Fig. 2: Examples of segmentation results.
Distance maps were build using original modification of IsotropicDistance function from MAMBA-Image library. Modification consisted in replacing isotropic structuring element by anisotropic element with x-to-y radius ratio 1:4. Anisotropic element was used to create the system of clusters that better corresponded to visually appreciated grouping of neurons, judging by cluster outlines overlapped to initial images.
Fig. 3: Examples of clusters, obtained by thresholding of distance maps at radius x=4; y=16 micrometers for areas 10m and 37.
In each distance map cluster population distributions were calculated for distances from 2 to 24 (structuring element from 2×8 to 24×96 micrometers). Numerical densities values were normalized to 1 millimeter squared. Normalized average distributions for each area under study. Examples of obtained “Area Clusterization Signatures” are presented at the top of this post.