modelarrayio.utils.mif_image.MifHeader

class modelarrayio.utils.mif_image.MifHeader(shape: tuple = (1,), zooms: tuple | None = None, layout: list[int] | None = None, dtype: dtype | None = None, transform: ndarray | None = None, intensity_offset: float = 0.0, intensity_scale: float = 1.0, keyval: dict | None = None)

Header for MIF (.mif / .mif.gz) image files.

The MIF format uses a text header with key: value pairs followed by END, then binary image data at the byte offset given by the file: . <offset> entry.

The transform stored in the file contains unit direction cosines for each voxel axis; voxel sizes are stored separately in the vox field. The nibabel 4x4 affine is reconstructed as:

affine[:3, :3] = transform[:3, :3] * zooms   # column-wise scaling
affine[:3,  3] = transform[:3, 3]             # translation unchanged

__init__(shape: tuple = (1,), zooms: tuple | None = None, layout: list[int] | None = None, dtype: dtype | None = None, transform: ndarray | None = None, intensity_offset: float = 0.0, intensity_scale: float = 1.0, keyval: dict | None = None) → None

Methods

__init__([shape, zooms, layout, dtype, ...])
copy()	Copy object to independent representation
from_fileobj(fileobj)	Read a MIF header from a binary file-like object.
from_header([header])
get_best_affine()	Return the 4x4 affine mapping canonical voxel indices to scanner space (mm).
get_data_dtype()
get_data_shape()
get_intensity_scaling()	Return (offset, scale) for intensity rescaling.
get_keyval()
get_layout()
get_transform()	Return a copy of the 3x4 direction-cosine + translation matrix.
get_zooms()
set_data_dtype(dtype)
set_data_shape(shape)
set_zooms(zooms)
write_to(fileobj[, data_offset])	Write the MIF header to fileobj, returning the data byte offset.