Asking for dti_diffeomorphic_population Error

[Peng Ren]

Dear Prof.Zhang,

Thanks for providing the powerful DTI-TK toolbox, when I try to run the dti_diffeomorphic_population command, there are errors like: 

(standard_in) 1: syntax error
(standard_in) 1: syntax error
/usr/local/dtitk-2.3.1-Linux-x86_64/scripts/dti_diffeomorphic_reg: line #141： [: -eq：unary operator expected  
iteration 1 done, not accepted ...
if you believe the registration terminates too early, you can relax the jacobian limit.
default jacobian limits can be changed by modifying dtitk_common.sh in the scripts directory of DTI-TK.

I've found some threads with similar questions in DTITK google group, including set origin to zero,  resample to 128x128x64 grid with 1.5x1.75x2.25 voxel size, and remove the 2 edge slices of all directions, I've done these work with the following codes, all seem to be met, but there keeps the same error. Besides, the first example goes well on my machine, could you give me some suggestions?

dti_template_bootstrap ixi_aging_template_resampled.nii.gz subjs.txt
dti_affine_population mean_initial.nii.gz subjs.txt EDS 3
TVtool -in mean_affine3.nii.gz -tr
BinaryThresholdImageFilter mean_affine3_tr.nii.gz mask.nii.gz 0.1 10000 1 0
# check before the next time-comsuming step, make sure none of the voxels in the top and bottome two slices are in the mask
# edit with AFNI version
3dZeropad -I -2 -S -2 -P -2 -A -2 -L -2 -R -2 -prefix mask_zcutup.nii.gz mask.nii.gz
3dZeropad -I 2 -S 2 -P 2 -A 2 -L 2 -R 2 -prefix newmask.nii.gz mask_zcutup.nii.gz
dti_diffeomorphic_population mean_affine3.nii.gz subjs_aff.txt newmask.nii.gz 0.002

the checking of the information:

(base) rp@master:/prot/PPMI/PreprossedDTI/test/3102/group$ VolumeInfo DTI_st_deoblique_dti_dtitk_unit_outlier_spd_aff.nii.gz
NIFTI Intent Code: NIFTI_INTENT_SYMMATRIX
NIFTI Orientation Code: LPI
Volume Info of DTI_st_deoblique_dti_dtitk_unit_outlier_spd_aff.nii.gz
size: 128x128x64, voxel size: 1.5x1.75x2.25, origin: [0, 0, 0]
(base) rp@master:/prot/PPMI/PreprossedDTI/test/3102/group$ VolumeInfo mean_affine3.nii.gz
NIFTI Intent Code: NIFTI_INTENT_SYMMATRIX
NIFTI Orientation Code: LPI
Volume Info of mean_affine3.nii.gz
size: 128x128x64, voxel size: 1.5x1.75x2.25, origin: [0, 0, 0]
(base) rp@master:/prot/PPMI/PreprossedDTI/test/3102/group$ VolumeInfo newmask.nii.gz
NIFTI Intent Code: NIFTI_INTENT_NONE
NIFTI Orientation Code: LPI
Volume Info of newmask.nii.gz
size: 128x128x64, voxel size: 1.5x1.75x2.25, origin: [0, 0, -0]

Best wishes,

Peng 

[Hui Zhang]

Hi Peng,

Were you able to complete the "First registration with DTI-TK" tutorial?

http://dti-tk.sourceforge.net/pmwiki/pmwiki.php?n=Documentation.FirstRegistration

Gary

--

---
You received this message because you are subscribed to the Google Groups "DTI-TK" group.
To unsubscribe from this group and stop receiving emails from it, send an email to dtitk+un...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/dtitk/c6752154-cf16-40a3-82d8-ae9499ad4b02n%40googlegroups.com.

[Peng Ren]

Dear Prof. Zhang,

I've tried that before, It goes exactly the same as the tutorial with iteration 1 done. 

All the best,

Peng

[Hui Zhang]

Thanks Peng for confirming that.  This suggests the issue you are experiencing most likely lies with your dataset.  I suggest that you carefully study the data provided as part of the First Registration tutorial.  By this, I mean you need to visually check all the input, intermediate, and output images so that you know what you should expect if everything is setup correctly.  Then you can compare these against what happens with your own dataset.

Gary

To view this discussion on the web visit https://groups.google.com/d/msgid/dtitk/32db18ab-912c-488e-8043-c493cd2fa659n%40googlegroups.com.

[Peng Ren]

Dear Prof. Zhang,

Finally figured it out, the problem is that I do preprocessing again myself after running fsl_to_dtitk, which automatically do these work already. Now, the registration seems to work fine. 

Just one more question to make sure:

I visualize the dtitk tensor file with 3d ellipsoid with TVglyphView, the first row represents the tensor image at scale=1, the second row represents the tensor image at scale=2, do this tensor image met the criteria?

 

Best wishes,

Peng

[Hui Zhang]

That is great news.  The tensors look fine to me.

Gary

To view this discussion on the web visit https://groups.google.com/d/msgid/dtitk/860dac28-6c00-44c7-a7e7-35b3fbda297fn%40googlegroups.com.

[Peng Ren]

Dear Prof. Zhang,

 I am registering the DTI images in PPMI with DTITK, I notice that there are many DTI images with potential artifacts like the figures below, are these effects acceptable? If not, would you mind giving some suggestions on operating quality control with DTITK?

#1

#2

#3

#4

All the best,

Peng

[Hui Zhang]

Hi Peng,

It depends on if these artefacts affect the estimated DTI maps.  Ultimately, you will need to review the results of the registration to know for sure.

Gary

To view this discussion on the web visit https://groups.google.com/d/msgid/dtitk/9f2df135-9642-420b-86a1-709a424167d6n%40googlegroups.com.

[Peng Ren]

Dear Prof. Zhang,

Thanks for the immediate reply, As for the review of the registration, I found that all the images are co-registered after running dti_diffeomorphic_population command. 

1. Do you think it's acceptable if large structures (ventricles and brain edges) are well-registered (as figures below) or do I need to check more small structures further?

2. Do you think these artefacts will affect the results of microstructure (FA, MD) analyses?

Underlay: intra-subject template; overlay: subject to Intra-subject

Best wishes,

Peng

[Gary Zhang]

Hi Peng,

If you need to do intra-subject alignment at the first stage, you may find the new DTI-TK addon created for this specific task useful

https://github.com/garyhuizhang/dtitk/tree/master/addons/IntraSubjGrpAlign

To assess the quality of your parameter maps, you may find another DTI-TK addon useful

https://github.com/garyhuizhang/dtitk/tree/master/addons/DPMqc

This add-on also provides a nice example of how to check quality of image registration as well.

Gary