mcmodels.models.voxel.RegionalizedModel¶

class mcmodels.models.voxel.RegionalizedModel(weights, nodes, source_key, target_key, ordering=None, dataframe=False)[source]¶

Regionalization/Parcelation of VoxelModel.

Regionalizes the connectivity model in VoxelModel given a brain parcelation.

Parameters:

source_key : array-like, shape=(n_source_voxels,): Flattened key relating each source voxel to a given brain region.
target_key : array-like, shape=(n_target_voxels,): Flattened key relating each target voxel to a given brain region.
ordering : array-like, optional (default=None): Order with which to arrange the source/target regions. If supplied, the ordering must contain at least every unique structure_id associated with each of the source/target regions.
dataframe : boolean, optional (default=False): If True, each metric of the regionalized model will be returned as a labeled pandas dataframe. Else, each metric will be returned as an unlabeled numpy array.

Examples

>>> from mcmodels.core import VoxelModelCache
>>> from mcmodels.models.voxel import RegionalizedModel
>>> cache = VoxelModelCache()
>>> # pull voxel-scale model from cache
>>> voxel_array, source_mask, target_mask = cache.get_voxel_connectivity_array()
>>> # regionalize to summary structures (region 934 was removed in new ccf)
>>> regions = cache.get_structures_by_set_id(
>>> region_ids = [r['id'] for r in regions if r['id'] != 934]
>>> # get array keys
>>> source_key = source_mask.get_key(region_ids)
>>> target_key = source_mask.get_key(region_ids)
>>> # regionalize model
>>> regional_model = RegionalizedModel.from_voxel_array(
...     voxel_array, source_key, target_key)
>>> regional_model.normalized_connection_density.shape
(291, 577)

Attributes:	`connection_density` \(w_{ij} / \|Y\|\) `connection_strength` \(w_{ij}\) `normalized_connection_density` \(w_{ij} / (\|X\| \|Y\|)\) `normalized_connection_strength` \(w_{ij} / \|X\|\)

Methods

`from_voxel_array`(voxel_array, \args, \\*kwargs)	Alternative constructor.
`predict`(self, X[, normalize])	Predict regional projection.

__init__(self, weights, nodes, source_key, target_key, ordering=None, dataframe=False)[source]¶

Methods

`__init__`(self, weights, nodes, source_key, …)
`from_voxel_array`(voxel_array, \args, \\*kwargs)	Alternative constructor.
`predict`(self, X[, normalize])	Predict regional projection.

connection_density¶

\(w_{ij} / |Y|\)

The average voxel-scale connectivity between each source voxel to each source region.

connection_strength¶

\(w_{ij}\)

The sum of the voxel-scale connectivity between each pair of source-target regions.

classmethod from_voxel_array(voxel_array, *args, **kwargs)[source]¶

Alternative constructor.

Parameters:

voxel_array : VoxelConnectivityArray object: The voxel-scale model in the form of a VoxelConnectivityArray object.
source_key : array-like, shape=(n_source_voxels,): Flattened key relating each source voxel to a given brain region.
target_key : array-like, shape=(n_target_voxels,): Flattened key relating each target voxel to a given brain region.
ordering : array-like, optional (default=None): Order with which to arrange the source/target regions. If supplied, the ordering must contain at least every unique structure_id associated with each of the source/target regions.
dataframe : boolean, optional (default=False): If True, each metric of the regionalized model will be returned as a labeled pandas dataframe. Else, each metric will be returned as an unlabeled numpy array.

Returns:

An instantiated RegionalizedModel object.

normalized_connection_density¶

\(w_{ij} / (|X| |Y|)\)

The average voxel-scale connectivity between each pair of source-target regions

normalized_connection_strength¶

\(w_{ij} / |X|\)

The average voxel-scale connectivity between each source region to each target voxel”

predict(self, X, normalize=False)[source]¶: Predict regional projection.