# Between-volumes Motion Correction on DWI datasets

## Overview

During a dMRI acquisition, the subject motion inevitable. This motion implies a misalignment between N volumes on a dMRI dataset. A common way to solve this issue is to perform a registration on each acquired volume to a reference b = 0. [JenkinsonSmith01]

This preprocessing is an highly recommended step that should be executed before any dMRI dataset analysis.

Let’s import some essential functions.

from dipy.align import motion_correction
from dipy.data import get_fnames


We choose one of the data from the datasets in dipy. However, you can replace the following line with the path of your image.

dwi_fname, dwi_bval_fname, dwi_bvec_fname = get_fnames('sherbrooke_3shell')


We load the image and the affine of the image. The affine is the transformation matrix which maps image coordinates to world (mm) coordinates. We also load the b-values and b-vectors.

data, affine = load_nifti(dwi_fname)


This data has 193 volumes. For this demo purpose, we decide to reduce the number of volumes to 5. However, we do not recommended to perform a motion correction with less than 10 volumes.

data_small = data[..., 5]
bvals_small = bvals[5]
bvecs_small = bvecs[5, ...]


Start motion correction of our reduced DWI dataset(between-volumes motion correction).

data_corrected, reg_afines = motion_correction(data, gtab, affine)


Save our DWI dataset corrected to a new Nifti file.

save_nifti('motion_correction.nii.gz', data_corrected.get_fdata(), affine)


## References

Jenkinson, M., Smith, S., 2001. A global optimisation method for robust affine registration of brain images. Med Image Anal 5 (2), 143–56.

Example source code

You can download the full source code of this example. This same script is also included in the dipy source distribution under the doc/examples/ directory.