Table 5 Analysis of Adam optimizer using different batch sizes.

From: FasNet: a hybrid deep learning model with attention mechanisms and uncertainty estimation for liver tumor segmentation on LiTS17

Batch Size

Validation Loss

Accuracy

Dice Coefficient

Jaccard Index

Precision

Recall

Specificity

F1 Score

16

0.0315

0.9631

0.8166

0.7887

0.7286

0.7930

0.9694

0.7063

32

0.0251

0.9954

0.8766

0.8487

0.8499

0.8560

0.9695

0.8363

64

0.0452

0.9302

0.7773

0.7577

0.7076

0.7792

0.9313

0.79252

128

0.0392

0.9137

0.7743

0.7472

0.6959

0.8350

0.9195

0.8044

Back to article page