Hello Mrtrix developers,
here I’ve recently got some a bit strange DWI data which have multiple b0.
Here attached the bvals.
It has 64 directions.
Does anybody have any experience with this kind of data? Is there any special step to process it or not?
Any suggestion would be helpful.
Under other topics, Rob have posted that the mrtrix could automatically recognize your data, so I think there would be no any other extra step to perform when you use the mrtrix.
According to my own experience of dealing with those kind of data(such as data acquired from simense) using FSL, firstly, I extracted all b0 images together and mean them, used the mean b0 to perform topup and create b0 mask, then combined the mean b0 with rest images (bvalue≠0) as new DWI and perform eddy_correct.
Hope that could help you and others could help to correct me if there were errors in my process step
Thanks a lot!
That’s very helpful.
You don’t need to worry about having multiple b0 in the middle. In fact this is benefits eddy.
For this kind of data I just get the mean of all b0 to obtain the mask. For the rest I just leave the data as it is. If I ran topup I just concatenate the first b0 with a reversed phased encoding acquired b0, and for eddy I just use the original dataset as it is (you can denoise and correct for gibbs rings artifact the data before if you want). I hope this helps.
That’s very useful information.
I was confused by the multiple b0. But now it’s clear.
I got DWI images from OASIS database. I found that it is difficult to preprocess and create FOD image.
Appreciate if someone can give some suggestions to use this data.
Thank you very much.
I’m not sure where you got these data from, but they’re clearly not multi-shell, and not HARDI. There’s only 23 non b=0 volumes, and each b-value is unique. There is no way to process such data using our standard recommended tools. You might be able to perform regular diffusion tensor imaging though, as long as all these b-values are associated with different directions. Otherwise, I doubt there’s much you can do with data of this nature, unfortunately…
Thank you very much sir for the clarification.