arrays (e.g. RGB data)

Works around older numpy versions

these return 3D images with the binary or probabilistic mask
in a new image.Image with the name set to the label name
and the same NIFTI header as the atlas

makes the output

resampleToPixdims and resampleToReference

package). Refactored, cleaned up, and enhanced to accept an arbitrary affine
transform - can be used to resample an image into any other space.

alignemnt of the input and resampled images to be adjusted.

winding order, to account for the possibility that different vertex sets
require different orders.

change flag/create a copy themselves

compressed voxelwise EVs

- (..., 3, 3) full diffusion tensors
- (..., 6) unique components
- tuple of eigenvectors & eigenvalues

There should be full test coverage of these new functions,
but coverage.py is currently not working for me, so not 100% sure.

are read-only, so we needd to create copies.

data. Will prpbably need tweaking.

guessing the data type of a file/directory path.

used.

VoxelwiseEV and VoxelwiseConfoundEV classes share a common base class.

this is all handled by nibabel. Also never keeps its own ref to file object.
Image.save always saves to a temp file, then moves that to the destination -
avoid any problems with mmap/indexed_gzip.

reduce the number of series that need to be loaded.