Link to code

Conform an image to a given orientation

Flips and reorder the image data array so that the axes match the directions indicated in orientation. The default RAS orientation corresponds to the first axis being ordered from left to right, the second axis from posterior to anterior, and the third axis from inferior to superior.

For oblique images, the original orientation is considered to be the closest plumb orientation.

No resampling is performed, and thus the output image is not de-obliqued or registered to any other image or template.

The effective transform is calculated from the original affine matrix to the reoriented affine matrix.


If an image is not reoriented, the original file is not modified

>>> import numpy as np
>>> from nipype.interfaces.image import Reorient
>>> reorient = Reorient(orientation='LPS')
>>> reorient.inputs.in_file = 'segmentation0.nii.gz'
>>> res =
>>> res.outputs.out_file
>>> print_affine(np.loadtxt(res.outputs.transform))
1.  0.  0.  0.
0.  1.  0.  0.
0.  0.  1.  0.
0.  0.  0.  1.
>>> reorient.inputs.orientation = 'RAS'
>>> res =
>>> res.outputs.out_file  # doctest: +ELLIPSIS
>>> print_affine(np.loadtxt(res.outputs.transform))
-1.   0.   0.  60.
 0.  -1.   0.  72.
 0.   0.   1.   0.
 0.   0.   0.   1.


in_file: (an existing file name)
        Input image

orientation: ('RAS' or 'RAI' or 'RPS' or 'RPI' or 'LAS' or 'LAI' or
          'LPS' or 'LPI' or 'RSA' or 'RSP' or 'RIA' or 'RIP' or 'LSA' or
          'LSP' or 'LIA' or 'LIP' or 'ARS' or 'ARI' or 'ALS' or 'ALI' or
          'PRS' or 'PRI' or 'PLS' or 'PLI' or 'ASR' or 'ASL' or 'AIR' or
          'AIL' or 'PSR' or 'PSL' or 'PIR' or 'PIL' or 'SRA' or 'SRP' or
          'SLA' or 'SLP' or 'IRA' or 'IRP' or 'ILA' or 'ILP' or 'SAR' or
          'SAL' or 'SPR' or 'SPL' or 'IAR' or 'IAL' or 'IPR' or 'IPL',
          nipype default value: RAS)
        Target axis orientation


out_file: (an existing file name)
        Reoriented image
transform: (an existing file name)
        Affine transform from input orientation to output


Link to code

Rescale an image

Rescales the non-zero portion of in_file to match the bounds of the non-zero portion of ref_file. Reference values in the input and reference images are defined by the percentile parameter, and the reference values in each image are identified and the remaining values are scaled accordingly. In the case of percentile == 0, the reference values are the maxima and minima of each image. If the invert parameter is set, the input file is inverted prior to rescaling.


To use a high-resolution T1w image as a registration target for a T2* image, it may be useful to invert the T1w image and rescale to the T2* range. Using the 1st and 99th percentiles may reduce the impact of outlier voxels.

>>> from nipype.interfaces.image import Rescale
>>> invert_t1w = Rescale(invert=True)
>>> invert_t1w.inputs.in_file = 'structural.nii'
>>> invert_t1w.inputs.ref_file = 'functional.nii'
>>> invert_t1w.inputs.percentile = 1.
>>> res =  # doctest: +SKIP


in_file: (an existing file name)
        Skull-stripped image to rescale
ref_file: (an existing file name)
        Skull-stripped reference image

invert: (a boolean)
        Invert contrast of rescaled image
percentile: (0.0 <= a floating point number <= 50.0, nipype default
          value: 0.0)
        Percentile to use for reference to allow for outliers - 1 indicates
        the 1st and 99th percentiles in the input file will be mapped to the
        99th and 1st percentiles in the reference; 0 indicates minima and
        maxima will be mapped


out_file: (an existing file name)
        Rescaled image