conn_comp

conn_comp(
    x,
    threshold=0,
    cluster_table=True,
    local_maxima=True,
    local_maxima_dist=15,
    connect='26-connect',
    *,
    mask=None,
    two_tailed=False,
)

Find connected components in an image.

Identifies and labels spatially connected regions in neuroimaging data, supporting both binary masks and thresholded volumes.

Parameters

Name Type Description Default
x NeuroVol The image object. required
threshold float Threshold defining lower intensity bound for the binary mask. Default is 0. See two_tailed for two-sided thresholding. 0
cluster_table bool Whether to return the cluster statistics table. Default True. True
local_maxima bool Whether to compute local maxima within clusters. Default True. True
local_maxima_dist float Minimum distance between local maxima in mm. Default 15. 15
connect str Connectivity pattern: “26-connect”, “18-connect”, or “6-connect”. Default “26-connect”. '26-connect'
mask LogicalNeuroVol, optional (keyword only) Restrict clustering to voxels where mask is True. When supplied, verify.assert_same_space(x, mask) is invoked first so a foreign-affine mask raises before any clustering. None
two_tailed bool, optional (keyword only) When True, threshold the absolute value (\|x.data\| > threshold) so both positive- and negative-tail clusters survive. peak_value in the cluster table is read back from the original signed data so the sign of each cluster is retained. Default False (one-tailed x.data > threshold, the legacy behaviour). False

Returns

Name Type Description
ConnCompResult index (ClusteredNeuroVol), size (NeuroVol), voxels, cluster_table, local_maxima, and provenance (populated with method_name="conn_comp" and the input space + mask hashes).
Back to top