EDDY CORRECTION PROBLEM

[Ítalo César Porto Xavier]

The following doesn't output the volume with eddy current, movement and susceptibility correction. What is the problem?

Regards,

Ítalo

[Fang-Cheng Yeh]

You may redirect the output by adding  > log.txt at the end of the command line.

The log output may give a hint.

You may also need to update DSI Studio if it is more than 6 month old.

Frank

--
You received this message because you are subscribed to the Google Groups "DSI Studio" group.
To unsubscribe from this group and stop receiving emails from it, send an email to dsi-studio+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/dsi-studio/b93efd0d-ed47-4eb6-a30b-6f69e9b7a0edn%40googlegroups.com.

[Ítalo César Porto Xavier]

Ops, thanks.

Now the following error occurred. I have already downloaded and installed it again.

Ítalo

[Ítalo César Porto Xavier]

Just for confirm, can I the following CL to do the same? How can I create the mask before executing it?

C:\Users\italo\Downloads\dsi_studio_win\plugin\eddy_cuda.exe --imain=1dwi.nii.src.gz.nii.gz --mask=1dwi.nii.src.gz.mask.nii.gz --acqp=1dwi.topup.acqparams.txt --index=1dwi.nii.src.gz.index.txt --bvecs=1dwi.nii.src.gz.bvec --bvals=1dwi.nii.src.gz.bval --out=1dwi.nii.src.gz.corrected --verbose=1

Thanks,

Ítalo

[Fang-Cheng Yeh]

You need to install cuda toolkit as mentioned at https://dsi-studio.labsolver.org/download.html

or you would need to use the CPU version.

Frank

To view this discussion on the web visit https://groups.google.com/d/msgid/dsi-studio/4e8d8e00-398f-4743-986b-7029e4cb45a3n%40googlegroups.com.

[Ítalo César Porto Xavier]

I have already CUDA11.7 installed.

My log.txt:

+ [1;34mDSI Studio version: Chen"陳" command line [0m
| + [1;34mchecking CUDA drivers [0m
| | CUDA Driver Version: 11070 CUDA Run Time Version: 11050
| | Device Count:1
| | + [1;34mDevice Number:0 [0m
| | | Arch: 50
| | | Device name: NVIDIA GeForce 840M
| | | Memory Clock Rate (KHz): 900000
| | | Memory Bus Width (bits): 64
| | | Peak Memory Bandwidth (GB/s): 14.4
| | |_0 ms
| |_115 ms
| DSI Studio version: Chen"陳" CPU/GPU computation enabled
| action=rec
| source=1dwi.nii.src.gz
| + [1;34mrun rec [0m
| | + [1;34mopen SRC file [0m
| | | + [1;34mopening 1dwi.nii.src.gz [0m
| | | |_1.022 s
| | |_2.369 s
| | + [1;34mreconstruction parameters: [0m
| | | method=4
| | | odf_resolving=0
| | | record_odf=0
| | | dti_no_high_b=1
| | | check_btable=0
| | | other_output=fa,ad,rd,md,iso,rdi,nrdi
| | | r2_weighted=0
| | | thread_count=4
| | | param0=1.25
| | | param1=3000
| | | param2=0.05
| | |_0 ms
| | + [1;34mspecify mask [0m
| | | mask=1
| | |_1 ms
| | + [1;34mpreprocessing [0m
| | | preprocessing=0
| | | motion_correction=0
| | |_0 ms
| | + [1;34madditional processing steps [0m
| | | cmd=[Step T2][Corrections][EDDY]
| | | + [1;34m[Step T2][Corrections][EDDY] [0m
| | | | run eddy
| | | | eddy without topup
| | | | get the bounding box for speeding up topup/eddy
| | | | trim "1dwi.nii.src.gz" for eddy
| | | | range: (0,0,0) to (256,256,70)
| | | | + [1;34msaving [0m
| | | | |_12.971 s
| | | | + [1;34msaving [0m
| | | | |_29 ms
| | | | run C:/Users/italo/Downloads/dsi_studio_win/plugin/eddy_cuda.exe
| | | | path:C:/Users/italo/Desktop/testehoje/teste DSI

| | | | --imain=1dwi.nii.src.gz.nii.gz
| | | | --mask=1dwi.nii.src.gz.mask.nii.gz
| | | | --acqp=1dwi.topup.acqparams.txt
| | | | --index=1dwi.nii.src.gz.index.txt
| | | | --bvecs=1dwi.nii.src.gz.bvec
| | | | --bvals=1dwi.nii.src.gz.bval
| | | | --out=1dwi.nii.src.gz.corrected
| | | | --verbose=1

| | | | completed.
| | | |_1.01883 m
| | | [1;31mERROR:eddy failed to process data [0m
| | |_1.01885 m
| |_1.0591 m
|_1.06118 m

[Fang-Cheng Yeh]

Sorry about the problem.

