unravel.coordinates.labeled_voxel_img_to_points module#
Use coords_labeled_voxel_img_to_points or labeled_voxel_img_to_points from UNRAVEL
to decode a labeled voxel image back into a table of label coordinates.
This is useful after a unique-voxel labeled image has been reoriented, resampled, or warped into another space, e.g. MERFISH-CCF space.
- Input:
labeled .nii.gz image where each nonzero value is a voxel_label
optional mapping CSV from
points_to_labeled_voxel_imglinking voxel_label to original metadata
- Output:
CSV with one row per voxel_label and coordinates in the current image space
Note
If warping creates multiple voxels for the same label, coordinates are summarized by centroid, median, and mode-like first voxel.
Use centroid_x/y/z for sphere-mask creation unless you prefer an integer voxel coordinate. The rounded centroid columns are also provided.
Usage:#
- labeled_voxel_img_to_points
-i unique_voxels_labeled_MERFISH.nii.gz -map unique_voxels/units__unique_voxel_mapping.csv -o unique_voxels_MERFISH_coords.csv -v
- unravel.coordinates.labeled_voxel_img_to_points.summarize_label_coords(label, coords, mode='centroid_rounded')[source]#
Summarize coordinates for one label and return a dict.
Parameters:#
- labelint
The voxel label being summarized.
- coordsndarray
An Nx3 array of voxel coordinates corresponding to the given label.
- modestr
The method for summarizing coordinates. Options are: - ‘centroid_rounded’: Compute the centroid and round to the nearest integer voxel coordinate - ‘centroid_float’: Compute the centroid and keep as float (physical space) - ‘median’: Compute the median coordinate across all voxels - ‘first’: Take the coordinates of the first voxel (mode-like)
Returns:#
- dict
A dictionary for a row with keys ‘voxel_label’, ‘x’, ‘y’, ‘z’, and ‘n_voxels_after_warp’.