Subdivision PILR analysis

[meganrm]

Try it in 2D first:

1. 1st sub division is in half along R, which is the nuclear boundary
2. get the total intensity for each of the subdivisions
3. compare histograms
   • stop if all intensity in one is the opposite of the other (need some threshold that means noise) . similarity of 0
   • If you have a subdivision that is not above noise threshold, also halt analysis of that subdivision
   • if two of the same subdivisions are sufficiently similar, continue to the next subdivision (division in half) and repeat these steps

Once we have a version of 2D in R we're happy with, we should compute it on a few other example structures from variance paper

question: then do it in each of the 3 dimensions independently or try doing subdivisions as "voxels"

[meganrm]

Need to define thresholds for both lower bound (ie noise) and how to compare histograms, what is "similar/the same" verses different histograms