Could you send me the src.gz file to debug?

You may also try the CPU version.

Frank

To view this discussion on the web visit https://groups.google.com/d/msgid/dsi-studio/dc21b389-9075-47b9-a031-97eb38ab93bfn%40googlegroups.com.

[Ítalo César Porto Xavier]

Please don't be sorry mate!

https://drive.google.com/drive/folders/1j5_ZGXOM_JNM-ZrpQ8uhwi0DP9MNNx35?usp=sharing

Ítalo

[Fang-Cheng Yeh]

It needs permission, or you may send the data using the upload link provided on the DSI Studio website.

To view this discussion on the web visit https://groups.google.com/d/msgid/dsi-studio/2afede62-d8e5-41c4-976a-de450efe9798n%40googlegroups.com.

[Ítalo César Porto Xavier]

Ops my fault

I changed the permission, try again, please

[Fang-Cheng Yeh]

I checked the file, and eddy worked on my windows desktop.
It seems that the problem is still the CUDA toolkit.

Perhaps you may also try the CPU version?
Frank

On Wed, Oct 5, 2022 at 6:41 PM Fang-Cheng Yeh <fran...@gmail.com> wrote:
>
> You need to install cuda toolkit as mentioned at https://dsi-studio.labsolver.org/download.html
> or you would need to use the CPU version.
> Frank
>
> On Wed, Oct 5, 2022 at 5:26 PM Ítalo César Porto Xavier <italo....@gmail.com> wrote:
>>
>> Just for confirm, can I the following CL to do the same? How can I create the mask before executing it?
>>
>> C:\Users\italo\Downloads\dsi_studio_win\plugin\eddy_cuda.exe --imain=1dwi.nii.src.gz.nii.gz --mask=1dwi.nii.src.gz.mask.nii.gz --acqp=1dwi.topup.acqparams.txt --index=1dwi.nii.src.gz.index.txt --bvecs=1dwi.nii.src.gz.bvec --bvals=1dwi.nii.src.gz.bval --out=1dwi.nii.src.gz.corrected --verbose=1
>>
>> Thanks,
>>
>> Ítalo
>> Em quarta-feira, 5 de outubro de 2022 às 18:11:12 UTC-3, Ítalo César Porto Xavier escreveu:
>>>
>>> Ops, thanks.
>>>
>>> Now the following error occurred. I have already downloaded and installed it again.
>>>
>>>

>>> Ítalo
>>>
>>> Em quarta-feira, 5 de outubro de 2022 às 18:01:58 UTC-3, Frank Yeh escreveu:
>>>>
>>>> You may redirect the output by adding > log.txt at the end of the command line.
>>>> The log output may give a hint.
>>>> You may also need to update DSI Studio if it is more than 6 month old.
>>>> Frank
>>>>
>>>>
>>>> On Wed, Oct 5, 2022 at 4:57 PM Ítalo César Porto Xavier <italo....@gmail.com> wrote:
>>>>>
>>>>> The following doesn't output the volume with eddy current, movement and susceptibility correction. What is the problem?
>>>>>
>>>>>

[Ítalo César Porto Xavier]

I tried CPU version as you said. I got my CPU on 95%, 4Gb of memory and passed 1 hour processing didn't finish, I had to end the task.

I downloaded the CUDA 11 (local) again and still the error, in GUI works but in the command line doesn't.

With Batch:

Ítalo Xavier

[Fang-Cheng Yeh]

I tried CPU version as you said. I got my CPU on 95%, 4Gb of memory and passed 1 hour processing didn't finish, I had to end the task.

CPU eddy usually runs for several hours, especially for more advanced DWI.

 

I downloaded the CUDA 11 (local) again and still the error, in GUI works but in the command line doesn't.

Interesting. Is CLI here DSI Studio's CLI or did you call the eddy_cuda directly? 

Frank

 

To view this discussion on the web visit https://groups.google.com/d/msgid/dsi-studio/74721ddd-2f06-4efd-98a3-add1cd504c25n%40googlegroups.com.

[Ítalo César Porto Xavier]

Executed in GUI (15 min) but I want to execute in CMD. How do I call eddy_cuda directly?

