Skip to main content
. 2020 Dec 22;4:72. doi: 10.1186/s41927-020-00170-3

Table 2.

Optimal parameter set Ƥ as determined by RF training

RF number of trees 30
RF maximal tree depth 30
RF samples per node 0.0001%
GF kernel size 21 × 21
GF Gaussian sigma 1
GF Sinusoidal wavelength 1
ELBP (inner) radius 3 (1.5)
ELBP sample 8

Optimal parameter set Ƥ, showing the highest average Dice ratio during the validation step of the RF training