Table 4 Comparison of predictive performance among different reconstruction losses on the multi-center independent test datasets

From: An integrated MRI-based diagnostic framework for glioma with incomplete imaging sequences and imperfect annotations

Dataset cohort

Losses

IDH

1p/19q

Pathology types

ACC

AUC

SPE

SEN

ACC

AUC

SPE

SEN

ACC

AUC

SPE

SEN

Public Test Set

L1

0.84

0.91

0.86

0.84

0.86

0.89

0.76

0.86

0.85

0.87

0.86

0.72

SSIM

0.84

0.91

0.86

0.84

0.86

0.87

0.75

0.86

0.86

0.89

0.87

0.73

SSIM-L1

0.84

0.87

0.86

0.84

0.87

0.90

0.79

0.87

0.85

0.89

0.89

0.74

MSE

0.87

0.93

0.88

0.87

0.89

0.92

0.82

0.89

0.89

0.91

0.91

0.79

In-house Test Set

L1

0.74

0.83

0.80

0.74

0.77

0.81

0.70

0.77

0.72

0.80

0.83

0.68

SSIM

0.74

0.80

0.80

0.74

0.78

0.79

0.73

0.78

0.72

0.76

0.81

0.67

SSIM-L1

0.75

0.83

0.80

0.75

0.78

0.78

0.71

0.78

0.71

0.77

0.82

0.66

MSE

0.80

0.85

0.81

0.80

0.80

0.84

0.80

0.80

0.79

0.81

0.83

0.71