.. _scil_lesions_harmonize_labels:

scil_lesions_harmonize_labels
=============================

::

	usage: __main__.py [-h] [--max_adjacency MAX_ADJACENCY]
	                   [--min_voxel_overlap MIN_VOXEL_OVERLAP]
	                   [--incremental_lesions] [--debug_mode] [-f]
	                   in_images [in_images ...] out_dir
	
	This script harmonizes labels across a set of lesion files represented in
	NIfTI format. It ensures that labels are consistent across multiple input
	images by matching labels between images based on spatial proximity and
	overlap criteria.
	
	The script works iteratively, so the multiple inputs should be in chronological
	order (and changing the order affects the output). All images should be
	co-registered.
	
	To obtain labels from binary mask use scil_labels_from_mask.
	
	WARNING: this script requires all files to have all lesions segmented.
	If your data only show new lesions at each timepoints (common in manual
	segmentation), use the option --incremental_lesions to merge past timepoints.
	    T1 = T1, T2 = T1 + T2, T3 = T1 + T2 + T3
	
	positional arguments:
	  in_images             Input file name, in nifti format.
	  out_dir               Output directory.
	
	options:
	  -h, --help            show this help message and exit
	  --max_adjacency MAX_ADJACENCY
	                        Maximum adjacency distance between lesions for them to be considered as the potential match [5.0].
	  --min_voxel_overlap MIN_VOXEL_OVERLAP
	                        Minimum number of overlapping voxels between lesions for them to be considered as the potential match [1].
	  --incremental_lesions
	                        If lesions files only show new lesions at each timepoint, this will merge past timepoints.
	  --debug_mode          Add a fake voxel to the corner to ensure consistent colors in MI-Brain.
	  -f                    Force overwriting of the output files.