Hi,

I'm trying to extract a tractography based on DSI_MRI.

I have some questions about how to obtain the tractography:

 

1. OUTPUT orientation from "diff_unpack".

I have DICOM images, and I want to convert them to NIFTI using "diff_unpack".

 

If the DICOM orientation is LPS, the NIFTI-1 output is also in LPS orientation or in LAS or RAS ¿?

The standard orientation for NIFTI-1 is LAS or RAS.

If "diff_unpack" doesn't change the orientation, the ImageOrientationPatient (which, in my case is Sagittal referenced to LPS) saved in the header doesn't change, right?

 

If there is a change into LAS or RAS, this change also affects to the ImageOrientationPatient reference.

In that case, does "diff_unpack" modify this reference or just keep the same as in DICOM header?

 

——

2. Gradient Vectors

My acquisition is different from the requirements of your software. I have 157 gradient vectors, and only 124 are needed. I interpolate this 157 gradient vectors to obtain just the 124 (with Matlab and load/save_untouch_nii, which keeps the orientation in the header).

My 157 gradients vectors are obtained from DICOM header, specifically from:

hdr=spm_dicom_headers(ind_file);

hdr.CSAImageHeaderInfo(ind_DiffusionGradientDirection).item(1:3).val

 

In order to do the interpolation, I would like to know the default reference system of these "124raw.txt" (also LPS ¿?). I guess DTK reorient the gradient vectors if there is a orientation information in the header, but I have to interpolate the gradient vectors before using DTK.

My 157 gradient vectors are in LPS. Do I also have to apply the sagittal conversion, or just LPS in the default orientation (axial)?.
I need to unify these SRIs.

 

—

3. Is the "default" reference system for both "124raw.txt" and "181vec.txt" (used for calculating the resulting odf) the same? In that case, which one?

 

—

4. Tractography

Finally, when I do the tractography, the _odf.nii and _max.nii are correct, but the tractography doesn't look correct.

It looks like, when computing the tractography with the resulting reconstruction parameters ("_odf.nii","_max.nii","_dwi.nii","_b0.nii") the dimensions for "_odf.nii" and "_max.nii" are [181 x y z], and the header transformation is not applied correctly.

 

It may be because of the reference rotation? I mean, if in the original MRI images the rotation reference is:

[3 1 2], with the _odf.nii it should be [4 2 3], and the order of the 1st dimension (181) should change too (because we need to keep the same SRI between the image and the odf vectors sampling).

 

I tried to obtain the DTI and DSI tractography with exactly the same parameters (auto-orientation and same gradient vectors) and, while in DTI it seems to work, in DSI it looks like the tractography orientation is not right.

(See snapshots: input MRI data with NIFTI-1 interpolated (bvec = 124raw.txt and without applying the header reorientation matrix (using load_untouch_nii in Matlab for the interploation))

 

(I couldn't upload images with higher resolution)

 

I guess the best option is to change the orientation at first, and work with the default reference system (this way I'll avoid the _odf.nii reorientation problem).

 

The main problem is about how DTK deals with different SRIs (LPS,RAS,LAS,…).

I hope you can help me with some of these questions.

Thanks,

 

Laura