method extracts one signal per region from the input fMRI image. Internally, this uses nilearn functionality to summarize the voxel-wise data within each labelled brain region. The result is returned as a pandas DataFrame, where each column corresponds to a region and each row corresponds to one time point of the fMRI recording. We also pass the CSF confound signal so that nilearn can regress it out during signal extraction.
workingmemory_signals = julichbrain.extract_signals_with_nilearn(
workingmemory_fmri,
confounds=workingmemory_confounds["csf"].values,
)
workingmemory_signals
[NiftiLabelsMasker.wrapped] Loading data from <nibabel.nifti1.Nifti1Image object at 0x7f04846890d0>
[NiftiLabelsMasker.wrapped] Loading regions from <nibabel.nifti1.Nifti1Image object at 0x7f0545c92590>
[NiftiLabelsMasker.wrapped] Resampling regions
/actions-runner/_work/siibra-python/siibra-python/siibra/volumes/parcellationmap.py:1713: UserWarning: After resampling the label image to the data image, the following labels were removed: {np.uint16(17), np.uint16(125), np.uint16(61), np.uint16(223)}. Label image only contains 411 labels (including background).
masker.fit_transform(source, confounds=confounds, sample_mask=sample_mask)
[NiftiLabelsMasker.wrapped] Finished fit
/actions-runner/_work/siibra-python/siibra-python/siibra/volumes/parcellationmap.py:1713: FutureWarning: boolean values for 'standardize' will be deprecated in nilearn 0.15.0.
Use 'zscore_sample' instead of 'True' or use 'None' instead of 'False'.
masker.fit_transform(source, confounds=confounds, sample_mask=sample_mask)
________________________________________________________________________________
[Memory] Calling nilearn.maskers.base_masker.filter_and_extract...
filter_and_extract(<nibabel.nifti1.Nifti1Image object at 0x7f04846890d0>, <nilearn.maskers.nifti_labels_masker._ExtractionFunctor object at 0x7f04cbb6d6d0>,
{ 'background_label': 0,
'clean_args': None,
'clean_kwargs': {},
'cmap': None,
'detrend': False,
'dtype': None,
'high_pass': None,
'high_variance_confounds': False,
'keep_masked_labels': False,
'labels': None,
'labels_img': <nibabel.nifti1.Nifti1Image object at 0x7f0545c92590>,
'low_pass': None,
'lut': index name color
0 1 Area 3b (PostCG) left #f63e00
1 2 Area hOc4la (LOC) left #e9a8bc
2 3 Area PF (IPL) left #e1d23c
3 4 Area CoS1 (CoS) left #007880
4 5 Astr (Amygdala) left NaN
.. ... ... ...
409 410 Area Fo2 (OFC) right #00ff00
410 411 Area Te 2.2 (STG) right #8a3ec0
411 412 Area hIP7 (IPS) right #46c4da
412 413 Area 7PC (SPL) right #fb581c
413 414 Area TeI (STG) right #164e5c
[414 rows x 3 columns],
'mask_img': None,
'reports': True,
'smoothing_fwhm': None,
'standardize': False,
'standardize_confounds': True,
'strategy': 'mean',
't_r': None,
'target_affine': None,
'target_shape': None}, confounds=[array([34159.246094, ..., 34129.640625], shape=(160,))], sample_mask=None, memory=Memory(location=/actions-runner/_work/siibra-python/siibra-python/.ci-docs-37911713724-3-notebooks/siibra-cache/joblib/joblib), memory_level=1, verbose=1, sklearn_output_config=None)
[NiftiLabelsMasker.wrapped] Loading data from <nibabel.nifti1.Nifti1Image object at 0x7f04846890d0>
[NiftiLabelsMasker.wrapped] Extracting region signals
[NiftiLabelsMasker.wrapped] Cleaning extracted signals
/actions-runner/_work/siibra-python/siibra-python/siibra/volumes/parcellationmap.py:1713: FutureWarning: boolean values for 'standardize' will be deprecated in nilearn 0.15.0.
Use 'zscore_sample' instead of 'True' or use 'None' instead of 'False'.
masker.fit_transform(source, confounds=confounds, sample_mask=sample_mask)
_______________________________________________filter_and_extract - 1.4s, 0.0min
After extracting the time series, we summarize each regional signal using pandas. The describe() method computes statistics such as the mean, standard deviation, minimum, maximum, and quartiles. We transpose the result so that each row corresponds to one brain region, then sort the regions by their mean signal value.
workingmemory_signals_stats = workingmemory_signals.describe().T.sort_values(
"mean", ascending=False
)
workingmemory_signals_stats
To inspect the regional time series, we select the 20 regions with the highest mean signal values during the working-memory task. The selected signals are transposed so that regions are shown along one axis and time points along the other. This provides a compact overview of how the strongest regional signals vary over time.
imshow(workingmemory_signals[workingmemory_signals_stats.index[:20]].T)
We repeat the same extraction workflow for the resting-state recording. This gives us a second region-wise signal matrix, again with time points as rows and atlas regions as columns. We then compute summary statistics for each region and sort them by the mean signal value, allowing direct comparison with the working-memory results.
restingstate_signals = julichbrain.extract_signals_with_nilearn(
restingstate_fmri,
confounds=restingstate_confounds["csf"].values,
)
restingstate_signals_stats = restingstate_signals.describe().T.sort_values(
"mean", ascending=False
)
restingstate_signals_stats
[NiftiLabelsMasker.wrapped] Loading data from <nibabel.nifti1.Nifti1Image object at 0x7f05436bbc90>
[NiftiLabelsMasker.wrapped] Loading regions from <nibabel.nifti1.Nifti1Image object at 0x7f0545c92590>
[NiftiLabelsMasker.wrapped] Resampling regions
/actions-runner/_work/siibra-python/siibra-python/siibra/volumes/parcellationmap.py:1713: UserWarning: After resampling the label image to the data image, the following labels were removed: {np.uint16(17), np.uint16(125), np.uint16(61), np.uint16(223)}. Label image only contains 411 labels (including background).
masker.fit_transform(source, confounds=confounds, sample_mask=sample_mask)
[NiftiLabelsMasker.wrapped] Finished fit
/actions-runner/_work/siibra-python/siibra-python/siibra/volumes/parcellationmap.py:1713: FutureWarning: boolean values for 'standardize' will be deprecated in nilearn 0.15.0.
Use 'zscore_sample' instead of 'True' or use 'None' instead of 'False'.
masker.fit_transform(source, confounds=confounds, sample_mask=sample_mask)
________________________________________________________________________________
[Memory] Calling nilearn.maskers.base_masker.filter_and_extract...
filter_and_extract(<nibabel.nifti1.Nifti1Image object at 0x7f05436bbc90>, <nilearn.maskers.nifti_labels_masker._ExtractionFunctor object at 0x7f0542b80190>,
{ 'background_label': 0,
'clean_args': None,
'clean_kwargs': {},
'cmap': None,
'detrend': False,
'dtype': None,
'high_pass': None,
'high_variance_confounds': False,
'keep_masked_labels': False,
'labels': None,
'labels_img': <nibabel.nifti1.Nifti1Image object at 0x7f0545c92590>,
'low_pass': None,
'lut': index name color
0 1 Area 3b (PostCG) left #f63e00
1 2 Area hOc4la (LOC) left #e9a8bc
2 3 Area PF (IPL) left #e1d23c
3 4 Area CoS1 (CoS) left #007880
4 5 Astr (Amygdala) left NaN
.. ... ... ...
409 410 Area Fo2 (OFC) right #00ff00
410 411 Area Te 2.2 (STG) right #8a3ec0
411 412 Area hIP7 (IPS) right #46c4da
412 413 Area 7PC (SPL) right #fb581c
413 414 Area TeI (STG) right #164e5c
[414 rows x 3 columns],
'mask_img': None,
'reports': True,
'smoothing_fwhm': None,
'standardize': False,
'standardize_confounds': True,
'strategy': 'mean',
't_r': None,
'target_affine': None,
'target_shape': None}, confounds=[array([34207.414062, ..., 34218.066406], shape=(240,))], sample_mask=None, memory=Memory(location=/actions-runner/_work/siibra-python/siibra-python/.ci-docs-37911713724-3-notebooks/siibra-cache/joblib/joblib), memory_level=1, verbose=1, sklearn_output_config=None)
[NiftiLabelsMasker.wrapped] Loading data from <nibabel.nifti1.Nifti1Image object at 0x7f05436bbc90>
[NiftiLabelsMasker.wrapped] Extracting region signals
[NiftiLabelsMasker.wrapped] Cleaning extracted signals
/actions-runner/_work/siibra-python/siibra-python/siibra/volumes/parcellationmap.py:1713: FutureWarning: boolean values for 'standardize' will be deprecated in nilearn 0.15.0.
Use 'zscore_sample' instead of 'True' or use 'None' instead of 'False'.
masker.fit_transform(source, confounds=confounds, sample_mask=sample_mask)
_______________________________________________filter_and_extract - 1.7s, 0.0min
Here we create horizontal bar plots for the top 10 regions in each condition. The bar height represents the mean extracted signal, and the error bars show the standard deviation across time. These plots provide a simple summary of which regions have the strongest average signal in each recording.
fig, axs = plt.subplots(1, 2, figsize=(12, 5))
restingstate_signals_stats.iloc[:10, :].plot(
ax=axs[0], title="resting state", kind="barh", y="mean", xerr="std"
)
workingmemory_signals_stats.iloc[:10, :].plot(
ax=axs[1], title="working memory", kind="barh", y="mean", xerr="std"
)
fig.tight_layout()
To compare the two recordings, we take the union of the top-ranked regions from the working-memory and resting-state summaries. We then create boxplots for the selected regions in both conditions. This allows us to compare the distribution of signal values across time for the same anatomical regions in the two fMRI recordings.
union = workingmemory_signals_stats.index[:20].union(
restingstate_signals_stats.index[:20]
)
union_sorted = (
workingmemory_signals_stats.loc[union, :].sort_values("mean", ascending=False).index
)
fig, ax = plt.subplots(figsize=(10, 10))
workingmemory_signals[union_sorted].boxplot(
ax=ax, rot=90, color="b", label="working memory",
flierprops={"marker": ".", "markeredgecolor": "b", "markersize": 3},
)
restingstate_signals[union_sorted].boxplot(
ax=ax, rot=90, color="r", label="resting state",
flierprops={"marker": ".", "markeredgecolor": "r", "markersize": 3},
)
ax.legend()
ax.set_title(f"Comparison of top {len(union_sorted)} regions for {subject}")
fig.tight_layout()
Instead of extracting the mean signal from each region, we can also use a different summary strategy. Here we use strategy=”variance” to compute the variance of the fMRI signal within each atlas region. This can be useful when we are interested in spatial or temporal variability rather than the average signal intensity.
workingmemory_signals_var = julichbrain.extract_signals_with_nilearn(
workingmemory_fmri,
confounds=workingmemory_confounds["csf"].values,
strategy="variance",
)
workingmemory_signals_var
[NiftiLabelsMasker.wrapped] Loading data from <nibabel.nifti1.Nifti1Image object at 0x7f05bb66cd90>
[NiftiLabelsMasker.wrapped] Loading regions from <nibabel.nifti1.Nifti1Image object at 0x7f0545d00cd0>
[NiftiLabelsMasker.wrapped] Resampling regions
/actions-runner/_work/siibra-python/siibra-python/siibra/volumes/parcellationmap.py:1713: UserWarning: After resampling the label image to the data image, the following labels were removed: {np.uint16(17), np.uint16(125), np.uint16(61), np.uint16(223)}. Label image only contains 411 labels (including background).
masker.fit_transform(source, confounds=confounds, sample_mask=sample_mask)
[NiftiLabelsMasker.wrapped] Finished fit
/actions-runner/_work/siibra-python/siibra-python/siibra/volumes/parcellationmap.py:1713: FutureWarning: boolean values for 'standardize' will be deprecated in nilearn 0.15.0.
