Table 9 Model performance comparison with and without TL by freezing/unfreezing the network layer. [Values in %].

From: An interpretable framework for gastric cancer classification using multi-channel attention mechanisms and transfer learning approach on histopathology images

Class

Layers

Sens

Spec

F1

Pre

Avg. Acc

Normal

Unfreeze

98.00

98.12

98.21

97.22

98.25

Freeze

98.80

98.91

99.00

98.24

98.87

Abnormal

Unfreeze

98.48

98.12

98.11

98.02

98.25

Freeze

98.97

99.20

98.91

98.60

98.87

Back to article page