You must be logged in to post Login Register


Lost Your Password?
Search Forums:


 






Wildcard Usage:
*    matches any number of characters
%    matches exactly one character

Forum Diffusion Toolkit Usage The inversion direction(x,y,z)in DTK?

The inversion direction(x,y,z)in DTK?
Page:   1  
UserPost

2:03 am
November 18, 2011


SusanZhong

Member

posts 8
Print this Post
1

Post edited 7:10 am – November 18, 2011 by SusanZhong

Dear  Ruopeng,

 

  Thank for your Diffusion toolkit & Trackvis.

 

   I am a new user. I want to process a set of data which are from different apparatus, for example, some are from Siemens, some are form Philips, with a script. But now I encounter a question ahout the inversion direction. If I need to try to find the inversion direction for each apparatus manually? I think The inversion direction is depended on the gradient matrix. The operation of inversion y in DTK equals to add '-' to all y direction elements in the gradient matrix.  If there are some relationship among the gradient matrix which can be used in fsl , the gradient matrix used in DTK and the inversion direction? I don't know if the parameter "image_orientation_patient" effect the inversion direction? or The inversion direction in the DTK is the fixed y direction or the other direction,as the operation in the DTIStudio?

Best Regards and Thank you again.

 

Zhong

12:18 pm
November 18, 2011


Ruopeng

Admin

posts 406
Print this Post
2

My short answer is, there's no definitive way to determine that. Just "play around with it" (meaning try different inversion flags) until the result makes sense.

"image orientation patient" will affect inversion direction. But I'd suggest not messing with it, just leave it as what it originally is.

Ruopeng

7:17 pm
November 20, 2011


SusanZhong

Member

posts 8
Print this Post
3

Post edited 3:33 pm – November 21, 2011 by SusanZhong

Dear Ruopeng,

Thank you for the answer.

I think I did not express myself exactly.I mean if I preprocess the data by fsl first and then input the preprocessed files(nifti) to the DTK. The gradient matrix is the same as the fsl used (only change the x,y,z format to match the right order;x,y,z in a row). In this case,if the inversion flag is constant y? I test several sets of data ,the inversion flag y- direction is needed in order to make sence.But I am not sure.

Thank you.

Zhong 

10:49 am
November 21, 2011


SusanZhong

Member

posts 8
Print this Post
4

Dear Ruopeng

Thank you very much for your answer.
If I preprocess the data by fsl first and then input the preprocessed file(nifti)to the DTK. The gradient matrix is the same as the fsl used(only change the x,y,z format to match the right order;x1,y1,z1 in a row).In this case, the inverse flag is also unconstant? I test several sets of data, the inversion flag y-direction is needed in order to make sense.But I am not sure.

There is another question about the coordination of the fiber point.I read the .trk file(the raw data be reconstructed to the tensor is nifti) in a matix and then picture the fiber track and the fa.nii in one space by matlab. I find that the flip y coordination is need in order to match the fiber and fa.nii well. I want to konw if the flip y coordination is constant if I reconstruct the tensor with the nifti(which is available in fsl)and the track the fiber. or the coordination(x?y?z?)need to flip depending on the other parameter?

Thank you~have a nice day.

Zhong

10:49 am
November 21, 2011


Ruopeng

Admin

posts 406
Print this Post
5

Hi Zhong,

This really depends case by base. If invert y makes sense for that kind of data, then that's the way to do it.

Best,

Ruopeng

10:50 am
November 21, 2011


SusanZhong

Member

posts 8
Print this Post
6

Dear Ruopeng

Thank you very much for your answer.
If I preprocess the data by fsl first and then input the preprocessed file(nifti)to the DTK. The gradient matrix is the same as the fsl used(only change the x,y,z format to match the right order;x1,y1,z1 in a row).In this case, the inverse flag is also unconstant? I test several sets of data, the inversion flag y-direction is needed in order to make sense.But I am not sure.

There is another question about the coordination of the fiber point.I read the .trk file(the raw data be reconstructed to the tensor is nifti) in a matix and then picture the fiber track and the fa.nii in one space by matlab. I find that the flip y coordination is need in order to match the fiber and fa.nii well. I want to konw if the flip y coordination is constant if I reconstruct the tensor with the nifti(which is available in fsl)and the track the fiber. or the coordination(x?y?z?)need to flip depending on the other parameter?

Thank you~have a nice day.

Zhong

11:14 am
November 21, 2011


Ruopeng

Admin

posts 406
Print this Post
7

Post edited 5:19 pm – November 22, 2011 by Ruopeng

I'm not sure I understood clearly what you mean by "constant". I guess you are trying to find a "theory" so that you know when and how to flip the x/y/z coordinate. As I said before, my experience is it is almost impossible to figure that out. Different scanner makers, different softwares all treat coordinates differently (even between their own different versions). If flip y suits this type of data for you, then do it that way for other similar data. But there is no guarantee. You should always error-check in the end.

Ruopeng
Page:   1