Note

This page is a reference documentation. It only explains the class signature, and not how to use it. Please refer to the user guide for the big picture.

nilearn.regions.RegionExtractor

class nilearn.regions.RegionExtractor(maps_img=None, mask_img=None, min_region_size=1350, threshold=1.0, thresholding_strategy='ratio_n_voxels', two_sided=False, extractor='local_regions', smoothing_fwhm=6, standardize=False, standardize_confounds=True, high_variance_confounds=False, detrend=False, low_pass=None, high_pass=None, t_r=None, dtype=None, resampling_target='data', keep_masked_maps=False, memory=None, memory_level=0, verbose=0, reports=True, cmap='CMRmap_r', allow_overlap=True, clean_args=None)[source]

Class for brain region extraction.

Region Extraction is a post processing technique which is implemented to automatically segment each brain atlas maps into different set of separated brain activated region. Particularly, to show that each decomposed brain maps can be used to focus on a target specific Regions of Interest analysis.

See Abraham et al.[1].

Added in Nilearn 0.2.

Parameters:
maps_img4D Niimg-like object or None, default=None

Image containing a set of whole brain atlas maps or statistically decomposed brain maps.

mask_imgNiimg-like object or None, optional

Mask to be applied to input data, passed to NiftiMapsMasker. If None, no masking is applied.

min_region_sizefloat, default=1350

Minimum volume in mm3 for a region to be kept. For example, if the voxel size is 3x3x3 mm then the volume of the voxel is 27mm^3. The default of 1350mm^3 means we take minimum size of 1350 / 27 = 50 voxels.

thresholdnumber, default=1.0

A value used either in ratio_n_voxels or img_value or percentile thresholding_strategy based upon the choice of selection.

thresholding_strategystr {‘ratio_n_voxels’, ‘img_value’, ‘percentile’}, default=’ratio_n_voxels’

If default ‘ratio_n_voxels’, we apply thresholding that will keep the more intense nonzero brain voxels (denoted as n_voxels) across all maps (n_voxels being the number of voxels in the brain volume). A float value given in threshold parameter indicates the ratio of voxels to keep meaning (if float=2. then maps will together have 2. x n_voxels non-zero voxels). If set to ‘percentile’, images are thresholded based on the score obtained with the given percentile on the data and the voxel intensities which are survived above this obtained score will be kept. If set to ‘img_value’, we apply thresholding based on the non-zero voxel intensities across all maps. A value given in threshold parameter indicates that we keep only those voxels which have intensities more than this value.

two_sidedbool, default=False

Whether the thresholding should yield both positive and negative part of the maps.

Added in Nilearn 0.11.1.

extractor{“local_regions”, “connected_components”}, default=”local_regions”

This option can take two values:

  • "connected_components": each component/region in the image is extracted automatically by labeling each region based upon the presence of unique features in their respective regions.

  • "local_regions": each component/region is extracted based on their maximum peak value to define a seed marker and then using random walker segmentation algorithm on these markers for region separation.

smoothing_fwhmfloat or int or None, optional.

If smoothing_fwhm is not None, it gives the full-width at half maximum in millimeters of the spatial smoothing to apply to the signal. Use this parameter to smooth an image to extract most sparser regions.

Note

This parameter is passed to nilearn.regions.connected_regions. It will be used only if extractor='local_regions'.

Note

Please set this parameter according to maps resolution, otherwise extraction will fail.

Default=6mm.

standardizeany of: ‘zscore_sample’, ‘zscore’, ‘psc’, True, False or None; default=False

Strategy to standardize the signal:

  • 'zscore_sample': The signal is z-scored. Timeseries are shifted to zero mean and scaled to unit variance. Uses sample std.

  • 'zscore': The signal is z-scored. Timeseries are shifted to zero mean and scaled to unit variance. Uses population std by calling default numpy.std with N - ddof=0.

    Deprecated since Nilearn 0.10.1: This option will be removed in Nilearn version 0.14.0. Use zscore_sample instead.

  • 'psc': Timeseries are shifted to zero mean value and scaled to percent signal change (as compared to original mean signal).