Ítalo Xavier

[Ítalo César Porto Xavier]

I mean, I don't know if in GUI, you call eddy.exe or eddy_cuda.exe by the way

Ítalo Xavier

[Fang-Cheng Yeh]

Open the [console...] mode and check out what it tells when you run GUI eddy.

To view this discussion on the web visit https://groups.google.com/d/msgid/dsi-studio/c77b0139-b99e-4b4e-9db3-d6236b3aabf2n%40googlegroups.com.

[Ítalo César Porto Xavier]

It runs eddy_cuda.exe

[Fang-Cheng Yeh]

You should be able to get the same from CLI.

One possible cause is the running directory.
eddy_cuda from TinyFSL disabled error reporting and may exist
prematurely without prompting the correct error.

On Thu, Oct 6, 2022 at 2:33 PM Ítalo César Porto Xavier

<italo....@gmail.com> wrote:
>
> It runs eddy_cuda.exe
>
> Em quinta-feira, 6 de outubro de 2022 às 15:08:46 UTC-3, Frank Yeh escreveu:
>>
>> Open the [console...] mode and check out what it tells when you run GUI eddy.
>>
>> On Thu, Oct 6, 2022 at 2:07 PM Ítalo César Porto Xavier <italo....@gmail.com> wrote:
>>>
>>> I mean, I don't know if in GUI, you call eddy.exe or eddy_cuda.exe by the way
>>>
>>> Ítalo Xavier
>>>
>>> Em quinta-feira, 6 de outubro de 2022 às 15:05:16 UTC-3, Ítalo César Porto Xavier escreveu:
>>>>
>>>> Executed in GUI (15 min) but I want to execute in CMD. How do I call eddy_cuda directly?
>>>>
>>>> Ítalo Xavier
>>>> Em quinta-feira, 6 de outubro de 2022 às 14:57:19 UTC-3, Frank Yeh escreveu:
>>>>>>
>>>>>>
>>>>>> I tried CPU version as you said. I got my CPU on 95%, 4Gb of memory and passed 1 hour processing didn't finish, I had to end the task.
>>>>>
>>>>>
>>>>> CPU eddy usually runs for several hours, especially for more advanced DWI.
>>>>>
>>>>>>
>>>>>>
>>>>>> I downloaded the CUDA 11 (local) again and still the error, in GUI works but in the command line doesn't.
>>>>>
>>>>>
>>>>> Interesting. Is CLI here DSI Studio's CLI or did you call the eddy_cuda directly?
>>>>> Frank
>>>>>
>>>>>>
>>>>>>
>>>>>> With Batch:
>>>>>>

> To view this discussion on the web visit https://groups.google.com/d/msgid/dsi-studio/d9345a85-6dfa-4746-a734-e3057e7f9da7n%40googlegroups.com.

[Ítalo César Porto Xavier]

The runnin directory is the same (C:\Users\italo\Desktop\testehoje\teste DSI). PS.: When I runs in cmd console the .exe path, it shows me the help, shows me the help for eddy.exe but for eddy_cuda.exe give me the same error:

Ok, so let me try calling the executables directly. What is the sequence, for calling eddy_cuda.exe and the parameters?

Waiting for your help mate.

Ítalo Xavier

[Fang-Cheng Yeh]

The calling parameters are listed in the console mode output.

As for the missing dll file, copying the dll file to the same directory may handle the error message. But I am not sure why it happened. I tested on my windows desktop, but did not show this error message.

Frank

To view this discussion on the web visit https://groups.google.com/d/msgid/dsi-studio/eec61240-6ad1-46a9-8f8b-094760571c20n%40googlegroups.com.

[Ítalo César Porto Xavier]

What's the output on your console?

[Fang-Cheng Yeh]

+ eddy

| + [Step T2][Corrections][EDDY]

| | run eddy

| | eddy without topup

| | get the bounding box for speeding up topup/eddy

| | trim "1dwi.nii.src.gz" for eddy

| | range: (50,29,0) to (208,231,70)

| | + saving

| | |_13.471 s

| | + saving

| | |_27 ms

| | run E:/Dropbox/work/GitHub/DSI-Studio-CMAKE/plugin/eddy_cuda.exe

| | path:C:/Users/YEHFC/Downloads

| | --imain=1dwi.nii.src.gz.nii.gz

| | --mask=1dwi.nii.src.gz.mask.nii.gz

| | --acqp=1dwi.topup.acqparams.txt

| | --index=1dwi.nii.src.gz.index.txt

| | --bvecs=1dwi.nii.src.gz.bvec

| | --bvals=1dwi.nii.src.gz.bval

| | --out=1dwi.nii.src.gz.corrected

| | --verbose=1

| | Reading images

| | Performing volume-to-volume registration

| | Running Register

To view this discussion on the web visit https://groups.google.com/d/msgid/dsi-studio/199794d1-6c41-4dad-abbe-899c4ead1653n%40googlegroups.com.

