cedalion.geometry.segmentation

Funtionality to work with segmented MRI scans.

Functions

binaryMaskGenerator(d)

cell_coordinates(volume[, flat])

Generate cell coordinates from a 3D volume.

segmentation_postprocessing(segmentation_dir)

Postprocessing of the segmented SPM12 MRI segmentation files.

sizeOfObject(img[, conn])

surface_from_segmentation(segmentation_mask, ...)

Create a surface from a segmentation mask.

voxels_from_segmentation(segmentation_mask, ...)

Generate voxels from a segmentation mask.
cedalion.geometry.segmentation.voxels_from_segmentation(
segmentation_mask: DataArray,
segmentation_types: List[str],
isovalue=0.9,
fill_holes_in_mask=False,
) → Voxels[source]

Generate voxels from a segmentation mask.

Parameters:

  • segmentation_mask – xr.DataArray Segmentation mask.

  • segmentation_types – List[str] List of segmentation types.

  • isovalue – float, optional Isovalue for marching cubes, by default 0.9.

  • fill_holes_in_mask – bool, optional Fill holes in the mask, by default False.

Returns:

cdc.Voxels

Voxels in voxel space.

cedalion.geometry.segmentation.surface_from_segmentation(
segmentation_mask: DataArray,
segmentation_types: List[str],
isovalue=0.9,
fill_holes_in_mask=False,
) → Surface[source]

Create a surface from a segmentation mask.

Parameters:

  • segmentation_mask (xr.DataArray) – Segmentation mask with dimensions segmentation type, i, j, k.

  • segmentation_types (List[str]) – A list of segmentation types to include in the surface.

  • isovalue (Float) – The isovalue to use for the marching cubes algorithm.

  • fill_holes_in_mask (Bool) – Whether to fill holes in the mask before creating the surface.

Returns:

A cedalion.Surface object.

cedalion.geometry.segmentation.cell_coordinates(volume, flat: bool = False)[source]

Generate cell coordinates from a 3D volume.

Parameters:

  • volume (np.ndarray) – 3D volume.

  • flat (bool, optional) – If True, return coordinates as a flat array, by default False.

  • Returns

  • -------

  • xr.DataArray – Cell coordinates in voxel space.

cedalion.geometry.segmentation.segmentation_postprocessing(
segmentation_dir: str,
mask_files: dict[str, str] = {'air': 'c6.nii', 'bone': 'c4.nii', 'csf': 'c3.nii', 'gray': 'c1.nii', 'skin': 'c5.nii', 'white': 'c2.nii'},
isSmooth: bool = True,
fixCSF: bool = True,
removeDisconnected: bool = True,
labelUnassigned: bool = True,
removeAir: bool = True,
subtractTissues: bool = True,
) → dict[source]

Postprocessing of the segmented SPM12 MRI segmentation files.

Parameters:

  • segmentation_dir (str) – Directory where the segmented files are stored.

  • mask_files (dict[str, str], optional) – Dictionary containing the filenames of the segmented tissues.

  • isSmooth (bool, optional) – Smooth the segmented tissues using Gaussian filter.

  • fixCSF (bool, optional) – Fix the CSF continuity.

  • removeDisconnected (bool, optional) – Remove disconnected voxels.

  • labelUnassigned (bool, optional) – Label empty voxels to the nearest tissue type.

  • removeAir (bool, optional) – Remove air cavities.

  • subtractTissues (bool, optional) – Subtract tissues from each others

Returns:

mask_filesdict

Dictionary containing the filenames of the postprocessed masks.

References:

This whole postprocessing is based on the following references: Huang et al. [HDS+13] Harmening and Miklody [HM22]

cedalion.geometry.segmentation.binaryMaskGenerator(d)[source]
cedalion.geometry.segmentation.sizeOfObject(img, conn=None)[source]