search

MIC-DKFZ / TractSeg

Automatic White Matter Bundle Segmentation
Apache License 2.0
222 stars 73 forks source link

beginnings.get_data(), NR_POINTS, dilate=DILATION, 16%|███████████▊ #123

Closed Renatar18 closed 3 years ago

Renatar18 commented 3 years ago

Hello!

The command stops at 16%: Tractometry -i ${i}/tractseg_output/TOM_trackings/ -o results/FA/${i}_FA.csv -e ${i}/tractseg_output/endings_segmentations/ -s ${i}/FA.nii.gz --tracking_format tck

What is wrong?

Thank you!

wasserth commented 3 years ago

Which version of nibabel are you using? Using an older version might be a quick workaround.

Renatar18 commented 3 years ago

Hi Jakob, Thank you very much for your reply! I adressed your question to our IT engineer, and he replied:

'Now I've installed older nibabel version (2.5.2) in the conda environment for tractseg on your server. But just as a warning, dipy 1.1.1 requires nibabel>=3.0.0. So I'll really need an exact version to not have compatibility issues'.

What do you say? Which version should we install?

Thank you, Renata.

On Mon, Oct 5, 2020 at 8:10 AM Jakob Wasserthal notifications@github.com wrote:

Which version of nibabel are you using? Using an older version might be a quick workaround.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/MIC-DKFZ/TractSeg/issues/123#issuecomment-703590103, or unsubscribe https://github.com/notifications/unsubscribe-auth/AIQXUVRMZ4OWZ5JLPG4TDQLSJGZSDANCNFSM4RPHVD7Q .

wasserth commented 3 years ago

Which nibabel version did you use when the error occured?

Renatar18 commented 3 years ago

We had version 3.0.1 on it before. When downgraded it to 2.5.2 it told: "dipy 1.1.1 has requirement nibabel>=3.0.0"

On Tue, Oct 6, 2020, 03:00 Jakob Wasserthal notifications@github.com wrote:

Which nibabel version did you use when the error occured?

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/MIC-DKFZ/TractSeg/issues/123#issuecomment-704073552, or unsubscribe https://github.com/notifications/unsubscribe-auth/AIQXUVXYHMDUREK76DCBYKDSJK6AXANCNFSM4RPHVD7Q .

wasserth commented 3 years ago

I just tested TractSeg with nibabel 3.1.1 and it worked fine for me. I guess there is another error. What is the complete error message when it stops?

Renatar18 commented 3 years ago

This is the complete message when it stops. For some reason it works well for some subjects and doesn't work for the others from the same sample. What could be the problem?

On Fri, Oct 9, 2020 at 8:49 AM Jakob Wasserthal notifications@github.com wrote:

I just tested TractSeg with nibabel 3.1.1 and it worked fine for me. I guess there is another error. What is the complete error message when it stops?

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/MIC-DKFZ/TractSeg/issues/123#issuecomment-706161155, or unsubscribe https://github.com/notifications/unsubscribe-auth/AIQXUVSXUP3VJZ7S5KIJVALSJ4BFBANCNFSM4RPHVD7Q .

wasserth commented 3 years ago

Probably your FA image has different dimensions than your endpoint_segmentation images. Could you post the dimensions / affine of those? Did you use the --preprocess option?

Renatar18 commented 3 years ago

On the preprocessing I ran eddy_correct and rotated the vectors before starting tractSeg.

On Mon, Oct 12, 2020 at 10:47 AM Jakob Wasserthal notifications@github.com wrote:

Probably your FA image has different dimensions than your endpoint_segmentation images. Could you post the dimensions / affine of those? Did you use the --preprocess option?

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/MIC-DKFZ/TractSeg/issues/123#issuecomment-707165652, or unsubscribe https://github.com/notifications/unsubscribe-auth/AIQXUVQ6Z6XCTU6KKMZ2EJ3SKMJHZANCNFSM4RPHVD7Q .

wasserth commented 3 years ago

The affine matrices in the header of the images (your diffusion image, peak image, segmentation and FA image) have to be the same. Probably this is causing the error.

Renatar18 commented 3 years ago

They all seem to be the same. Are you willing to see the data? Should I upload one of the "wrong" subjects?

On Wed, Oct 14, 2020 at 3:38 AM Jakob Wasserthal notifications@github.com wrote:

The affine matrices in the header of the images (your diffusion image, peak image, segmentation and FA image) have to be the same. Probably this is causing the error.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/MIC-DKFZ/TractSeg/issues/123#issuecomment-708221134, or unsubscribe https://github.com/notifications/unsubscribe-auth/AIQXUVWMTDECCGKOYFRZJO3SKVIPTANCNFSM4RPHVD7Q .