Use 'zscore_sample' instead of 'True' or use 'None' instead of 'False'.
masker.fit_transform(source, confounds=confounds, sample_mask=sample_mask)
________________________________________________________________________________
[Memory] Calling nilearn.maskers.base_masker.filter_and_extract...
filter_and_extract(<nibabel.nifti1.Nifti1Image object at 0x7f05bb66cd90>, <nilearn.maskers.nifti_labels_masker._ExtractionFunctor object at 0x7f05bb541a50>,
{ 'background_label': 0,
'clean_args': None,
'clean_kwargs': {},
'cmap': None,
'detrend': False,
'dtype': None,
'high_pass': None,
'high_variance_confounds': False,
'keep_masked_labels': False,
'labels': None,
'labels_img': <nibabel.nifti1.Nifti1Image object at 0x7f0545d00cd0>,
'low_pass': None,
'lut': index name color
0 1 Area 3b (PostCG) left #f63e00
1 2 Area hOc4la (LOC) left #e9a8bc
2 3 Area PF (IPL) left #e1d23c
3 4 Area CoS1 (CoS) left #007880
4 5 Astr (Amygdala) left NaN
.. ... ... ...
409 410 Area Fo2 (OFC) right #00ff00
410 411 Area Te 2.2 (STG) right #8a3ec0
411 412 Area hIP7 (IPS) right #46c4da
412 413 Area 7PC (SPL) right #fb581c
413 414 Area TeI (STG) right #164e5c
[414 rows x 3 columns],
'mask_img': None,
'reports': True,
'smoothing_fwhm': None,
'standardize': False,
'standardize_confounds': True,
'strategy': 'variance',
't_r': None,
'target_affine': None,
'target_shape': None}, confounds=[array([34159.246094, ..., 34129.640625], shape=(160,))], sample_mask=None, memory=Memory(location=/actions-runner/_work/siibra-python/siibra-python/.ci-docs-37911713724-3-notebooks/siibra-cache/joblib/joblib), memory_level=1, verbose=1, sklearn_output_config=None)
[NiftiLabelsMasker.wrapped] Loading data from <nibabel.nifti1.Nifti1Image object at 0x7f05bb66cd90>
[NiftiLabelsMasker.wrapped] Extracting region signals
/actions-runner/_work/siibra-python/siibra-python/.ci-docs-37911713724-3-notebooks/venv/lib/python3.11/site-packages/scipy/ndimage/_measurements.py:647: RuntimeWarning: invalid value encountered in divide
means = sums / counts
[NiftiLabelsMasker.wrapped] Cleaning extracted signals
/actions-runner/_work/siibra-python/siibra-python/siibra/volumes/parcellationmap.py:1713: FutureWarning: boolean values for 'standardize' will be deprecated in nilearn 0.15.0.
Use 'zscore_sample' instead of 'True' or use 'None' instead of 'False'.
masker.fit_transform(source, confounds=confounds, sample_mask=sample_mask)
_______________________________________________filter_and_extract - 2.0s, 0.0min
The extracted region-wise variance values are mapped back onto the labelled atlas using colorize method. This creates a new image where each region is assigned the corresponding extracted value. We then visualize one time point of the resulting image with nilearn.plotting.view_img, using a magma colormap to show regional differences.
colored_map = julichbrain.colorize(workingmemory_signals_var.mean())
plotting.view_img(colored_map, cmap="magma", symmetric_cmap=False)
[NiftiLabelsMasker.fit] Loading regions from <nibabel.nifti1.Nifti1Image object at 0x7f0545d00cd0>
[NiftiLabelsMasker.fit] Finished fit
[NiftiLabelsMasker.inverse_transform] Computing image from signals
/actions-runner/_work/siibra-python/siibra-python/.ci-docs-37911713724-3-notebooks/venv/lib/python3.11/site-packages/numpy/_core/fromnumeric.py:840: UserWarning: Warning: 'partition' will ignore the 'mask' of the MaskedArray.
a.partition(kth, axis=axis, kind=kind, order=order)
So far, we used a labelled atlas map, where each voxel belongs to a discrete region. As an alternative, siibra can also work with probabilistic maps. Here we load the DiFuMo 64-component statistical map and use it to extract signals from the working-memory fMRI data. Probabilistic maps can represent spatially overlapping components and are often useful for functional parcellations.
difumo64_pmaps = siibra.get_map("DiFuMo 64", "mni152", "statistical")
difumo64_pmaps.extract_signals_with_nilearn(workingmemory_fmri)
[NiftiMapsMasker.wrapped] Loading data from <nibabel.nifti1.Nifti1Image object at 0x7f05bbc3c350>
[NiftiMapsMasker.wrapped] Loading regions from <nibabel.nifti1.Nifti1Image object at 0x7f0542a7fd50>
[NiftiMapsMasker.wrapped] Resampling regions
[NiftiMapsMasker.wrapped] Finished fit
/actions-runner/_work/siibra-python/siibra-python/siibra/volumes/parcellationmap.py:1713: FutureWarning: boolean values for 'standardize' will be deprecated in nilearn 0.15.0.
