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.

Sign up for GitHub

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

Jump to bottom

Remove DCF from iterative SENSE algorithm #481

Open
ckolbPTB opened this issue Oct 25, 2024 · 3 comments · May be fixed by #539
Open

Remove DCF from iterative SENSE algorithm #481

ckolbPTB opened this issue Oct 25, 2024 · 3 comments · May be fixed by #539

Comments

@ckolbPTB
Copy link
Collaborator 

          How important is the weighted objective compared to only using a DCF corrected x0?

The one comparison in the Prussmann paper gives a slight advantage for >30 iterations, disadvantage for fewer.

https://arxiv.org/abs/1902.09657 ( Ong, Uecker, Lustig) also argue that the reweighting of the objective function is suboptimal due to the noise reshaping. the maximum likelihood solution would be without the dcf. they suggest proper preconditioning using different algorithms.

Originally posted by @fzimmermann89 in #388 (comment)
@ckolbPTB
Copy link
Collaborator Author

related to #473

@fzimmermann89
Copy link
Member

image
fessler

@fzimmermann89
Copy link
Member

fzimmermann89 commented Oct 25, 2024 •
edited
Loading

@Stef-Martin and me tried it out for his data.
We had to change the stopping tolerance for cg, otherwise the non-dcf version terminated after fewer iterations and had more artifacts. after fixed 10 iterations, the subjective quality in our ad-hoc tests was the same with and without dcf in the operator.
we required ~20 iterations if not starting with a dcf x0 but with zeros.

There also exist really cheap thus fast approximations for the dcf (pipe& menon or even a simple grid approximation) that should be good enough if only used for x0

@fzimmermann89 fzimmermann89 linked a pull request Nov 14, 2024 that will close this issue
Draft
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

Successfully merging a pull request may close this issue.

Use Gram Operator in iterative reco and operator norm PTB-MR/mrpro
2 participants
@fzimmermann89 @ckolbPTB