So I tested on a synthetic image (attached) with 3 ROIs of increasing sizes.
Coarseness for the small, medium and large ROIs were respectively:
Larger the ROI, smaller the values, which is consistent with my findings (T2_FLAIR Brain MRI analysis)
I did not verify yet but it seems that the glrlm and glszm _GrayLevelNonUniformity show also an association with the size of the ROI (Larger the ROI, Larger the values or Larger the ROI smaller the values if you look at GrayLevelNonUniformityNormalized)
Any insight? I have looked at the formulas but I don’t get why I observe that.
( This synthetic image is a modified image of a carpet )