[Ítalo César Porto Xavier]

Hm not much differences.

Two questions:

1. Whats is the command for the acquisition parameters (.txt)?

2. Why my source image fliped 180º after eddy correction reconstruction?

Ítalo Xavier

[Fang-Cheng Yeh]

Perhaps I can check your SRC file at my side (Have you sent me your
SRC to check? Sorry for my poor memory.)
Frank

On Thu, Oct 6, 2022 at 4:46 PM Ítalo César Porto Xavier

> To view this discussion on the web visit https://groups.google.com/d/msgid/dsi-studio/a6fe4f9a-6280-4317-a257-602a892c3755n%40googlegroups.com.

[Ítalo César Porto Xavier]

All my files is on the link: https://drive.google.com/drive/folders/1j5_ZGXOM_JNM-ZrpQ8uhwi0DP9MNNx35

You forget the first question too XD

Ítalo Xavier

[Fang-Cheng Yeh]

The first question is about FSL's eddy which you can find answers at https://fsl.fmrib.ox.ac.uk/fsl/fslwiki/eddy

To view this discussion on the web visit https://groups.google.com/d/msgid/dsi-studio/c81095a9-f056-4288-a8df-1dd92f1b509bn%40googlegroups.com.

[Fang-Cheng Yeh]

From the files you showed, the eddy was done correctly.

> 2. Why my source image fliped 180º after eddy correction reconstruction?

NIFTI file can be saved in LPS and RAS provided the srow is written correctly: https://nifti.nimh.nih.gov/

[Ítalo César Porto Xavier]

Hi! Appreciate your response,

eddy_cuda worked after updating drives and the VS.

How can I identify the differences after eddy?

Ítalo Xavier

[Fang-Cheng Yeh]

You would need to visually check the eddy distortion: https://www.youtube.com/watch?v=stL4GMeTC1I

For the command line, use quality control (--action=qc) and check the NDC value.

Frank

To view this discussion on the web visit https://groups.google.com/d/msgid/dsi-studio/9f7e65a4-fc1e-4070-b94d-8f4a5bf817b4n%40googlegroups.com.

[Ítalo César Porto Xavier]

Actually the source file fliped not the corrected file after eddy. I didn't understand this translation.

Quality control tells what about eddy?

Here the NDC (Step T1a) = 0.794313

Ítalo Xavier

[Fang-Cheng Yeh]

NDC needs to compare before/after eddy correction.

Frank

To view this discussion on the web visit https://groups.google.com/d/msgid/dsi-studio/f64af396-67c7-4b63-9915-0f5a09a19c35n%40googlegroups.com.

[Fang-Cheng Yeh]

> Actually the source file fliped not the corrected file after eddy. I didn't understand this translation.

Could you let me know how you confirmed it is flipped?
Frank

[Ítalo César Porto Xavier]

I've confirmed the source file fliped visually using O1: View Images. 

I am afraid NDC checked the same file, because quality control ask for the folder not the file.

Ítalo Xavier

[Fang-Cheng Yeh]

Thanks for attaching the screenshot. 

As I mentioned, the voxel orders are only saved in different orders (LPS vs RPS), and the head position is still the same.

===> NIFTI file can be saved in LPS and RAS provided the srow is written correctly: https://nifti.nimh.nih.gov/

--

You received this message because you are subscribed to the Google Groups "DSI Studio" group.
To unsubscribe from this group and stop receiving emails from it, send an email to dsi-studio+...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/dsi-studio/f930b1de-687a-4be3-b03c-dbaa2b0ebeb9n%40googlegroups.com.

[Fang-Cheng Yeh]

Hi Ítalo,

    I checked the screenshot in detail and the DSI Studio code again. You are right that the SRC file is flipped. This is a visualization bug when showing the content of the SRC file. The file content is okay.

    I am updating DSI Studio to avoid this issue: https://github.com/frankyeh/DSI-Studio/actions/runs/3221971871 (should be ready in an hour).

     Thanks a lot for the help, and I am sorry for the inconvenience caused.

Best,

Frank

[Ítalo César Porto Xavier]

Hi Frank,

How you concluded that is ok with the .src? 

How you compare before and after eddy? The quality control only check the source files as your site describes:

I would like to know how can I defend that the eddy actually corrected the image.

