Hi, Ajay--
Re. flipping gradients: different softwares and scanner systems just have different ways of referring to positive/negative x,y,z directions. I don't know why it's not standard, but many things in MRI like this aren't (very unfortunately!). I have usually found dcm2nii/dcm2niix to be reliable and low maintenance: I always check, though, to be sure, esp. when I get data from a new scanner, protocol, center, etc.
Basically, +/- differences don't affect magnitude values of DTI fitting, but they do affect the relative orientations in space (which itself is pretty important). I use tractography to help me feel confident that the system is consistent, and the tool in AFNI-FATCAT for this is @GradFlipTest-- I'd recommend using that. It basically tests out WB tractographic results for each flip (including no flip), and counts tracts to guess a winner: usually, then difference is quite large and clear, and you also get prompted to look at each result to verify.
Note that moving to/from software tools will change the flipping, likely: for example, TORTOISE internally does use a different system than AFNI (such is life), but one can always fix this pretty straightforwardly with @GradFlipTest and 1dDW_Grad_o_Mat++.
--pt