I’ve been working on fixel based analysis of fibre density and cross section. I wanted to apply bias filed correction and intensity normalisation with mtbin. However, i couldn’t have the bias field corrected FOD images for all subjects. I have 8 subjects in my population and i could have only one bias field corrected FOD, for the other subjects the output images are empty. I don’t really understand what would be the problem because i didn’t have any error or warning.
Are you able to share (via link) a single problematic dataset (with the wm FOD, gm and csf compartment maps) with me via a personal message?
@Dave: this sounds like exactly the same problem I mentioned yesterday. When it occurred, no output images were created in our case too.
@Gizem: if you try to run
mtbin on one of those subjects, what command line output do you get?
I get the uncompressing process for WM , GM and CSF FOD images, mask and intensity normalisation and bias field correction process.
I was referring to the explicit command line output in case of such a dataset where it fails, because I reckon the issue you get is similar to one we’ve been getting ourselves for 3 separate studies in 1 day. Starting to look like a pattern…
So to clarify, do you see a command line output akin to this one for a failing dataset?
mtbin: [ . ] performing intensity normalisation and bias field correction......
mtbin: [SYSTEM FATAL CODE: SIGSEGV (11)] Segmentation fault: Invalid memory access
At the beginning I had the segmentation fault only for 2 subject’s data, then i used mask option and didn’t get any segmentation fault but the outputs are empty.
Thanks again for reporting and providing more information, @Gizem. Via your feedback and our own observations, we were able to implement a series of fixes that should hopefully deal with most/all causes that may lead to this problem. If you update your MRtrix3 install via…
…you can give the new version of
mtbin a go. You’ll notice it’s now mandatory to provide the
-mask option: supply it with the brain mask you computed earlier in the pipeline, and potentially upsampled as well (the one that you provided to
dwi2fod msmt_csd via the
-mask option too).
Let us know if you encounter any further issues!