Regards,

Ítalo Xavier

[Fang-Cheng Yeh]

I would save a new SRC file after eddy

There is a --save_src parameter with --action=rec for this purpose  

To view this discussion on the web visit https://groups.google.com/d/msgid/dsi-studio/e50d0014-2e34-4c59-9fb2-851a7fd07caan%40googlegroups.com.

[Ítalo César Porto Xavier]

I collected 3 examples before and after eddy:

Renamed the corrected file as the same as the .src file, with other prefix:

Couldn't load the files after eddy:

Ítalo Xavier

[Fang-Cheng Yeh]

SRC files are not NIFTI files. Renaming them won't work out.

You would need to save SRC files from DSI Studio.

To view this discussion on the web visit https://groups.google.com/d/msgid/dsi-studio/89e7d7b1-bbcf-493f-991f-b0fba3d9e5a8n%40googlegroups.com.

[Ítalo César Porto Xavier]

What? The corrected file after eddy correction has .src in name. Do you want me to do what with the corrected file if the file is already a src?

I don't get it...

[Fang-Cheng Yeh]

Renamed the corrected file as the same as the .src file, with other prefix <= this won't work correctly.

To view this discussion on the web visit https://groups.google.com/d/msgid/dsi-studio/b864d5b6-4315-4539-8844-0cad94c0b82en%40googlegroups.com.

[Ítalo César Porto Xavier]

This I understand mate.

Do I have to create a copy of bvec and bval for the same subject, create the source file for the corrected, and then rerun the Quality Control?

[Fang-Cheng Yeh]

It is one solution.

What I usually did is use --save_src with --action=rec to save another copy, or in the GUI, use the [File] menu to save a corrected copy.

Frank

To view this discussion on the web visit https://groups.google.com/d/msgid/dsi-studio/62401764-2906-4148-888d-2813de43589dn%40googlegroups.com.

[Ítalo César Porto Xavier]

1. '--save_src' ask for input?

2. Must I put --action=rec also --action=qc?

3. Can I do it in one line?

Ítalo Xavier

[Fang-Cheng Yeh]

1. --save_src=new.src.gz

2. No

3. No

To view this discussion on the web visit https://groups.google.com/d/msgid/dsi-studio/c059f3eb-dcef-4965-a12e-14321db298e7n%40googlegroups.com.

[Ítalo César Porto Xavier]

Thank you Frank. Do you know how to solve the bug while executing eddy_cuda?

[Fang-Cheng Yeh]

You need to install cuda toolkit.

If still does not work, then use the CPU version of DSI Studio.

Frank

To view this discussion on the web visit https://groups.google.com/d/msgid/dsi-studio/cf57d92c-63db-43fb-b19b-d6ea12e26db8n%40googlegroups.com.

[Ítalo César Porto Xavier]

Thank you!

[Ítalo César Porto Xavier]

Hi mate, I'm here again.

I am executing the eddy_cuda.exe in command line and sometime I get the error on the image, Do you know what it is?

Regards,

Ítalo

[Ítalo César Porto Xavier]

See that I ran again and didn't get any error:

You received this message because you are subscribed to a topic in the Google Groups "DSI Studio" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/dsi-studio/F_YvhnkYHm0/unsubscribe.
To unsubscribe from this group and all its topics, send an email to dsi-studio+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/dsi-studio/52969337-d0a1-4ad6-92e5-07bb6d1cce6dn%40googlegroups.com.

[Fang-Cheng Yeh]

It would be a problem with the CUDA driver, but there could be other causes.

unfortunataely, i am not familiar with CUDA in Ubuntu.

Frank

To view this discussion on the web visit https://groups.google.com/d/msgid/dsi-studio/52969337-d0a1-4ad6-92e5-07bb6d1cce6dn%40googlegroups.com.

[Ítalo César Porto Xavier]

I'm using the CUDA in Windows. I'll check the driver update.

Thank you Fang

To view this discussion on the web visit https://groups.google.com/d/msgid/dsi-studio/CAG_Qrt%2BaVcFUdvTG3v%2Btyvz3tNou1ACgs84yL%3Dza56K_-QfVcw%40mail.gmail.com.

[Ítalo César Porto Xavier]

Hi Frank!

My DWI data has distortion from posterior to anterior (image below), and I've checked that the MR uses the following parameters: 0 1 0 0.1 (dir=+y)

When I ran the Eddy correction in GUI, generated the .txt with 0 -1 0 0.5, this change the results, what should I do to change it in command line?

Regards,

Ítalo Xavier