Table 9 Ablation analysis-capsule layer effectiveness.

From: A federated learning-based privacy-preserving image processing framework for brain tumor detection from CT scans

Model variant

Accuracy (%)

Precision (%)

Sensitivity (%)

F1-score (%)

ResNet-50 only

94.56

92.3

93.85

92.95

ResNet-50 + capsule networks

96.87

95.4

96.2

95.8

ResNet-50 + HGBOA (without capsule networks)

95.62

93.85

94.2

94.02

Full model: ResNet-50 + capsule networks + HGBOA

99.07

98.54

99.82

99.18

Back to article page