1

I have two matrices MRI scans of completely different sizes (141x172x110 and 176x208x176). I need to map the larger matrix B to the smaller one A. The values of matrix A are regions of a brain - an atlas, and the values of matrix B are the type of brain matter. I need to figure out from which region of the brain each index in B comes. Each index is a 1mm square area of space in both matrices. Each matrix is centered at the middle index in each matrix (median of each dimension), so overlaying the two matrices will create some overlap with boundaries of each index. I imagine some solution like Euclidean distance or majority vote is feasible, but not sure where to begin. Any ideas?

user2680312
  • 63
  • 6
  • Think you need to clarify a bit, here. You must have some registration information between the two matrices, right? And you want to find (say) the cell from A which is nearest to a given cell in B, for all cells in B? So the actual content of A and B does not need to be manipulated? Also, how do you want to handle out-of-sample data (B being larger than A)? Do the two matrices share a coordinate system? Are their dimensions parallel (e.g. are they both represented in standard stereotaxic coordinates?)? Bit more info, and there are straightforward solutions to this type of problem. – Rattus Ex Machina May 12 '15 at 11:25

1 Answers1

0

Looks like what you really need is to scale up the smaller matrix to the same (probably proportional) size as the larger matrix and use interpolation. You should be able to use this answer:

resizing 3D matrix (image) in MATLAB

Community
  • 1
  • 1
Justas
  • 5,718
  • 2
  • 34
  • 36