Between-volumes Motion Correction on DWI datasets

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 [1].

This preprocessing is a 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.core.gradients import gradient_table
from dipy.data import get_fnames
from dipy.io.gradients import read_bvals_bvecs
from dipy.io.image import load_nifti, save_nifti

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(name="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)
bvals, bvecs = read_bvals_bvecs(dwi_bval_fname, dwi_bvec_fname)

This data has 193 volumes. For this demo purpose, we decide to reduce the number of volumes to 3. However, we do not recommend to perform a motion correction with less than 10 volumes.

data_small = data[..., :3]
bvals_small = bvals[:3]
bvecs_small = bvecs[:3]
gtab = gradient_table(bvals_small, bvecs=bvecs_small)

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

data_corrected, reg_affines = motion_correction(data_small, gtab, affine=affine)

Optimizing level 2 [max iter: 10000]
Optimizing level 1 [max iter: 1000]
Optimizing level 0 [max iter: 100]
Optimizing level 2 [max iter: 10000]
Optimizing level 1 [max iter: 1000]
Optimizing level 0 [max iter: 100]
Optimizing level 2 [max iter: 10000]
Optimizing level 1 [max iter: 1000]
Optimizing level 0 [max iter: 100]
Optimizing level 2 [max iter: 10000]
Optimizing level 1 [max iter: 1000]
Optimizing level 0 [max iter: 100]
Optimizing level 2 [max iter: 10000]
Optimizing level 1 [max iter: 1000]
Optimizing level 0 [max iter: 100]
Optimizing level 2 [max iter: 10000]
Optimizing level 1 [max iter: 1000]
Optimizing level 0 [max iter: 100]

Save our DWI dataset corrected to a new Nifti file.

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

References

[1]

Mark Jenkinson and Stephen Smith. A global optimisation method for robust affine registration of brain images. Medical Image Analysis, 5(2):143–156, 2001. URL: https://doi.org/10.1016/S1361-8415(01)00036-6, doi:10.1016/S1361-8415(01)00036-6.