medpy.graphcut.energy_label.regional_atlas

medpy.graphcut.energy_label.regional_atlas(graph, label_image, xxx_todo_changeme1)

Regional term based on a probability atlas.

An implementation of a regional term, suitable to be used with the graph_from_labels function.

This regional term introduces statistical probability of a voxel to belong to the object to segment. It computes the sum of all statistical atlas voxels under each region and uses this value as terminal node weight for the graph cut.

Parameters:

graphGCGraph

The graph to add the weights to.

label_imagendarray

The label image.

probability_mapndarray

The probability atlas image associated with the object to segment.

alphafloat

The energy terms alpha value, balancing between boundary and regional term.

Notes

This function requires a probability atlas image of the same shape as the original image to be passed along. That means that graph_from_labels has to be called with regional_term_args set to the probability atlas image.

This function is tested on 2D and 3D images and theoretically works for all dimensionalities.