Skip to main content
. 2019 Feb 28;13:97. doi: 10.3389/fnins.2019.00097

Table 1.

Model parametrization.

Optimized performance measure/ Hyperparameter Dice coefficient 95% Hausdorff distance Averaged Hausdorff distance
U-NET MODEL PARAMETRIZATION
Learning rate 1e-4 1e-4 1e-5
Batch size 16 64 8
Dropout rate 0 0 0
HALF U-NET MODEL PARAMETRIZATION
Learning rate 1e-4 1e-4 1e-4
Batch size 64 32 32
Dropout rate 0.1 0.2 0.1
GRAPH CUT PARAMETRIZATION
Hyperparameter Value
Weights 10