labs.spatial_models.structural_bfls

Module: labs.spatial_models.structural_bfls

Inheritance diagram for nipy.labs.spatial_models.structural_bfls:

Inheritance diagram of nipy.labs.spatial_models.structural_bfls

The main routine of this module implement the LandmarkRegions class, that is used to represent Regions of interest at the population level (in a template space).

This has been used in Thirion et al. Structural Analysis of fMRI Data Revisited: Improving the Sensitivity and Reliability of fMRI Group Studies. IEEE TMI 2007

Author : Bertrand Thirion, 2006-2013

LandmarkRegions

class nipy.labs.spatial_models.structural_bfls.LandmarkRegions(domain, k, indiv_coord, subjects, confidence)

Bases: object

This class is intended to represent a set of inter-subject regions It should inherit from some abstract multiple ROI class, not implemented yet.

__init__(domain, k, indiv_coord, subjects, confidence)

Building the landmark_region

Parameters:

domain: ROI instance

defines the spatial context of the SubDomains

k: int,

the number of landmark regions considered

indiv_coord: k-length list of arrays,

coordinates of the nodes in some embedding space.

subjects: k-length list of integers

these correspond to an ROI feature: the subject index of individual regions

confidence: k-length list of arrays,

confidence values for the regions (0 is low, 1 is high)

centers()

returns the average of the coordinates for each region

kernel_density(k=None, coord=None, sigma=1.0)

Compute the density of a component as a kde

Parameters:

k: int (<= self.k) or None

component upon which the density is computed if None, the sum is taken over k

coord: array of shape(n, self.dom.em_dim), optional

a set of input coordinates

sigma: float, optional

kernel size

Returns:

kde: array of shape(n)

the density sampled at the coords

map_label(coord=None, pval=1.0, sigma=1.0)

Sample the set of landmark regions on the proposed coordiante set cs, assuming a Gaussian shape

Parameters:

coord: array of shape(n,dim), optional,

a set of input coordinates

pval: float in [0,1]), optional

cutoff for the CR, i.e. highest posterior density threshold

sigma: float, positive, optional

spatial scale of the spatial model

Returns:

label: array of shape (n): the posterior labelling

roi_prevalence()

Return a confidence index over the different rois

Returns:

confid: array of shape self.k

the population_prevalence

show()

function to print basic information on self

nipy.labs.spatial_models.structural_bfls.build_landmarks(domain, coords, subjects, labels, confidence=None, prevalence_pval=0.95, prevalence_threshold=0, sigma=1.0)

Given a list of hierarchical ROIs, and an associated labelling, this creates an Amer structure wuch groups ROIs with the same label.

Parameters:

domain: discrete_domain.DiscreteDomain instance,

description of the spatial context of the landmarks

coords: array of shape(n, 3)

Sets of coordinates for the different objects

subjects: array of shape (n), dtype = np.int

indicators of the dataset the objects come from

labels: array of shape (n), dtype = np.int

index of the landmark the object is associated with

confidence: array of shape (n),

measure of the significance of the regions

prevalence_pval: float, optional

prevalence_threshold: float, optional,

(c) A label should be present in prevalence_threshold subjects with a probability>prevalence_pval in order to be valid

sigma: float optional,

regularizing constant that defines a prior on the region extent

Returns:

LR : None or structural_bfls.LR instance

describing a cross-subject set of ROIs. If inference yields a null result, LR is set to None

newlabel: array of shape (n)

a relabelling of the individual ROIs, similar to u, that discards labels that do not fulfill the condition (c)