-setorient RPI not responding

[ihattan]

Hi SCT Experts,

hope you are doing well. I’m still new to sct software and excited to use it as a main tool to analyse my Ex-vivo dwi dataset.

I wonder how I may solve the following issue to align dwi dataset with T1 mask orientation (RPI)?

T1 dataset and (DWI) were scanned in the sme space. So,I resampled the dwi dataset to T1 resol. Then I tried to re-orient my dwi data to RPI orientation to use T1 mask to generate dwi template but unfortunately it didn’t seem to work out.

I followed the following steps:

I used the following commands for T1 dwi dataset to get RPI orientation that matches the mask. Please see screenshots when I tried to overlay the dwi and t1 dataset with Charleys’ mask (nothing were changed).
sct_image -i dwi_top_resampled.nii.gz -setorient RPI dwi_top_resampled_RPI.nii.gz

I checked the orientation for all dataset using sct_getorient and fslinfo I got that the orientations and resolutions are alright. (please see the screenshots below).

In case if you would like to investigate from your side, I’m happy to share data files with you.

Could you please guide me whether I could bring t1 and dwi dataset to T1 mask space with correct RPI orientation or not?

Many thanks in advance …

Cheers,

Ibrahim

[jcohenadad]

Hi Ibrahim,

Thank you for reaching out to us.

What I would do is:

• segment the DWI data (in addition to the T1w data)
• register using a combination of the segmentations and the image.

If you can add a link to the images and segmentations, I’m happy to write a script for you.

Cheers
Julien

[jcohenadad]

Hi @ihattan,

Just checking in to see how I can help. Would you like to send me 1-2 images so I can check on my end how to tune the registration? You can send it to me via email if the data cannot be publicly disclosed.

Cheers,
Julien

[ihattan]

Hi @jcohenadad,

Thank you very much for following up regarding this issue. I did share dataset via email. Let me know please if further information is needed.

by the way, I wonder whether SCT tools could handle to straighten 4D dwi dataset or not.

I know it can handle it by option -slpit then use the 3D output. Here, dataset has to be in 4D before and after straightening.

Any other alternative way to do so in case sct tools can not fix it?

Many thank in advance,

Cheers,

Ibrahim

[jcohenadad]

Hi Ibrahim,

I did share dataset via email. Let me know please if further information is needed.

The link didn’t work-- see my email.

by the way, I wonder whether SCT tools could handle to straighten 4D dwi dataset or not.

To straighten a 4D volume, you can first estimate the curve2straight warping field using a 3D data, and then apply it to the 4D volume. Below is an example code:

# Download example data
sct_download_data -d sct_example_data
# Go to DWI folder
cd sct_example_data/dmri/
# Separate b=0 and DWI and average
sct_dmri_separate_b0_and_dwi -i dmri.nii.gz -bvec bvecs.txt
# Segment cord
sct_deepseg_sc -i dmri_dwi_mean.nii.gz -c dwi
# Straighten DWI mean data
sct_straighten_spinalcord -i dmri_dwi_mean.nii.gz -s dmri_dwi_mean_seg.nii.gz 
# Apply straightening to 4D dmri data
sct_apply_transfo -i dmri.nii.gz -d straight_ref.nii.gz -w warp_curve2straight.nii.gz 

Cheers,
Julien

[ihattan]

Thank you very much @jcohenadad. It’s very robust method and is working very well now.
In regard to the dataset, a link has been sent to your email. Let me know please if there is any issues while downloading.

Many thanks in advance…

Cheers,

Ibrahim

[jcohenadad]

Hi Ibrahim,

I successfully downloaded the data but didn’t have a chance to look at them yet (been very busy lately). I promise I will come back to you within the next few days.

[ihattan]

Great… no problem take your time…

[jcohenadad]

Hi Ibrahim,

I had a look at the images you sent me. First of all, I noticed that 4225_top_sc.nii.gz is not the segmentation of 4225_top_1.nii but of another image. In order to register your DWI scans to 4225_top_sc.nii.gz, I would also need the image that this mask is based on.

Also, I noticed that the two masks are flipped A-P:

Which means that one set of images has wrong headers. It is likely the dwi_top_mask.nii.gz (and its mask and 4225_top_1.nii), because the ventral horn are pointing towards “P” (it should be “A”).