  • True: The signal is z-scored (same as option zscore). Timeseries are shifted to zero mean and scaled to unit variance.

    Deprecated since Nilearn 0.13.0dev: In nilearn version 0.15.0, True will be replaced by 'zscore_sample'.

  • False: Do not standardize the data.

    Deprecated since Nilearn 0.13.0dev: In nilearn version 0.15.0, False will be replaced by None.

Deprecated since Nilearn 0.15.0dev: The default will be changed to None in version 0.15.0.

Note

Recommended to set to True if signals are not already standardized. Passed to NiftiMapsMasker.

standardize_confoundsbool, default=True

If set to True, the confounds are z-scored: their mean is put to 0 and their variance to 1 in the time dimension.

high_variance_confoundsbool, default=False

If True, high variance confounds are computed on provided image with nilearn.image.high_variance_confounds and default parameters and regressed out.

detrendbool, optional

Whether to detrend signals or not.

Note

Passed to nilearn.signal.clean.

Default=False.

low_passfloat or int or None, default=None

Low cutoff frequency in Hertz. If specified, signals above this frequency will be filtered out. If None, no low-pass filtering will be performed.

Note

Passed to nilearn.signal.clean.

high_passfloat or int or None, default=None

High cutoff frequency in Hertz. If specified, signals below this frequency will be filtered out.

Note

Passed to nilearn.signal.clean.

t_rfloat or int or None, default=None

Repetition time, in seconds (sampling period). Set to None if not provided.

Note

Passed to nilearn.signal.clean.

dtypedtype like, “auto” or None, default=None

Data type toward which the data should be converted. If “auto”, the data will be converted to int32 if dtype is discrete and float32 if it is continuous.

resampling_target{“data”, “mask”, “maps”, None}, default=”data”

Defines which image gives the final shape/size.

  • "data" means that the atlas is resampled to the shape of the data if needed

  • "mask" means that the maps_img and images provided to fit() are resampled to the shape and affine of mask_img

  • "maps" means the mask_img and images provided to fit() are resampled to the shape and affine of maps_img

  • None means no resampling: if shapes and affines do not match, a ValueError is raised.

keep_masked_mapsbool, optional

If True, masked atlas with invalid maps (maps that contain only zeros after applying the mask) will be retained in the output, resulting in corresponding time series containing zeros only. If False, the invalid maps will be removed from the trimmed atlas, resulting in no empty time series in the output.

Deprecated since Nilearn 0.10.2.

Changed in Nilearn 0.13.0dev: The keep_masked_maps parameter will be removed in 0.15.

memoryNone, instance of joblib.Memory, str, or pathlib.Path

Used to cache the masking process. By default, no caching is done. If a str is given, it is the path to the caching directory.

memory_levelint, default=0

Rough estimator of the amount of memory used by caching. Higher value means more memory for caching. Zero means no caching.

verboseint, default=0

Verbosity level (0 means no message).

reportsbool, default=True

If set to True, data is saved in order to produce a report.

cmapmatplotlib.colors.Colormap, or str, optional

The colormap to use. Either a string which is a name of a matplotlib colormap, or a matplotlib colormap object. default=”CMRmap_r” Only relevant for the report figures.

allow_overlapTrue

If False, an error is raised if the maps overlaps (ie at least two maps have a non-zero value for the same voxel).

clean_argsdict or None, default=None

Keyword arguments to be passed to clean called within the masker. Within clean, kwargs prefixed with 'butterworth__' will be passed to the Butterworth filter.

Added in Nilearn 0.12.1.

Attributes:
clean_args_dict

Keyword arguments to be passed to clean called within the masker. Within clean, kwargs prefixed with 'butterworth__' will be passed to the Butterworth filter.

index_numpy.ndarray

Array of list of indices where each index value is assigned to each separate region of its corresponding family of brain maps.

maps_img_nibabel.nifti1.Nifti1Image

The maps mask of the data.

mask_img_A 3D binary nibabel.nifti1.Nifti1Image or None.

The mask of the data. If no mask_img was passed at masker construction, then mask_img_ is None, otherwise is the resulting binarized version of mask_img where each voxel is True if all values across samples (for example across timepoints) is finite value different from 0.

memory_joblib memory cache
n_elements_int

The number of overlapping maps in the mask. This is equivalent to the number of volumes in the mask image.

Added in Nilearn 0.9.2.

regions_img_nibabel.nifti1.Nifti1Image

List of separated regions with each region lying on an original volume concatenated into a 4D image.

See also

nilearn.regions.connected_label_regions

A function can be readily used for extraction of regions on labels based atlas images.

References

__init__(maps_img=None, mask_img=None, min_region_size=1350, threshold=1.0, thresholding_strategy='ratio_n_voxels', two_sided=False, extractor='local_regions', smoothing_fwhm=6, standardize=False, standardize_confounds=True, high_variance_confounds=False, detrend=False, low_pass=None, high_pass=None, t_r=None, dtype=None, resampling_target='data', keep_masked_maps=False, memory=None, memory_level=0, verbose=0, reports=True, cmap='CMRmap_r', allow_overlap=True, clean_args=None)[source]
fit(imgs=None, y=None)[source]

Prepare signal extraction from regions.

Parameters:
imgslist of Niimg-like objects or None, default=None

See Input and output: neuroimaging data representation. Image data passed to the reporter.

yNone

This parameter is unused. It is solely included for scikit-learn compatibility.

fit_transform(imgs, y=None, confounds=None, sample_mask=None)[source]

Prepare and perform signal extraction.

Parameters:
imgs3D/4D Niimg-like object

See Input and output: neuroimaging data representation. Images to process. If a 3D niimg is provided, a 1D array is returned.

yNone

This parameter is unused. It is solely included for scikit-learn compatibility.

confoundsnumpy.ndarray, str, pathlib.Path, pandas.DataFrame or list of confounds timeseries, default=None

This parameter is passed to nilearn.signal.clean. Please see the related documentation for details. shape: (number of scans, number of confounds)

sample_maskAny type compatible with numpy-array indexing, default=None

shape = (total number of scans - number of scans removed) for explicit index (for example, sample_mask=np.asarray([1, 2, 4])), or shape = (number of scans) for binary mask (for example, sample_mask=np.asarray([False, True, True, False, True])). Masks the images along the last dimension to perform scrubbing: for example to remove volumes with high motion and/or non-steady-state volumes. This parameter is passed to nilearn.signal.clean.

Added in Nilearn 0.8.0.

Returns:
signalsnumpy.ndarray, pandas.DataFrame or polars.DataFrame

Signal for each element.

Changed in Nilearn 0.13.0dev: Added set_output support.

The type of the output is determined by set_output(): see the scikit-learn documentation.

Output shape for :

  • For Numpy outputs:

    • 3D images: (number of elements,)

    • 4D images: (number of scans, number of elements) array

  • For DataFrame outputs:

    • 3D or 4D images: (number of scans, number of elements) array

generate_report(displayed_maps=10)[source]

Generate an HTML report for the current NiftiMapsMasker object.

Note

This functionality requires to have Matplotlib installed.

Parameters:
displayed_mapsint, or list, or ndarray, or “all”, default=10

Indicates which maps will be displayed in the HTML report.

  • If "all": All maps will be displayed in the report.

masker.generate_report("all")
  • If a list or ndarray: This indicates

    the indices of the maps to be displayed in the report. For example, the following code will generate a report with maps 6, 3, and 12, displayed in this specific order:

masker.generate_report([6, 3, 12])
  • If an int: This will only display the first n maps,

    n being the value of the parameter. By default, the report will only contain the first 10 maps. Example to display the first 16 maps:

masker.generate_report(16)
Returns:
reportnilearn.reporting.html_report.HTMLReport

HTML report for the masker.

get_feature_names_out(input_features=None)

Get output feature names for transformation.

The feature names out will prefixed by the lowercased class name. For example, if the transformer outputs 3 features, then the feature names out are: [“class_name0”, “class_name1”, “class_name2”].

