Hello all,

I'm performing exploratory ROI analysis and would like to know whether there is any recommendation / advice regarding statistical tools that can be used to assign significance to contour-change gradients on the surface. I know that contours can be plotted on the surface in SUMA (thus plotting ‘edges’), but I would like to understand if the gradient of change on the contour edge exceeds chance (in some sense). I.e., are there zones where the change is less gradual than expected by chance.

One (perhaps too liberal) possibility might be to populate a simulated (matched) ROI with random values from Normal distribution, smooth them to match the data smoothness, and calculate the distribution of gradients next to contour edges, thus establishing a sampling distribution, but I was wondering if anyone (Ziad?) has already hacked something more systematic in the past.

Related, is there code in AFNI to calculate gradients or does one have to implement an edge detector (sobel?) using 3dcalc?

Thanks,
Oori