So:

• could you please also send me the image for 4225_top_sc.nii.gz
• if you’d like me to also register 4225_top_1.nii to the DWI, let me know. I don’t need the mask for this image because the two images are already very close.
• if you could fix the header information that would be great.

[ihattan]

Hi @jcohenadad

Thank you very much for your valuable information. I re-sent all data via email including T1 that belongs to the mask you had already. Could you please also re-produce -setorient RPI and sct_deepseg/straightening as well from your side?

I tested these options but results did not look good.

Many thanks in advance…

Cheers,

Ibrahim

[ihattan]

They are the same but the mask for t1 is oriented in RPI. So, I tried to re-orient the dwi/t1 to RPI but did not work out. I think there was something wrong from my side.

Please see the output for the data orientation

s4333238@cai-wks1:/30days/s4333238/TEST_MRTRIX_POPU_TEMPLATE/top_bott_updated/4225/top$ sct_image -i 4225_dwi_top.nii.gz -getorient

--
Spinal Cord Toolbox (4.2.0)

ASL
s4333238@cai-wks1:/30days/s4333238/TEST_MRTRIX_POPU_TEMPLATE/top_bott_updated/4225/top$ sct_image -i 4225_top_1.nii -getorient

--
Spinal Cord Toolbox (4.2.0)

pixdim[0] (qfac) should be 1 (default) or -1; setting qfac to 1
pixdim[0] (qfac) should be 1 (default) or -1; setting qfac to 1
LPI
s4333238@cai-wks1:/30days/s4333238/TEST_MRTRIX_POPU_TEMPLATE/top_bott_updated/4225/top$ sct_image -i 4225_top_sc.nii.gz -getorient

--
Spinal Cord Toolbox (4.2.0)

RPI
s4333238@cai-wks1:/30days/s4333238/TEST_MRTRIX_POPU_TEMPLATE/top_bott_updated/4225/top$

I think I’m facing the similar problem as this one

Sct_straighten_spinalcord fails for rat spinal cord

Glad it solved it! Please note that we recently improved the straightening results (more details here). Looking at your image I think that those recent changes can improve your results. Until we create a new release, you can already test it out in the master’s branch. Cheers, Julien

[jcohenadad]

Hi Ibrahim,

sct_image -setorient only alters the qform/sform of the header, not the actual data. In your case, the header is inconsistent with the data orientation, so you need to enforce another orientation between your data and the header-- this is done using the setorient-data flag. See below (red: before, green: after):

I ran this:

sct_image -i 4225_top_sc.nii.gz -setorient-data RAI -o 4225_top_sc_RAI.nii.gz

Now I am wondering why the mask had a wrong orientation at the first place. You mentioned it is “Charley’s mask”: but how was it produced? What command?

There are also problems with the file 4225_top_1.nii: some fields which are wrong, e.g. pixdim0:

filename	4225_top_1.nii
size of header	348
data_type	UINT16
dim0		3
dim1		256
dim2		256
dim3		1024
dim4		0
dim5		0
dim6		0
dim7		0
vox_units	mm
time_units	s
datatype	512
nbyper		2
bitpix		16
pixdim0		0.000000  --> WRONG VALUE

As mentioned in the NIfTI documentation, pixdim[0] should be only -1 or 1.

Also, the file 4225_dwi_top.nii.gz has forbidden values in the header:

filename	4225_dwi_top.nii.gz
size of header	348
data_type	UINT16
dim0		4
dim1		128
dim2		512
dim3		128
dim4		20
dim5		1
dim6		1
dim7		1
vox_units	mm
time_units	s
datatype	512
nbyper		2
bitpix		16
pixdim0		1.000000
pixdim1		0.156250
pixdim2		0.156250
pixdim3		0.156250
pixdim4		nan  --> WRONG VALUE

I would strongly recommend using neuroimaging apps that keep the integrity of the image headers, otherwise you will end up with unexpected consequences.

Now that the masks are properly oriented, we can straighten:

sct_straighten_spinalcord -i 4225_top_1.nii -s 4225_top_sc_RAI.nii.gz 

Which gives the following result: