CAN antspy use for CT abdomen registration？？ about antspy

[stnava]

Yes, it can and the range of solutions you can get is quite broad. this is why you typically want to Tune parameters. If you want additional help, you would need to share the images .

On Fri, Aug 16, 2019 at 3:40 AM chestnut111 ***@***.***> wrote: CAN antspy use for CT abdomen registration？？ I use antspy suyncc for CT abdomen registration， but ,It didn't work out so well Because of breathing, there's a big gap between the two 3d images — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#79?email_source=notifications&email_token=AACPE7WAXGZJM5AGTXCAGRTQEZKVLA5CNFSM4IME5ZVKYY3PNVWWK3TUL52HS4DFUVEXG43VMWVGG33NNVSW45C7NFSM4HFS4F4A>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AACPE7U5WMIN26QL4RQLM6LQEZKVLANCNFSM4IME5ZVA> .

-- brian

[chestnut111]

THX, my code is :
out = ants.registration(fixed, moving, type_of_transform='SyNCC') # no mask
and my out pic (mitk) is ：

left is fixed image ,middle is after reg image and the right is move image。
Because of the large deformation of the tissues such as breathing, liver, lungs, and heart, what appropriate parameters should I use? please! thank!

[stnava]

to properly answer this question, you would need to run an evaluation study. how would you do this?

• define an objective metric that measures performance quantitatively

• run a bunch of registrations to determine which performs best according to the above criterion

• change the similarity metric and its parameters - probably variants of MI for rigid/affine and CC for deformable will work

• try different number of multi-resolution strategies

• set the number of iterations to be large so you can achieve convergence

  • but start your experiments using only low-resolution so you can gain an intuition about performance without waiting a week

anyway, those are some general guidelines in addition to: https://www.ncbi.nlm.nih.gov/pubmed/24058331