Table 9 Performance comparison of models with or without the covariate on the testing sets in (a) classification of aggressive and non-aggressive EC, (b) TMB prediction for aggressive ECs, and (c) TMB prediction for non-aggressive ECs

From: Deep learning for endometrial cancer subtyping and predicting tumor mutational burden from histopathological slides

Task

Model selection

Backbone

 

Acc.

Sens.

Spec.

MeanSS

AUROC

(a) Classification of aggressive and non-aggressive EC

F1-score

Truncated ResNet50

with the covariate

0.89

0.93

0.84

0.89

0.97

  

Truncated ResNet101

 

0.84

0.80

0.89

0.84

0.86

  

Truncated ResNet152

 

0.86

0.83

0.90

0.87

0.93

  

Truncated ResNet50

without the covariate

0.67

0.76

0.54

0.65

0.73

  

Truncated ResNet101

 

0.77

0.71

0.86

0.78

0.87

  

Truncated ResNet152

 

0.81

0.82

0.79

0.80

0.85

(b) TMB prediction for aggressive ECs

Cross-Entropy

Truncated ResNet50

with the covariate

0.67

0.78

0.61

0.69

0.69

  

Truncated ResNet101

 

0.71

0.75

0.68

0.71

0.78

  

Truncated ResNet152

 

0.73

0.72

0.73

0.73

0.78

  

Truncated ResNet50

without the covariate

0.66

0.56

0.73

0.64

0.62

  

Truncated ResNet101

 

0.72

0.61

0.79

0.70

0.67

  

Truncated ResNet152

 

0.67

0.75

0.63

0.69

0.71

(c) TMB prediction for non-aggressive ECs

F1-score

Truncated ResNet50

with the covariate

0.64

0.57

0.67

0.62

0.69

  

Truncated ResNet101

 

0.69

0.38

0.84

0.61

0.65

  

Truncated ResNet152

 

0.66

0.76

0.61

0.68

0.70

  

Truncated ResNet50

without the covariate

0.53

0.52

0.54

0.53

0.58

  

Truncated ResNet101

 

0.69

0.48

0.79

0.63

0.60

  

Truncated ResNet152

 

0.63

0.48

0.70

0.59

0.60

  1. The bold case highlights the recommended setup of the proposed framework.