dMRI MoCo does not work on my IVIM data

SCT
1

Dear SCT team,

I have dataset of dwi with several b-vals for IVIM analyses. There is a movement between first b-val and the last one (and of course between others too but lower).

Preproc:

fslsplit Ax_IVIM.nii.gz OrigDWI -t
fslmaths OrigDWI0008 -add OrigDWI0009 -add OrigDWI0010 -div 3 OrigDWI_mean

0008 b-val=300; 0009 = 400; 00010 = 800

sct_deepseg_sc -i OrigDWI_mean.nii.gz -c dwi -qc qc

sct_create_mask -i OrigDWI_mean.nii.gz -p centerline,OrigDWI_mean_seg.nii.gz -f cylinder -size 35mm

manual corr of mask if needed 0011=B0

itksnap -g OrigDWI_mean.nii.gz -s OrigDWI_mean_seg.nii.gz -o OrigDWI0011.nii.gz

I tried MoCo

sct_dmri_moco -i Ax_IVIM.nii.gz -m mask_DWI_mean.nii.gz -bvec Ax_IVIM.bvec -qc qc -qc-seg OrigDWI_mean_seg.nii.gz

but it does not help. So I found this problem (Sct_dmri_moco doesn't seem to correct motion) and tried p=0 but it did not help either

sct_dmri_moco -i Ax_IVIM.nii.gz -bvec Ax_IVIM.bvec -m mask_OrigDWI_mean.nii.gz -g 1 -param poly=0,metric=MI,smooth=1

Here are my dependencies:

ct_check_dependencies

–
Spinal Cord Toolbox (6.1)

sct_check_dependencies

SYSTEM INFORMATION

SCT info:

  • version: 6.1
  • path: /home/marek/sct_6.1
    OS: linux (Linux-5.15.0-94-generic-x86_64-with-glibc2.31)
    CPU cores: Available: 16, Used by ITK functions: 16
    RAM: Total: 31980MB, Used: 2594MB, Available: 28882MB

OPTIONAL DEPENDENCIES

Check FSLeyes version…[OK] (1.3.0)

MANDATORY DEPENDENCIES

Check Python executable…[OK]
Using bundled python 3.9.18 (main, Sep 11 2023, 13:41:44)
[GCC 11.2.0] at /home/marek/sct_6.1/python/envs/venv_sct/bin/python
Check if data are installed…[OK]
Check if dipy is installed…[OK] (1.5.0)
Check if ivadomed is installed…[OK] (2.9.8)
Check if matplotlib is installed…[OK] (3.8.1)
Check if nibabel is installed…[OK] (3.2.2)
Check if nilearn is installed…[OK] (0.10.2)
Check if numpy is installed…[OK] (1.23.5)
Check if onnxruntime is installed…[OK] (1.16.1)
Check if pandas is installed…[OK] (1.4.4)
Check if portalocker is installed…[OK] (2.8.2)
Check if psutil is installed…[OK] (5.9.6)
Check if pyqt5 (5.12.3) is installed…[OK] (5.12.3)
Check if pytest is installed…[OK] (7.4.3)
Check if pytest-cov is installed…[OK] (4.1.0)
Check if requests is installed…[OK] (2.31.0)
Check if requirements-parser is installed…[OK]
Check if scipy is installed…[OK] (1.11.3)
Check if scikit-image is installed…[OK] (0.22.0)
Check if scikit-learn is installed…[OK] (1.3.2)
Check if xlwt is installed…[OK] (1.3.0)
Check if tqdm is installed…[OK] (4.66.1)
Check if transforms3d is installed…[OK] (0.4.1)
Check if urllib3 is installed…[OK] (2.0.7)
Check if pytest_console_scripts is installed…[OK]
Check if pyyaml is installed…[OK] (6.0.1)
Check if voxelmorph is installed…[OK] (0.2)
Check if wquantiles is installed…[OK] (0.4)
Check if xlsxwriter is installed…[OK] (3.1.9)
Check if spinalcordtoolbox is installed…[OK]
Check ANTs compatibility with OS …[OK]
Check PropSeg compatibility with OS …[OK]
Check if figure can be opened with PyQt…[OK]
Check if figure can be opened with matplotlib…[OK] (Using GUI backend: ‘QtAgg’)

Here are my testing data:

You can see that before MoCo (upper row 800 - 0) and after MoCo (lower row 800 - 0) there is some change in b800 image but still it is not registered with B0.

obrazek
obrazek767×675 55 KB

Please, can you help me?

Thank you,

Mark

2

Hi Mark,

Thank you for reaching out.

First, this is not a typical DWI volume, in that the first volume is not ‘b=0’ but ‘b=10’. See your Ax_IVIM.bval:

10	20	30	40	50	75	100	200	300	400	800	0

Second, your acquisitions are likely interleaved, which makes it harder to correct for motions caused by slowly (~Hz) varying offsets in B0, eg caused by breathing, as already discussed Sct_dmri_moco doesn't seem to correct motion - #6 by jcohenadad

That being said, I’ve looked into solutions for you. If you want to consider the first volume as a ‘b=0’ you can do the following:

sct_dmri_moco -i Ax_IVIM.nii.gz -bvec Ax_IVIM.bvec -bval Ax_IVIM.bval -bvalmin 11

However, this approach produces poor registration results. Some comments:

  • we need a mask and/or agressive cropping, as you rightfully figured
  • reducing smoothing seems to help
  • remove regularization seems to help
  • metric MI seems to be better than CC in this case
  • considering ‘b=0’ as all files below b=100 seems to help

Below the code that gives me best results (of course, more tweaking might produce even better results).

# Average DWI scans together, using custom bval threshold
sct_dmri_separate_b0_and_dwi -i Ax_IVIM.nii.gz -bvec Ax_IVIM.bvec -bal Ax_IVIM.bval -bvalmin 200
# Segment spinal cord
sct_deepseg_sc -i Ax_IVIM_dwi_mean.nii.gz -c dwi -qc qc
# Crop IVIM scan using dilated segmentation for better and faster performance of moco algorithm
sct_crop_image -i Ax_IVIM.nii.gz -m Ax_IVIM_dwi_mean_seg.nii.gz -dilate 15x15x0
# Crop segmentation (only for QC)
sct_crop_image -i Ax_IVIM_dwi_mean_seg.nii.gz -m Ax_IVIM_dwi_mean_seg.nii.gz -dilate 15x15x0
# Perform motion correction
sct_dmri_moco -i Ax_IVIM_crop.nii.gz -bvec Ax_IVIM.bvec -bval Ax_IVIM.bval -bvalmin 100 -g 1 -qc qc -qc-seg Ax_IVIM_dwi_mean_seg.nii.gz

And here is the QC report, showing all my experiments: :point_right: Link to QC

Hope that helps,
Julien

:information_source: If you wish to display properly formatted code in this forum, you need to use the characters ``` at the beginning and at the end of your code.