Hello
 In general, I would heed the warning generated by dcm2nii and upgrade to dcm2niix. However, in your case with such old data the older software may do a better job. Your data is from a Siemens system circa 2004, and pre-dates my archival sets
. Therefore, it is hard to provide advice, though your removal of private tags seems very extreme and goes far beyond normal anonymization. Perhaps you want to convert the raw data to NIfTI directly and skip the anonymization stage - it is much easier to create anonymized NIfTI images. My memory is hazy for such old images, in general the Siemens CSA headers are key to determine diffusion parameters (0029,1010 and 0029,1020). Please be careful to validate any bval and bvec files you get. Good luck!