Table 2 Hyperparameters for deep learning model for spine segmentation.

From: Attention LinkNet-152: a novel encoder-decoder based deep learning network for automated spine segmentation

Hyperparameter

Typical Range / Value

Learning Rate

0.001

Batch Size

8

Epochs

100

Optimizer

Adam

Dropout Rate

0.3

Loss Function

Dice Loss

Activation Function

ReLU

Input Image Size

256 × 256

Early Stopping Patience

10

Back to article page