Open dschmitz89 opened 5 years ago
normalized correlation is our standard go to metric for deformable registration. should work for this problem if you use appropriate regularization.
an experienced C++ developer should be able to implement this with some hints from existing implementation such as:
https://itk.org/Doxygen/html/classitk_1_1ANTSNeighborhoodCorrelationImageToImageMetricv4.html
brian
On Fri, May 17, 2019 at 3:03 AM dschmitz89 notifications@github.com wrote:
Hey guys,
we are using ANTs for multimodal registration of histological sections onto images of the tissue block before sectioning. The multimodality leaves us only with Mutual Information and Glocal Correlation as metrics. Normalized Gradient Fields (NGF) computes the sum of the angles between the normalized gradients of both images and should in consequence be more sensitive to local intensity variations than a global statistic as MI.
How complicated would it be to implement the NGF metric in ANTs? An older ITK implementation exists ( https://www.insight-journal.org/browse/publication/714 ). If you think that it is possible to do for an experienced C++ programmer within reasonable time, we might be able to do it ourselves.
Thanks in advance!
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/ANTsX/ANTs/issues/767?email_source=notifications&email_token=AACPE7W77QPTYG5MGPCATJTPVZKFBA5CNFSM4HNST6L2YY3PNVWWK3TUL52HS4DFUVEXG43VMWVGG33NNVSW45C7NFSM4GUKVCUQ, or mute the thread https://github.com/notifications/unsubscribe-auth/AACPE7V2IZS6SKFQXXRZY6DPVZKFBANCNFSM4HNST6LQ .
Has there been progress on this?
not to my knowledge.
Hey guys,
we are using ANTs for multimodal registration of histological sections onto images of the tissue block before sectioning. The multimodality leaves us only with Mutual Information and Glocal Correlation as metrics. Normalized Gradient Fields (NGF) computes the sum of the angles between the normalized gradients of both images and should in consequence be more sensitive to local intensity variations than a global statistic as MI.
How complicated would it be to implement the NGF metric in ANTs? An older ITK implementation exists ( https://www.insight-journal.org/browse/publication/714 ). If you think that it is possible to do for an experienced C++ programmer within reasonable time, we might be able to do it ourselves.
Thanks in advance!