I’m doing SEMS DTI on ex vivo mouse brains and using a high b-value. Thus my reference images have a very different intensity pattern from the diffusion images. I can use eddy_correct or a 2D version of it to improve the alignment of the diffusion images, but these programs always do a poor job with the reference. I have shuffled along by choosing the diffusion image which best aligns to my reference, then using that one to align the other images. After I get the aligned images, I then replace the reference image with the original and that gives reasonable DTI data. But I’ve noticed that my reference is often off by 3-4 pixels from the aligned set of diffusion images. Since this is consistent in all of my slices, I wanted to know if there was a way to translate a set of images by an arbitrary number of data points. A circular shift would work fine.
Thanks, Allen
|