Use 'zscore_sample' instead of 'True' or use 'None' instead of 'False'.
masker.fit_transform(source, confounds=confounds, sample_mask=sample_mask)
[NiftiMapsMasker.wrapped] Loading data from <nibabel.nifti1.Nifti1Image object at 0x7f05bbc3c350>
[NiftiMapsMasker.wrapped] Extracting region signals
[NiftiMapsMasker.wrapped] Cleaning extracted signals
A similar workflow for surface-based fMRI data is possible.
julichbrain_fs5 = siibra.get_map("julich 3.1", "fsaverage5")
fs_variant = "inflated"
file_template = (
f"{subject}_task-workingmemory_acq-seq_space-fsaverage5_hemi-" + "{}.func.gii"
)
url_template = dataset_base_url + folder + file_template
giivol = siibra.volumes.from_url(
{h: url_template.format(h[0]) for h in ["Left", "Right"]},
space="fsaverage5",
format="gii-timeseries",
time=list(range(160)),
)
signals = julichbrain_fs5.extract_signals_with_nilearn(
giivol, surface_variant=fs_variant
)
signals
[SurfaceLabelsMasker.wrapped] Loading data from <SurfaceImage (20484, 160)>
[SurfaceLabelsMasker.wrapped] Loading regions from <SurfaceImage (20484,)>
[SurfaceLabelsMasker.wrapped] Finished fit
/actions-runner/_work/siibra-python/siibra-python/siibra/volumes/parcellationmap.py:1713: FutureWarning: boolean values for 'standardize' will be deprecated in nilearn 0.15.0.
Use 'zscore_sample' instead of 'True' or use 'None' instead of 'False'.
masker.fit_transform(source, confounds=confounds, sample_mask=sample_mask)
[SurfaceLabelsMasker.wrapped] Extracting region signals
[SurfaceLabelsMasker.wrapped] Cleaning extracted signals
colorize` method can be used on surface maps as well. In which case, the output is a nilearn.surface.SurfaceImage for compatibility with nilearn.plotting
surf_im = julichbrain_fs5.colorize(signals.std(), surface_variant=fs_variant)
plotting.view_surf(
surf_map=surf_im,
hemi="both",
symmetric_cmap=False,
cmap="magma",
)
[SurfaceLabelsMasker.fit] Loading regions from <SurfaceImage (20484,)>
[SurfaceLabelsMasker.fit] Finished fit
[SurfaceLabelsMasker.inverse_transform] Computing image from signals
config_output_folder = "./custom-configurations"
def create_config(subject, task):
folder = f"derivatives/fmriprep/{subject}/func/"
file_prefix = f"{subject}_task-{task}_acq-seq_space-fsaverage5"
url_template = dataset_base_url + folder + file_prefix + "_hemi-{}.func.gii"
conf = {
"@type": "siibra/feature/voi/v0.1",
"name": f"{subject} - fMRI - ds002790",
"modality": f"fMRI - {task}",
"space": {"name": "freesurfer fsaverage5"},
"providers": {
"gii-timeseries": {h: url_template.format(h[0]) for h in ["Left", "Right"]}
},
"time": {"start": 0, "stop": 10, "num": 160},
}
# discard unavailable data
for url in conf["providers"]["gii-timeseries"].values():
req = requests.get(url, stream=True)
if not req.ok:
return
# create json config for future reuse
filepath = f"{config_output_folder}/features/images/vois/fmri/{file_prefix}.json"
pathlib.Path(filepath).parent.mkdir(parents=True, exist_ok=True)
with open(filepath, "wt") as fp:
json.dump(conf, fp=fp, indent="\t")
num_of_subjects = 10
tasks = ["restingstate", "workingmemory", "stopsignal"]
for task in tasks:
for subject in range(num_of_subjects):
create_config(f"sub-{subject + 1:04d}", task)
siibra.extend_configuration(config_output_folder)
fmri_vols = siibra.features.get(julichbrain_fs5.space, "FMRIVolumeOfInterest")
print(len(fmri_vols))
29
regional_signal_averages = {}
top_regions = pd.Index([])
for task in tasks:
regional_signal_averages[task] = pd.concat(
(
julichbrain_fs5.extract_signals_with_nilearn(
fmri_vol,
strategy="mean", # mean signal value per region per timepoint
surface_variant=fs_variant,
).std() # std over time per region
for fmri_vol in fmri_vols
if task in fmri_vol.name
),
axis=1,
)
# compute the average of all subjects
regional_signal_averages[task]["mean"] = regional_signal_averages[task].mean(axis=1)
# find top 25 regions
task_top_regions = regional_signal_averages[task].sort_values("mean", ascending=False).index[:25]
top_regions = top_regions.union(task_top_regions)
[SurfaceLabelsMasker.wrapped] Loading data from <SurfaceImage (20484, 240)>
[SurfaceLabelsMasker.wrapped] Loading regions from <SurfaceImage (20484,)>
[SurfaceLabelsMasker.wrapped] Finished fit
/actions-runner/_work/siibra-python/siibra-python/siibra/volumes/parcellationmap.py:1713: FutureWarning: boolean values for 'standardize' will be deprecated in nilearn 0.15.0.
Use 'zscore_sample' instead of 'True' or use 'None' instead of 'False'.
masker.fit_transform(source, confounds=confounds, sample_mask=sample_mask)
[SurfaceLabelsMasker.wrapped] Extracting region signals
[SurfaceLabelsMasker.wrapped] Cleaning extracted signals
[SurfaceLabelsMasker.wrapped] Loading data from <SurfaceImage (20484, 240)>
[SurfaceLabelsMasker.wrapped] Loading regions from <SurfaceImage (20484,)>
[SurfaceLabelsMasker.wrapped] Finished fit
/actions-runner/_work/siibra-python/siibra-python/siibra/volumes/parcellationmap.py:1713: FutureWarning: boolean values for 'standardize' will be deprecated in nilearn 0.15.0.
