Table 1 The depicted hyperparameter space was explored during model training on the CLEAR dataset

From: Generating cervical anatomy labels using a deep ensemble multi-class segmentation model applied to transvaginal ultrasound images

Model Type

Hyperparameters

 

Residual Units

Optimizer

Dropout

Learning Rate

SegResNet

NA

*Adam, SGD

0.1, 0.2, *0.3, 0.4

0.1, 0.05, 0.01, *0.001

UNet

NA

*Adam, SGD

0.1, 0.2, *0.3, 0.4

0.1, *0.05, 0.01, 0.001

Residual UNet

2, *4

*Adam, SGD

0.1, 0.2, *0.3, 0.4

0.1, 0.05, 0.01, *0.001

nn-Unet

NA

*Adam, SGD

0.1, 0.2, 0.3, *0.4

0.1, 0.05, 0.01, *0.001

Attention UNet

NA

*Adam, SGD

*0.1, 0.2, 0.3, 0.4

0.1, 0.05, 0.01, *0.001

UNETR

NA

Adam, *SGD

0.1, 0.2, 0.3, *0.4

*0.1, 0.05, 0.01, 0.001

  1. The experimentally-derived best hyperparameters for each model type are indicated with a preceding asterisk (*item).
Back to article page