Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Inter-subject registration #16

Draft
wants to merge 6 commits into
base: main
Choose a base branch
from
Draft

Conversation

EvanBeal
Copy link
Collaborator

This PR reunites all the files used to obtain results on inter-subject registration (pipeline + specific files to select data, register PMJ, evaluate registration after the different steps of the process).

Description of the process

The image pairs to be registered are formed by using the MRI image (of the modality specified) of each subject in the dataset as a fixed image and randomly selecting another MRI image (of the modality specified) of any other subject in the dataset as a moving image. Using existing labels of the intervertebral discs position, the discs between the C2/C3 and C7/T1 vertebrae in fixed and moving images are aligned using translation to approximately register the spinal cord. The position of the pontomedullary junction (PMJ) (detected using SCT) in both images is used to translate the moving image along the right-left and posterior-anterior axes to roughly align the brain structure. Finally, the SC segmentations are computed to perform an axial slice-by-slice alignment of the center of mass of the SC. Deep learning registration models then use these images to register the entire volume and refine the spinal cord registration.

inter_sub_process

1. Translation in the S-I, P-A and R-L directions to align the positions of the C2/C3 and C7/T1 intervertebral discs (white dots). 2. Translation in the R-L and P-A directions to align the pontomedullary junction of fixed and moving images (blue dot). 3. The SC segmentations of the fixed and moving images are used to perform a slice-by-slice axial alignment of the SC using the center of mass. 4. Deep learning deformable registration using cascaded models.

State of the PR

This PR is a work in progress:

  • Different options available for the intra-subject registration have not been implemented (sub volumes, multi-session)
  • The option to use predefined image pairs has not been tested
  • The process has not been tested on a different dataset than the spine generic dataset
  • The pipeline does not allow to easily specify the file name for the intervertebral discs labels

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

1 participant