NiBabel

Access a cacophony of neuro-imaging file formats

Table Of Contents

Previous topic

eulerangles

Next topic

imageclasses

Reggie -- the one

funcs

Processor functions for images

as_closest_canonical(img[, enforce_diag]) Return img with data reordered to be closest to canonical
concat_images(images[, check_affines, axis]) Concatenate images in list to single image, along specified dimension
four_to_three(img) Create 3D images from 4D image by slicing over last axis
squeeze_image(img) Return image, remove axes length 1 at end of image shape

as_closest_canonical

nibabel.funcs.as_closest_canonical(img, enforce_diag=False)

Return img with data reordered to be closest to canonical

Canonical order is the ordering of the output axes.

Parameters:

img : spatialimage

enforce_diag : {False, True}, optional

If True, before transforming image, check if the resulting image affine will be close to diagonal, and if not, raise an error

Returns:

canonical_img : spatialimage

Version of img where the underlying array may have been reordered and / or flipped so that axes 0,1,2 are those axes in the input data that are, respectively, closest to the output axis orientation. We modify the affine accordingly. If img is already has the correct data ordering, we just return img unmodified.

concat_images

nibabel.funcs.concat_images(images, check_affines=True, axis=None)

Concatenate images in list to single image, along specified dimension

Parameters:

images : sequence

sequence of SpatialImage or filenames of the same dimensionalitys

check_affines : {True, False}, optional

If True, then check that all the affines for images are nearly the same, raising a ValueError otherwise. Default is True

axis : None or int, optional

If None, concatenates on a new dimension. This requires all images to be the same shape. If not None, concatenates on the specified dimension. This requires all images to be the same shape, except on the specified dimension.

Returns :

——- :

concat_img : SpatialImage

New image resulting from concatenating images across last dimension

four_to_three

nibabel.funcs.four_to_three(img)

Create 3D images from 4D image by slicing over last axis

Parameters:

img : image

4D image instance of some class with methods get_data, header and affine, and a class constructor allowing klass(data, affine, header)

Returns:

imgs : list

list of 3D images

squeeze_image

nibabel.funcs.squeeze_image(img)

Return image, remove axes length 1 at end of image shape

For example, an image may have shape (10,20,30,1,1). In this case squeeze will result in an image with shape (10,20,30). See doctests for further description of behavior.

Parameters:

img : SpatialImage

Returns:

squeezed_img : SpatialImage

Copy of img, such that data, and data shape have been squeezed, for dimensions > 3rd, and at the end of the shape list

Examples

>>> import nibabel as nf
>>> shape = (10,20,30,1,1)
>>> data = np.arange(np.prod(shape)).reshape(shape)
>>> affine = np.eye(4)
>>> img = nf.Nifti1Image(data, affine)
>>> img.shape == (10, 20, 30, 1, 1)
True
>>> img2 = squeeze_image(img)
>>> img2.shape == (10, 20, 30)
True

If the data are 3D then last dimensions of 1 are ignored

>>> shape = (10,1,1)
>>> data = np.arange(np.prod(shape)).reshape(shape)
>>> img = nf.ni1.Nifti1Image(data, affine)
>>> img.shape == (10, 1, 1)
True
>>> img2 = squeeze_image(img)
>>> img2.shape == (10, 1, 1)
True

Only final dimensions of 1 are squeezed

>>> shape = (1, 1, 5, 1, 2, 1, 1)
>>> data = data.reshape(shape)
>>> img = nf.ni1.Nifti1Image(data, affine)
>>> img.shape == (1, 1, 5, 1, 2, 1, 1)
True
>>> img2 = squeeze_image(img)
>>> img2.shape == (1, 1, 5, 1, 2)
True