Skip to main content
. Author manuscript; available in PMC: 2017 Aug 3.
Published in final edited form as: IEEE Trans Med Imaging. 2014 Jul 8;33(12):2293–2310. doi: 10.1109/TMI.2014.2337057

TABLE III.

Parameter settings for random forest classification method in pathology detection.

GLCM # of bins per axis = 16
# of directions = 4
Offset = 2
Pixel intensity dynamic range = 16 bits

Rnd. fst # of trees in a forest = 70
% of training set used to build individual trees = 0.6

GLRLM # of directions = 4
# of levels = 8

Misc. ROI dynamic-range = 16-bits
ROI window = 7 × 7