Use 'zscore_sample' instead of 'True' or use 'None' instead of 'False'.
masker.fit_transform(source, confounds=confounds, sample_mask=sample_mask)
[SurfaceLabelsMasker.wrapped] Extracting region signals
[SurfaceLabelsMasker.wrapped] Cleaning extracted signals
[SurfaceLabelsMasker.wrapped] Loading data from <SurfaceImage (20484, 240)>
[SurfaceLabelsMasker.wrapped] Loading regions from <SurfaceImage (20484,)>
[SurfaceLabelsMasker.wrapped] Finished fit
/actions-runner/_work/siibra-python/siibra-python/siibra/volumes/parcellationmap.py:1713: FutureWarning: boolean values for 'standardize' will be deprecated in nilearn 0.15.0.
Use 'zscore_sample' instead of 'True' or use 'None' instead of 'False'.
masker.fit_transform(source, confounds=confounds, sample_mask=sample_mask)
[SurfaceLabelsMasker.wrapped] Extracting region signals
[SurfaceLabelsMasker.wrapped] Cleaning extracted signals
[SurfaceLabelsMasker.wrapped] Loading data from <SurfaceImage (20484, 240)>
[SurfaceLabelsMasker.wrapped] Loading regions from <SurfaceImage (20484,)>
[SurfaceLabelsMasker.wrapped] Finished fit
/actions-runner/_work/siibra-python/siibra-python/siibra/volumes/parcellationmap.py:1713: FutureWarning: boolean values for 'standardize' will be deprecated in nilearn 0.15.0.
Use 'zscore_sample' instead of 'True' or use 'None' instead of 'False'.
masker.fit_transform(source, confounds=confounds, sample_mask=sample_mask)
[SurfaceLabelsMasker.wrapped] Extracting region signals
[SurfaceLabelsMasker.wrapped] Cleaning extracted signals
[SurfaceLabelsMasker.wrapped] Loading data from <SurfaceImage (20484, 240)>
[SurfaceLabelsMasker.wrapped] Loading regions from <SurfaceImage (20484,)>
[SurfaceLabelsMasker.wrapped] Finished fit
/actions-runner/_work/siibra-python/siibra-python/siibra/volumes/parcellationmap.py:1713: FutureWarning: boolean values for 'standardize' will be deprecated in nilearn 0.15.0.
Use 'zscore_sample' instead of 'True' or use 'None' instead of 'False'.
masker.fit_transform(source, confounds=confounds, sample_mask=sample_mask)
[SurfaceLabelsMasker.wrapped] Extracting region signals
[SurfaceLabelsMasker.wrapped] Cleaning extracted signals
[SurfaceLabelsMasker.wrapped] Loading data from <SurfaceImage (20484, 240)>
[SurfaceLabelsMasker.wrapped] Loading regions from <SurfaceImage (20484,)>
[SurfaceLabelsMasker.wrapped] Finished fit
/actions-runner/_work/siibra-python/siibra-python/siibra/volumes/parcellationmap.py:1713: FutureWarning: boolean values for 'standardize' will be deprecated in nilearn 0.15.0.
Use 'zscore_sample' instead of 'True' or use 'None' instead of 'False'.
masker.fit_transform(source, confounds=confounds, sample_mask=sample_mask)
[SurfaceLabelsMasker.wrapped] Extracting region signals
[SurfaceLabelsMasker.wrapped] Cleaning extracted signals
[SurfaceLabelsMasker.wrapped] Loading data from <SurfaceImage (20484, 240)>
[SurfaceLabelsMasker.wrapped] Loading regions from <SurfaceImage (20484,)>
[SurfaceLabelsMasker.wrapped] Finished fit
/actions-runner/_work/siibra-python/siibra-python/siibra/volumes/parcellationmap.py:1713: FutureWarning: boolean values for 'standardize' will be deprecated in nilearn 0.15.0.
Use 'zscore_sample' instead of 'True' or use 'None' instead of 'False'.
masker.fit_transform(source, confounds=confounds, sample_mask=sample_mask)
[SurfaceLabelsMasker.wrapped] Extracting region signals
[SurfaceLabelsMasker.wrapped] Cleaning extracted signals
[SurfaceLabelsMasker.wrapped] Loading data from <SurfaceImage (20484, 240)>
[SurfaceLabelsMasker.wrapped] Loading regions from <SurfaceImage (20484,)>
[SurfaceLabelsMasker.wrapped] Finished fit
/actions-runner/_work/siibra-python/siibra-python/siibra/volumes/parcellationmap.py:1713: FutureWarning: boolean values for 'standardize' will be deprecated in nilearn 0.15.0.
Use 'zscore_sample' instead of 'True' or use 'None' instead of 'False'.
masker.fit_transform(source, confounds=confounds, sample_mask=sample_mask)
[SurfaceLabelsMasker.wrapped] Extracting region signals
[SurfaceLabelsMasker.wrapped] Cleaning extracted signals
[SurfaceLabelsMasker.wrapped] Loading data from <SurfaceImage (20484, 240)>
[SurfaceLabelsMasker.wrapped] Loading regions from <SurfaceImage (20484,)>
[SurfaceLabelsMasker.wrapped] Finished fit
/actions-runner/_work/siibra-python/siibra-python/siibra/volumes/parcellationmap.py:1713: FutureWarning: boolean values for 'standardize' will be deprecated in nilearn 0.15.0.
Use 'zscore_sample' instead of 'True' or use 'None' instead of 'False'.
masker.fit_transform(source, confounds=confounds, sample_mask=sample_mask)
[SurfaceLabelsMasker.wrapped] Extracting region signals
[SurfaceLabelsMasker.wrapped] Cleaning extracted signals
[SurfaceLabelsMasker.wrapped] Loading data from <SurfaceImage (20484, 160)>
[SurfaceLabelsMasker.wrapped] Loading regions from <SurfaceImage (20484,)>
[SurfaceLabelsMasker.wrapped] Finished fit
/actions-runner/_work/siibra-python/siibra-python/siibra/volumes/parcellationmap.py:1713: FutureWarning: boolean values for 'standardize' will be deprecated in nilearn 0.15.0.
