Thanks for the positive comments. A mask option might solve your problem with the automask. I'll try to look into that.

The 12-parameter affine transformation might be accomplished with 3dWarpDrive. I don't think it would be too useful to include it inside 3dDWItoDT.

The trace and some other anisotropy measures mentioned in the presentation are possible other computations we could add. Of course 3dcalc can make short work of that.

I think the cumulative weight factors might be useful for quality control (outliers). I put this in for my own debugging, but I think it might have additional utility. If you find one gradient gives a much lower cumulative weight factor, say 0.75, it might trigger you to look at that gradient volume more closely. The cumulative weight factors and the individual voxel weight factors used in reweighting are normalized to be around 1.0. As far as absolute numbers for quality control, you will have to get a feel for what's out of the usual range or construct your own database for that.

Thanks again for giving us some feedback.