Hello,
  I used Mricron (dcm2nii) and MRIConvert (dcm2niix) to convert dicom to 4D nifti. But I found that bval from dcm2niix is wrong (all zero), while from dcm2nii is correct (3 b0 + 30 b1000). And the bvec are different between dcm2nii and dcm2niix. I don't know which one I should trust? Here are the bvecs from the same participant. The scanner is GE. The scan date is 2016.
  Thank you in advance!

Yuting.

I suspect they are both correct. The bvec lists the gradient directions in image space, so one must consider how the rows, columns and slices are saved to disk. For example, DICOM images are written with the first row at the top of the image, with subsequent rows below it (the way we write English), while NIfTI describes ascending rows as being higher on the screen, the way we draw the Y axis on a cartesian graph. By default, dcm2niix reverses row order to match NIfTI expectations.

I would not recommend using dcm2nii - development ceased half a decade ago and both the DICOM standard and manufacturers usage has changed a lot since then. Please use an up to date version of dcm2niix. 

I would encourage you to test the vectors generated by dcm2niix. The "dedicated document" that can be downloaded from the dcm2niix manual explains how:

https://www.nitrc.org/plugins/mwiki/inde...

According to your suggestion, I have checked the .bvec from MRIConvert and dcm2nii, and yes, you are right, both of them are correct.

Thanks a lot.

yuting.