Use 'zscore_sample' instead of 'True' or use 'None' instead of 'False'.
masker.fit_transform(source, confounds=confounds, sample_mask=sample_mask)
[SurfaceLabelsMasker.wrapped] Extracting region signals
[SurfaceLabelsMasker.wrapped] Cleaning extracted signals
[SurfaceLabelsMasker.wrapped] Loading data from <SurfaceImage (20484, 160)>
[SurfaceLabelsMasker.wrapped] Loading regions from <SurfaceImage (20484,)>
[SurfaceLabelsMasker.wrapped] Finished fit
/actions-runner/_work/siibra-python/siibra-python/siibra/volumes/parcellationmap.py:1713: FutureWarning: boolean values for 'standardize' will be deprecated in nilearn 0.15.0.
Use 'zscore_sample' instead of 'True' or use 'None' instead of 'False'.
masker.fit_transform(source, confounds=confounds, sample_mask=sample_mask)
[SurfaceLabelsMasker.wrapped] Extracting region signals
[SurfaceLabelsMasker.wrapped] Cleaning extracted signals
[SurfaceLabelsMasker.wrapped] Loading data from <SurfaceImage (20484, 160)>
[SurfaceLabelsMasker.wrapped] Loading regions from <SurfaceImage (20484,)>
[SurfaceLabelsMasker.wrapped] Finished fit
/actions-runner/_work/siibra-python/siibra-python/siibra/volumes/parcellationmap.py:1713: FutureWarning: boolean values for 'standardize' will be deprecated in nilearn 0.15.0.
Use 'zscore_sample' instead of 'True' or use 'None' instead of 'False'.
masker.fit_transform(source, confounds=confounds, sample_mask=sample_mask)
[SurfaceLabelsMasker.wrapped] Extracting region signals
[SurfaceLabelsMasker.wrapped] Cleaning extracted signals
[SurfaceLabelsMasker.wrapped] Loading data from <SurfaceImage (20484, 160)>
[SurfaceLabelsMasker.wrapped] Loading regions from <SurfaceImage (20484,)>
[SurfaceLabelsMasker.wrapped] Finished fit
/actions-runner/_work/siibra-python/siibra-python/siibra/volumes/parcellationmap.py:1713: FutureWarning: boolean values for 'standardize' will be deprecated in nilearn 0.15.0.
Use 'zscore_sample' instead of 'True' or use 'None' instead of 'False'.
masker.fit_transform(source, confounds=confounds, sample_mask=sample_mask)
[SurfaceLabelsMasker.wrapped] Extracting region signals
[SurfaceLabelsMasker.wrapped] Cleaning extracted signals
[SurfaceLabelsMasker.wrapped] Loading data from <SurfaceImage (20484, 160)>
[SurfaceLabelsMasker.wrapped] Loading regions from <SurfaceImage (20484,)>
[SurfaceLabelsMasker.wrapped] Finished fit
/actions-runner/_work/siibra-python/siibra-python/siibra/volumes/parcellationmap.py:1713: FutureWarning: boolean values for 'standardize' will be deprecated in nilearn 0.15.0.
Use 'zscore_sample' instead of 'True' or use 'None' instead of 'False'.
masker.fit_transform(source, confounds=confounds, sample_mask=sample_mask)
[SurfaceLabelsMasker.wrapped] Extracting region signals
[SurfaceLabelsMasker.wrapped] Cleaning extracted signals
[SurfaceLabelsMasker.wrapped] Loading data from <SurfaceImage (20484, 160)>
[SurfaceLabelsMasker.wrapped] Loading regions from <SurfaceImage (20484,)>
[SurfaceLabelsMasker.wrapped] Finished fit
/actions-runner/_work/siibra-python/siibra-python/siibra/volumes/parcellationmap.py:1713: FutureWarning: boolean values for 'standardize' will be deprecated in nilearn 0.15.0.
Use 'zscore_sample' instead of 'True' or use 'None' instead of 'False'.
masker.fit_transform(source, confounds=confounds, sample_mask=sample_mask)
[SurfaceLabelsMasker.wrapped] Extracting region signals
[SurfaceLabelsMasker.wrapped] Cleaning extracted signals
[SurfaceLabelsMasker.wrapped] Loading data from <SurfaceImage (20484, 160)>
[SurfaceLabelsMasker.wrapped] Loading regions from <SurfaceImage (20484,)>
[SurfaceLabelsMasker.wrapped] Finished fit
/actions-runner/_work/siibra-python/siibra-python/siibra/volumes/parcellationmap.py:1713: FutureWarning: boolean values for 'standardize' will be deprecated in nilearn 0.15.0.
Use 'zscore_sample' instead of 'True' or use 'None' instead of 'False'.
masker.fit_transform(source, confounds=confounds, sample_mask=sample_mask)
[SurfaceLabelsMasker.wrapped] Extracting region signals
[SurfaceLabelsMasker.wrapped] Cleaning extracted signals
[SurfaceLabelsMasker.wrapped] Loading data from <SurfaceImage (20484, 160)>
[SurfaceLabelsMasker.wrapped] Loading regions from <SurfaceImage (20484,)>
[SurfaceLabelsMasker.wrapped] Finished fit
/actions-runner/_work/siibra-python/siibra-python/siibra/volumes/parcellationmap.py:1713: FutureWarning: boolean values for 'standardize' will be deprecated in nilearn 0.15.0.
Use 'zscore_sample' instead of 'True' or use 'None' instead of 'False'.
masker.fit_transform(source, confounds=confounds, sample_mask=sample_mask)
[SurfaceLabelsMasker.wrapped] Extracting region signals
[SurfaceLabelsMasker.wrapped] Cleaning extracted signals
[SurfaceLabelsMasker.wrapped] Loading data from <SurfaceImage (20484, 160)>
[SurfaceLabelsMasker.wrapped] Loading regions from <SurfaceImage (20484,)>
[SurfaceLabelsMasker.wrapped] Finished fit
/actions-runner/_work/siibra-python/siibra-python/siibra/volumes/parcellationmap.py:1713: FutureWarning: boolean values for 'standardize' will be deprecated in nilearn 0.15.0.
