Table 4 Performance metrics of the CSHG-CervixNet with hold-out and k-fold cross-validation.

From: A hybrid compound scaling hypergraph neural network for robust cervical cancer subtype classification using whole slide cytology images

Model

Validation type

Accuracy (%)

Precision (%)

Recall (%)

F1-score (%)

Specificity (%)

GCN

Hold-out

97.58

97.67

97.59

97.57

99.12

GCN

5-Fold CV

98.96

98.95

98.93

98.62

99.99

CSHG-CervixNet

Hold-out

98.6

97.60

96.92

97.24

99.52

CSHG-CervixNet

5-Fold CV

99.31

98.97

99.38

99.34

99.77

Back to article page