Failed to read volume /OUTPUTS/b0_all.nii.gz Error : No image files match: /OUTPUTS/b0_all

[tasciseda]

Hello, I ran the synb0-DISCO docker container with the following command, sudo docker run --rm -v $(pwd)/INPUTS/:/INPUTS/ -v $(pwd)/OUTPUTS:/OUTPUTS/ -v $(pwd)/license.txt:/extra/freesurfer/license.txt --user $(id -u):$(id -g) leonyichencai/synb0-disco:v3.1 I got the following error; "Failed to read volume /OUTPUTS/b0_all.nii.gz Error : No image files match: /OUTPUTS/b0_all FINISHED!!!"

I attached my OUTPUTS file. As far as I understand topup and eddy steps didn't go through. Can you please help me with this issue. Thank you very much. Seda

[schillkg]

Can you post the full log output on the terminal?

[tasciseda]

Sure! my_log_output.pdf Thank you for your help.

[schillkg]

This is a memory issue. How much RAM are you allowing your container to access? We recommend >12Gb and it will work successfully.

Thank you, Kurt

[tasciseda]

I increased the RAM to 20gb and ran again. I got the other outputs except topup outputs and now I am getting this error "Topup: msg=topup_clp::topup_clp: Subsampling levels incompatible with image data." I should have 64 slices in my image. What might be the problem ? Thank you Seda

[schillkg]

What is the size of the data? TOPUP from FSL will give that error if your topup configuration file is written requiring subsampling by a factor of 2 or 4. Ours does indeed assume a matrix size where all values are even, but you can use other configuration files (b02b0_2.cnf I believe is provided by FSL). Let me know if you need help implementing that.

[tasciseda]

I attached the size of the data. I am a bit confused and your help is highly appreciated to implement this. Thank you!

[schillkg]

The 3rd dimension is an odd number. Our default configuration file will fail with odd numbers. There are a few options. You can replace this was an FSL provided file (they provide one called b02b0_2.cnf that does not subsample the data and does not require even numbers). Or you can zero-pad your dwi volume so that it is a multiple of two (adding one slice so that there are 74 slices). You can also just run the synthesis and run topup outside of the container/synb0 pipeline. I think zero-padding the image will likely be the easiest solution.

[tasciseda]

I see. I was checking to add zero to make it 74 slices however I couldn't find any documentation. Is there an fsl command to make it or should I use another software? Thank you, Seda

[schillkg]

yes - there are a few options - I typically use MRTrix functions. I believe "mrpad" can zero-pad images. Also FSL should have tools, something like "fslroi" should work. I have not done this, and it may actually be easier to remove the first or last slice than to actually 0-pad! Here is some info from FSL developers on the challenges with odd slices: https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=FSL;6c4c9591.2002

[tasciseda]

I have been trying couple methods, I extract one slice and padded with zeros, however when I try to add that new padded slice to the original image it said "header transformations of input images do not match" and then like you suggested I removed the last slice and topup worked in container. I have run eddy outside the container, I used the acqparams.txt that I used for the container which was ; 0 -1 0 0.0438149 0 -1 0 0.000000 my question is should I change acqparams.txt to; 0 -1 0 0.0438149 0 1 0 0.0438149 after the topup for other steps including eddy? Thank you. Seda

[schillkg]

acqparams should stay constant for the entire process. You will also want to be sure that the input to eddy is the same size as the input to topup.

[schillkg]

Do the results look reasonable after the synb0 process after zero-padding the image?

[tasciseda]

I wasn't able to do zero-padding, however I removed the last slice and the results looked reasonable for one subject. I will test with couple more subjects and see. Thank you! Seda