mrinfo: exported bvecs differ from eddy_rotated_bvecs

Dear MRtrix experts:

I ran the following code to pre-process my diffusion data:

dwifslpreproc Data_den_unr.mif Data_den_unr_preproc.mif -rpe_all -pe_dir AP -readout_time 0.096 -nthreads 6 -eddy_options " --repol " -nocleanup

I used the mrinfo command to view the information of the output file (Data_den_unr_preproc.mif) and the bvecs information was as follows:

0.4634783483,-0.763593366,0.4495697855,5.0

0.4634478604,-0.7630779467,0.4504754466,5.0

0.5537714268,0.2257738444,-0.8014757501,1490.0

-0.4345934173,0.8995436095,-0.04415717654,2990.0

Then, I ran the following code to export the bvecs in FSL format:

mrinfo Data_den_unr_preproc.mif -export_grad_fsl ./bvec ./bval

and the bvecs information was as follows:

-0.494588056728427 -0.564401680223864 0.660933731533426

-0.494552474519908 -0.563622249014687 0.661625128301939

-0.535234482446633 -0.030392123088012 -0.844156601353036

0.464855071207986 0.824815867639136 -0.321851747324264

I also check the dwi_post_eddy.eddy_rotated_bvecs file in the scratch directory and the bvecs information was as follows:

-0.4856681123 -0.5655381308 0.6665531542

-0.484980488 -0.5649692755 0.6675355002

-0.543031251 -0.03326990226 -0.8390531414

0.4579956025 0.8264646381 -0.3274083538

If I need to use the .nii format diffusion image data and bvecs files in FSL format for subsequent analysis, which of the above bvecs files should I use?

I will be appreciated for your reply.

All the best,

XinY