Use 'zscore_sample' instead of 'True' or use 'None' instead of 'False'.
masker.fit_transform(source, confounds=confounds, sample_mask=sample_mask)
[SurfaceLabelsMasker.wrapped] Extracting region signals
[SurfaceLabelsMasker.wrapped] Cleaning extracted signals
[SurfaceLabelsMasker.wrapped] Loading data from <SurfaceImage (20484, 160)>
[SurfaceLabelsMasker.wrapped] Loading regions from <SurfaceImage (20484,)>
[SurfaceLabelsMasker.wrapped] Finished fit
/actions-runner/_work/siibra-python/siibra-python/siibra/volumes/parcellationmap.py:1713: FutureWarning: boolean values for 'standardize' will be deprecated in nilearn 0.15.0.
Use 'zscore_sample' instead of 'True' or use 'None' instead of 'False'.
masker.fit_transform(source, confounds=confounds, sample_mask=sample_mask)
[SurfaceLabelsMasker.wrapped] Extracting region signals
[SurfaceLabelsMasker.wrapped] Cleaning extracted signals
[SurfaceLabelsMasker.wrapped] Loading data from <SurfaceImage (20484, 228)>
[SurfaceLabelsMasker.wrapped] Loading regions from <SurfaceImage (20484,)>
[SurfaceLabelsMasker.wrapped] Finished fit
/actions-runner/_work/siibra-python/siibra-python/siibra/volumes/parcellationmap.py:1713: FutureWarning: boolean values for 'standardize' will be deprecated in nilearn 0.15.0.
Use 'zscore_sample' instead of 'True' or use 'None' instead of 'False'.
masker.fit_transform(source, confounds=confounds, sample_mask=sample_mask)
[SurfaceLabelsMasker.wrapped] Extracting region signals
[SurfaceLabelsMasker.wrapped] Cleaning extracted signals
[SurfaceLabelsMasker.wrapped] Loading data from <SurfaceImage (20484, 223)>
[SurfaceLabelsMasker.wrapped] Loading regions from <SurfaceImage (20484,)>
[SurfaceLabelsMasker.wrapped] Finished fit
/actions-runner/_work/siibra-python/siibra-python/siibra/volumes/parcellationmap.py:1713: FutureWarning: boolean values for 'standardize' will be deprecated in nilearn 0.15.0.
Use 'zscore_sample' instead of 'True' or use 'None' instead of 'False'.
masker.fit_transform(source, confounds=confounds, sample_mask=sample_mask)
[SurfaceLabelsMasker.wrapped] Extracting region signals
[SurfaceLabelsMasker.wrapped] Cleaning extracted signals
[SurfaceLabelsMasker.wrapped] Loading data from <SurfaceImage (20484, 221)>
[SurfaceLabelsMasker.wrapped] Loading regions from <SurfaceImage (20484,)>
[SurfaceLabelsMasker.wrapped] Finished fit
/actions-runner/_work/siibra-python/siibra-python/siibra/volumes/parcellationmap.py:1713: FutureWarning: boolean values for 'standardize' will be deprecated in nilearn 0.15.0.
Use 'zscore_sample' instead of 'True' or use 'None' instead of 'False'.
masker.fit_transform(source, confounds=confounds, sample_mask=sample_mask)
[SurfaceLabelsMasker.wrapped] Extracting region signals
[SurfaceLabelsMasker.wrapped] Cleaning extracted signals
[SurfaceLabelsMasker.wrapped] Loading data from <SurfaceImage (20484, 220)>
[SurfaceLabelsMasker.wrapped] Loading regions from <SurfaceImage (20484,)>
[SurfaceLabelsMasker.wrapped] Finished fit
/actions-runner/_work/siibra-python/siibra-python/siibra/volumes/parcellationmap.py:1713: FutureWarning: boolean values for 'standardize' will be deprecated in nilearn 0.15.0.
Use 'zscore_sample' instead of 'True' or use 'None' instead of 'False'.
masker.fit_transform(source, confounds=confounds, sample_mask=sample_mask)
[SurfaceLabelsMasker.wrapped] Extracting region signals
[SurfaceLabelsMasker.wrapped] Cleaning extracted signals
[SurfaceLabelsMasker.wrapped] Loading data from <SurfaceImage (20484, 231)>
[SurfaceLabelsMasker.wrapped] Loading regions from <SurfaceImage (20484,)>
[SurfaceLabelsMasker.wrapped] Finished fit
/actions-runner/_work/siibra-python/siibra-python/siibra/volumes/parcellationmap.py:1713: FutureWarning: boolean values for 'standardize' will be deprecated in nilearn 0.15.0.
Use 'zscore_sample' instead of 'True' or use 'None' instead of 'False'.
masker.fit_transform(source, confounds=confounds, sample_mask=sample_mask)
[SurfaceLabelsMasker.wrapped] Extracting region signals
[SurfaceLabelsMasker.wrapped] Cleaning extracted signals
[SurfaceLabelsMasker.wrapped] Loading data from <SurfaceImage (20484, 225)>
[SurfaceLabelsMasker.wrapped] Loading regions from <SurfaceImage (20484,)>
[SurfaceLabelsMasker.wrapped] Finished fit
/actions-runner/_work/siibra-python/siibra-python/siibra/volumes/parcellationmap.py:1713: FutureWarning: boolean values for 'standardize' will be deprecated in nilearn 0.15.0.
