Hi, I’m still trying to follow this tutorial : http://userdocs.mrtrix.org/en/latest/tutorials/hcp_connectome.html I had first problems with 5ttgen but that has been solved (for more details see 5ttgen ). I was next able to run all subsequent “Structural image processing” steps. Regarding, the “Di…

Yes, I’m afraid that first error does indicate that the process ran out of memory. I’m not sure what data you’re using, but it’s likely that data.nii.gz file uncompresses to something much larger, maybe 3Gb say. Any processing on this file will need to produce an output file too, and hold that in RA…

I once ran into a similar issue. What I did was (spatially) crop the data to a field of view that just contains the brain. You’d be surprised how that’s typically only quite a small portion of the full image space (in 3D)…

That’s a good point: a decent fraction of the compression provided by saving as e.g. .nii.gz actually arises because of the vast volumes of zero-padding outside the brain, which can be represented very efficiently in a compressed format. In my experience, cropping the FoV of such images can reduce t…

I’ve just run the conversion from the uncompressed file data.nii $ gunzip data.nii.gz $ mrconvert data.nii DWI.mif -fslgrad bvecs bvals -datatype float32 -stride 0,0,0,1 and surprisingly it worked ! The memory usage was at 100%, my computer swapped for few minutes, and then the DWI.mif file was ge…

[image] deledalle: I’ve just run the conversion from the uncompressed file data.nii gunzip data.nii.gz mrconvert data.nii DWI.mif -fslgrad bvecs bvals -datatype float32 -stride 0,0,0,1 and surprisingly it worked ! The memory usage was at 100%, my computer swapped for few minutes, and then th…

I have now completed the “Diffusion image processing” steps with success. I now get a new error at step 2 of the “Connectome generation” (SIFT). tcksift 100M.tck WM_FODs.mif 10M_SIFT.tck -act 5TT.mif -term_number 10M tcksift: [100%] resampling ACT 5TT image to fixel image space tcksift: [100%] segm…

[image] deledalle: Thanks for the advice. To clarify, it’s a collegue of mine that will be doing the analysis and she knows this kind of stuff and knows about the HCP data. Since I have a more powerful laptop, I am in charge of the generation of the connectome. :+1: [image] deledalle: I …

The M suffix is indeed not working. It is just generating 10 tracks

@deledalle , a general recommendation that will make your life with MRtrix much easier is to inspect the outcome of every command, to make sure it makes sense. So each time you run a command that has an image as output, inspect it using mrinfo and load it into mrview. Each time you run a command th…

OK, looks like the integer multiplier postfix wasn’t working for that specific command & option; so instead of generating 100 million tracks, it probably just generated 100. I’ll push the fix once it’s finished testing. If anybody manages to find another instance where this mechanism fails, please l…

MRtrix3 Community

HCP connectome tutorial

show post in topic

Home
Categories
Guidelines
Terms of Service
Privacy Policy

Powered by Discourse, best viewed with JavaScript enabled