NOTES

Versions and Tags:
------------------
(version information moved to NEWS file)
ver_0-3-1: switched to using NC_SHORT as the internal type to
           represent the gradient volumes. 
ver_0-3: Initialisation code more flexible, used in third pipe-run.
ver_0-2-1: Same as previous version, but actually compiles on IRIX.
ver_0-2: Second run of lmu pipeline
ver_0-1: Very first working prototype, still with many, um, issues.

Unresolved Issues:
------------------
* Two surface intersection.
* Surface thickness greater than one pixel.
	* This I should be able to solve through setting all of the areas
	  outside of the brain to a certain value, all of the areas inside
	  the brain as another, leaving all voxels identified as being 
	  part of either surface as part of the mantle.

Features Needed:
----------------
* Convergence criteria which measures change across iterations in 
  determining the mantle and stops once num changed increases.

Bugs and Bad Design:
--------------------
* Resource overrun check awkward
* Doesn't appear to check only mantle values
* Fails when h set to 0.5