Parameters:
input_featuresarray-like of str or None, default=None

Only used to validate feature names with the names seen in fit.

Returns:
feature_names_outndarray of str objects

Transformed feature names.

get_metadata_routing()

Get metadata routing of this object.

Please check User Guide on how the routing mechanism works.

Returns:
routingMetadataRequest

A MetadataRequest encapsulating routing information.

get_params(deep=True)

Get parameters for this estimator.

Parameters:
deepbool, default=True

If True, will return the parameters for this estimator and contained subobjects that are estimators.

Returns:
paramsdict

Parameter names mapped to their values.

inverse_transform(region_signals)[source]

Compute voxel signals from region signals.

Any mask given at initialization is taken into account.

Parameters:
signals1D/2D numpy.ndarray

Extracted signal. If a 1D array is provided, then the shape should be (number of elements,). If a 2D array is provided, then the shape should be (number of scans, number of elements).

Returns:
imgnibabel.nifti1.Nifti1Image

Transformed image in brain space. Output shape for :

See Input and output: neuroimaging data representation.

set_fit_request(*, imgs='$UNCHANGED$')

Configure whether metadata should be requested to be passed to the fit method.

Note that this method is only relevant when this estimator is used as a sub-estimator within a meta-estimator and metadata routing is enabled with enable_metadata_routing=True (see sklearn.set_config). Please check the User Guide on how the routing mechanism works.

The options for each parameter are:

  • True: metadata is requested, and passed to fit if provided. The request is ignored if metadata is not provided.

  • False: metadata is not requested and the meta-estimator will not pass it to fit.

  • None: metadata is not requested, and the meta-estimator will raise an error if the user provides it.

  • str: metadata should be passed to the meta-estimator with this given alias instead of the original name.

The default (sklearn.utils.metadata_routing.UNCHANGED) retains the existing request. This allows you to change the request for some parameters and not others.

Added in version 1.3.

Parameters:
imgsstr, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED

Metadata routing for imgs parameter in fit.

Returns:
selfobject

The updated object.

set_inverse_transform_request(*, region_signals='$UNCHANGED$')

Configure whether metadata should be requested to be passed to the inverse_transform method.

Note that this method is only relevant when this estimator is used as a sub-estimator within a meta-estimator and metadata routing is enabled with enable_metadata_routing=True (see sklearn.set_config). Please check the User Guide on how the routing mechanism works.

The options for each parameter are:

  • True: metadata is requested, and passed to inverse_transform if provided. The request is ignored if metadata is not provided.

  • False: metadata is not requested and the meta-estimator will not pass it to inverse_transform.

  • None: metadata is not requested, and the meta-estimator will raise an error if the user provides it.

  • str: metadata should be passed to the meta-estimator with this given alias instead of the original name.

The default (sklearn.utils.metadata_routing.UNCHANGED) retains the existing request. This allows you to change the request for some parameters and not others.

Added in version 1.3.

Parameters:
region_signalsstr, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED

Metadata routing for region_signals parameter in inverse_transform.

Returns:
selfobject

The updated object.

set_output(*, transform=None)

Set output container.

See Introducing the set_output API for an example on how to use the API.

Parameters:
transform{“default”, “pandas”, “polars”}, default=None

Configure output of transform and fit_transform.

  • “default”: Default output format of a transformer

  • “pandas”: DataFrame output

  • “polars”: Polars output

  • None: Transform configuration is unchanged

Added in version 1.4: “polars” option was added.

Returns:
selfestimator instance

Estimator instance.

set_params(**params)

Set the parameters of this estimator.

The method works on simple estimators as well as on nested objects (such as Pipeline). The latter have parameters of the form <component>__<parameter> so that it’s possible to update each component of a nested object.

Parameters:
**paramsdict

Estimator parameters.

Returns:
selfestimator instance

Estimator instance.

set_transform_request(*, confounds='$UNCHANGED$', imgs='$UNCHANGED$', sample_mask='$UNCHANGED$')

Configure whether metadata should be requested to be passed to the transform method.