Use 'zscore_sample' instead of 'True' or use 'None' instead of 'False'.
masker.fit_transform(source, confounds=confounds, sample_mask=sample_mask)
[SurfaceLabelsMasker.wrapped] Extracting region signals
[SurfaceLabelsMasker.wrapped] Cleaning extracted signals
[SurfaceLabelsMasker.wrapped] Loading data from <SurfaceImage (20484, 217)>
[SurfaceLabelsMasker.wrapped] Loading regions from <SurfaceImage (20484,)>
[SurfaceLabelsMasker.wrapped] Finished fit
/actions-runner/_work/siibra-python/siibra-python/siibra/volumes/parcellationmap.py:1713: FutureWarning: boolean values for 'standardize' will be deprecated in nilearn 0.15.0.
Use 'zscore_sample' instead of 'True' or use 'None' instead of 'False'.
masker.fit_transform(source, confounds=confounds, sample_mask=sample_mask)
[SurfaceLabelsMasker.wrapped] Extracting region signals
[SurfaceLabelsMasker.wrapped] Cleaning extracted signals
[SurfaceLabelsMasker.wrapped] Loading data from <SurfaceImage (20484, 224)>
[SurfaceLabelsMasker.wrapped] Loading regions from <SurfaceImage (20484,)>
[SurfaceLabelsMasker.wrapped] Finished fit
/actions-runner/_work/siibra-python/siibra-python/siibra/volumes/parcellationmap.py:1713: FutureWarning: boolean values for 'standardize' will be deprecated in nilearn 0.15.0.
Use 'zscore_sample' instead of 'True' or use 'None' instead of 'False'.
masker.fit_transform(source, confounds=confounds, sample_mask=sample_mask)
[SurfaceLabelsMasker.wrapped] Extracting region signals
[SurfaceLabelsMasker.wrapped] Cleaning extracted signals
[SurfaceLabelsMasker.wrapped] Loading data from <SurfaceImage (20484, 223)>
[SurfaceLabelsMasker.wrapped] Loading regions from <SurfaceImage (20484,)>
[SurfaceLabelsMasker.wrapped] Finished fit
/actions-runner/_work/siibra-python/siibra-python/siibra/volumes/parcellationmap.py:1713: FutureWarning: boolean values for 'standardize' will be deprecated in nilearn 0.15.0.
Use 'zscore_sample' instead of 'True' or use 'None' instead of 'False'.
masker.fit_transform(source, confounds=confounds, sample_mask=sample_mask)
[SurfaceLabelsMasker.wrapped] Extracting region signals
[SurfaceLabelsMasker.wrapped] Cleaning extracted signals
[SurfaceLabelsMasker.wrapped] Loading data from <SurfaceImage (20484, 216)>
[SurfaceLabelsMasker.wrapped] Loading regions from <SurfaceImage (20484,)>
[SurfaceLabelsMasker.wrapped] Finished fit
/actions-runner/_work/siibra-python/siibra-python/siibra/volumes/parcellationmap.py:1713: FutureWarning: boolean values for 'standardize' will be deprecated in nilearn 0.15.0.
Use 'zscore_sample' instead of 'True' or use 'None' instead of 'False'.
masker.fit_transform(source, confounds=confounds, sample_mask=sample_mask)
[SurfaceLabelsMasker.wrapped] Extracting region signals
[SurfaceLabelsMasker.wrapped] Cleaning extracted signals
top_regions_sorted = (
regional_signal_averages["stopsignal"]
.loc[top_regions, :]
.sort_values("mean", ascending=False)
.index
)
fig, ax = plt.subplots(figsize=(12, 10))
colors = ["r", "g", "b"]
for i, task in enumerate(tasks):
regional_signal_averages[task].drop(columns="mean", inplace=True)
regional_signal_averages[task].loc[top_regions_sorted, :].T.boxplot(
ax=ax, rot=90, label=task, color=colors[i],
flierprops={"marker": ".", "markersize": 1},
)
ax.legend()
ax.set_title("Comparison of top regions between tasks across subjects")
fig.tight_layout()
fs_variant = "pial"
vmin = min([regional_signal_averages[task].min().min() for task in tasks])
vmax = max([regional_signal_averages[task].max().max() for task in tasks])
for task in tasks:
surf_im = julichbrain_fs5.colorize(
regional_signal_averages[task].mean(axis=1),
surface_variant=fs_variant
)
plotting.plot_surf(
surf_map=surf_im,
hemi="both",
cmap="magma",
title=task,
vmin=vmin,
vmax=vmax,
)
[SurfaceLabelsMasker.fit] Loading regions from <SurfaceImage (20484,)>
[SurfaceLabelsMasker.fit] Finished fit
[SurfaceLabelsMasker.inverse_transform] Computing image from signals
/actions-runner/_work/siibra-python/siibra-python/.ci-docs-37911713724-3-notebooks/venv/lib/python3.11/site-packages/numpy/lib/_function_base_impl.py:2599: RuntimeWarning: invalid value encountered in <lambda> (vectorized)
outputs = ufunc(*args, out=...)
[SurfaceLabelsMasker.fit] Loading regions from <SurfaceImage (20484,)>
[SurfaceLabelsMasker.fit] Finished fit
[SurfaceLabelsMasker.inverse_transform] Computing image from signals
/actions-runner/_work/siibra-python/siibra-python/.ci-docs-37911713724-3-notebooks/venv/lib/python3.11/site-packages/numpy/lib/_function_base_impl.py:2599: RuntimeWarning: invalid value encountered in <lambda> (vectorized)
outputs = ufunc(*args, out=...)
[SurfaceLabelsMasker.fit] Loading regions from <SurfaceImage (20484,)>
[SurfaceLabelsMasker.fit] Finished fit
[SurfaceLabelsMasker.inverse_transform] Computing image from signals
/actions-runner/_work/siibra-python/siibra-python/.ci-docs-37911713724-3-notebooks/venv/lib/python3.11/site-packages/numpy/lib/_function_base_impl.py:2599: RuntimeWarning: invalid value encountered in <lambda> (vectorized)
outputs = ufunc(*args, out=...)
Total running time of the script: (20 minutes 22.715 seconds)
Estimated memory usage: 101 MB