Table 3 Shows the overall performance metrics of the EMLM when trained with the Weka selected DTI features of the tumorous region, NAWM region, and tumorous + NAWM region for the FWE DTI model.

From: Accurate low and high grade glioma classification using free water eliminated diffusion tensor metrics and ensemble machine learning

Region of interest

Weka attribute selection

Class

Sensitivity

Specificity

AUC-ROC score

Tumorous region

FWE FA of the core and edema

LGG

0.86

0.84

0.84

HGG

0.87

0.85

0.86

NAWM

FWE FA and MD of the NAWM

LGG

0.84

0.75

0.79

HGG

0.80

0.79

0.78

Tumorous + NAWM region

FWE FA of the core and edema, FA and MD of the NAWM

LGG

0.92

0.90

0.90

HGG

0.89

0.88

0.88

Back to article page