Note that this method is only relevant when this estimator is used as a sub-estimator within a meta-estimator and metadata routing is enabled with enable_metadata_routing=True (see sklearn.set_config). Please check the User Guide on how the routing mechanism works.

The options for each parameter are:

  • True: metadata is requested, and passed to transform if provided. The request is ignored if metadata is not provided.

  • False: metadata is not requested and the meta-estimator will not pass it to transform.

  • None: metadata is not requested, and the meta-estimator will raise an error if the user provides it.

  • str: metadata should be passed to the meta-estimator with this given alias instead of the original name.

The default (sklearn.utils.metadata_routing.UNCHANGED) retains the existing request. This allows you to change the request for some parameters and not others.

Added in version 1.3.

Parameters:
confoundsstr, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED

Metadata routing for confounds parameter in transform.

imgsstr, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED

Metadata routing for imgs parameter in transform.

sample_maskstr, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED

Metadata routing for sample_mask parameter in transform.

Returns:
selfobject

The updated object.

transform(imgs, confounds=None, sample_mask=None)[source]

Apply mask, spatial and temporal preprocessing.

Parameters:
imgs3D/4D Niimg-like object

See Input and output: neuroimaging data representation. Images to process. If a 3D niimg is provided, a 1D array is returned.

confoundsnumpy.ndarray, str, pathlib.Path, pandas.DataFrame or list of confounds timeseries, default=None

This parameter is passed to nilearn.signal.clean. Please see the related documentation for details. shape: (number of scans, number of confounds)

sample_maskAny type compatible with numpy-array indexing, default=None

shape = (total number of scans - number of scans removed) for explicit index (for example, sample_mask=np.asarray([1, 2, 4])), or shape = (number of scans) for binary mask (for example, sample_mask=np.asarray([False, True, True, False, True])). Masks the images along the last dimension to perform scrubbing: for example to remove volumes with high motion and/or non-steady-state volumes. This parameter is passed to nilearn.signal.clean.

Added in Nilearn 0.8.0.

Returns:
signalsnumpy.ndarray, pandas.DataFrame or polars.DataFrame

Signal for each element.

Changed in Nilearn 0.13.0dev: Added set_output support.

The type of the output is determined by set_output(): see the scikit-learn documentation.

Output shape for :

  • For Numpy outputs:

    • 3D images: (number of elements,)

    • 4D images: (number of scans, number of elements) array

  • For DataFrame outputs:

    • 3D or 4D images: (number of scans, number of elements) array

transform_single_imgs(imgs, confounds=None, sample_mask=None)[source]

Extract signals from a single 4D niimg.

Parameters:
imgs3D/4D Niimg-like object

See Input and output: neuroimaging data representation. Images to process.

confoundsCSV file or array-like, default=None

This parameter is passed to nilearn.signal.clean. Please see the related documentation for details. shape: (number of scans, number of confounds)

sample_maskAny type compatible with numpy-array indexing, default=None

shape = (total number of scans - number of scans removed) for explicit index (for example, sample_mask=np.asarray([1, 2, 4])), or shape = (number of scans) for binary mask (for example, sample_mask=np.asarray([False, True, True, False, True])). Masks the images along the last dimension to perform scrubbing: for example to remove volumes with high motion and/or non-steady-state volumes. This parameter is passed to nilearn.signal.clean.

Added in Nilearn 0.8.0.

Returns:
signalsnumpy.ndarray, pandas.DataFrame or polars.DataFrame

Signal for each element.

Changed in Nilearn 0.13.0dev: Added set_output support.

The type of the output is determined by set_output(): see the scikit-learn documentation.

Output shape for :

  • For Numpy outputs:

    • 3D images: (number of elements,)

    • 4D images: (number of scans, number of elements) array

  • For DataFrame outputs:

    • 3D or 4D images: (number of scans, number of elements) array

Examples using nilearn.regions.RegionExtractor

Regions extraction using dictionary learning and functional connectomes

Regions extraction using dictionary learning and functional connectomes

Regions Extraction of Default Mode Networks using Smith Atlas

Regions Extraction of Default Mode Networks using Smith Atlas