Table 2 Multiple comparisons for MSI status prediction of endometrium cancer on TCGA dataset: LSD test

From: Deep learning to assess microsatellite instability directly from histopathological whole slide images in endometrial cancer

LSD multiple comparison

    

Measurement

(I) Method

(J) Method

Mean difference(I-J)

Std. error

Sig.

95%C.I.

      

Lower bound

Upper bound

Accuracy

Proposed method

ClassicMIL

***0.39

0.003

<0.001

0.38

0.39

  

CLAM

***0.25

0.003

<0.001

0.24

0.25

  

TOAD

***0.24

0.003

<0.001

0.23

0.24

  

KAT

***0.29

0.003

<0.001

0.28

0.29

Precision

Proposed method

ClassicMIL

***0.24

0.005

<0.001

0.23

0.25

  

CLAM

***0.12

0.005

<0.001

0.11

0.13

  

TOAD

***0.14

0.005

<0.001

0.13

0.15

  

KAT

***0.19

0.005

<0.001

0.18

0.20

Sensitivity

Proposed method

ClassicMIL

***0.45

0.01

<0.001

0.43

0.47

  

CLAM

***0.31

0.01

<0.001

0.29

0.34

  

TOAD

***0.26

0.01

<0.001

0.24

0.29

  

KAT

***0.25

0.01

<0.001

0.23

0.27

F-measure

Proposed method

ClassicMIL

***0.34

0.006

<0.001

0.33

0.35

  

CLAM

***0.22

0.006

<0.001

0.21

0.23

  

TOAD

***0.22

0.006

<0.001

0.21

0.23

  

KAT

***0.21

0.006

<0.001

0.20

0.22

  1. The mean difference is significant at the level of *0.05, **0.01 and ***0.001.
Back to article page