Dear Mahinda,
For the mris_preproc, I think Doug posted already explaining how it works - it uses a jacobian correction in a way that makes it mass-conservative. You may want to search the archives. For the ROIs, the standard method is to add up the areas of their constituent vertices, so it should be straightforward to describe I believe. And the statistics are based on the GLM, which is already well entrenched in the literature, with no need for details other than describing carefully the design you used and how you corrected for multiple testing.