Print

---

Hi Stam,

I transposed the bvecs because the toolbox I used requies a N*3 bvec rather than 3*N generated by dcm2nii.

I will try FSL-dtifit. Thank you very much.

Hi Liqin,
Looking at whole-brain tracking results (which I am assuming you show in this figure) isn't the most straightforward way to find out if and what has gone wrong. Having said that, I agree that the results on the right look strange.
Why did you transpose the bvecs output by dcm2nii?
It is sometimes possible that dcm2nii has failed. The easiest way to test that is to dtifit and see if the dti principal eigenvectors make sense. If this easy test fails, you are certain that the something in your orientations is wrong.
Cheers
Stam
On 29 Nov 2012, at 08:46, Liqin Yang wrote:
> Hi Stam,
> Thanks for help, but i 'm still not sure. Attached is the four results I obtained (all shown are with a length>30). The results with or without eddy_correct looks similar, but the results with individual bvecs were quite strange and some fibers (the long violet fibers) were lost. I have transposed the rows and columns of the bvec file generated by dcm2nii in Matlab with x=x'. I haven't used the FSL's tracking methods because I wanted to first have a look at the traditional FACT results and wll try FSL later. Maybe it is the problem of the tools I used and it will be right to use individual bvecs in FSL?
>
>
> ----- 原始邮件 -----
> 发件人：Stamatios Sotiropoulos <[log in to unmask]>
> 收件人：[log in to unmask]
> 主题：Re: [FSL] siemens standard bvec or individual bvec? eddy_correct or not?
> 日期：2012年11月28日 16点53分
>
> Hi Mozi,
> 1. Use the individual bvecs generated by dcm2nii. These will in general be different from the standard siemens, as they are rotated for slice angulation used in the acquisition.
> 2. It is true that, depending on the sequence, eddy current artifacts can be alleviated, but not fully removed. Eddy_correct will take care of any residual artifacts and also of subject motion that cannot be corrected during acquisition. Note that FSL5 now provides much better ways for distortion and eddy current correction (see topup and eddy toolboxes).
> Cheers
> Stam
> On 28 Nov 2012, at 07:37, Liqin Yang wrote:
>> hi experts,
>> I'm new for DTI, and I have two questions for DTI data analyss.
>> 1: For each subject, the bvec generated by dcm2nii or mriconvert are similar, but neither was same with the standard siemens directions provided by diffusion toolkit or others. Should i use individual bvec or standard siemens bvec? I tried both for fiber tracking, but the results are quite different.
>> 2, I used both eddy_crrected (by FSL) and raw data for fiber tracking using traditional FACT, the results were different for quite a few fibers (the direction, length and streamlines couts). As my aim is to find the differences of fibers in two group, I wonder if the eddy_correct step will make my results more credible or suspect? A Siemens engineer told me that the system did some eddy current correction during the scan, but maybe the eddy_correct more precise?
>> Thank you for any advice or discussion.
>>
>> Mozi
>>
> <test.png>