Skip to main content

Table 3 Comparison between FOR-PROCESSING and FOR-PRESENTATION features computed on “dense” area and “non-dense” area

From: Radiomic features of breast parenchyma: assessing differences between FOR PROCESSING and FOR PRESENTATION digital mammography

Feat group

Feature name

w

r (dense)

r (no dense)

Can Cor

PD FOR PRES dense

PD FOR PROC dense

PD FOR PRES no-dense

PD FOR PROC no-dense

1st or

10 percentile

 

0.99

1.00

1.00

− 0.21

− 0.16

− 0.15

− 0.17

 

90 percentile

 

0.99

1.00

1.00

− 0.21

− 0.11

− 0.16

− 0.15

 

Energy

 

1.00

1.00

1.00

0.51

0.54

− 0.45

− 0.45

 

Entropy

 

0.57

0.64

0.71

0.13

0.55

0.25

0.39

 

Interquartile Range

 

0.50

0.30

0.50

0.17

0.71

− 0.24

0.36

 

Kurtosis

 

0.57

0.92

0.92

− 0.37

− 0.62

(a)

− 0.15

 

Maximum

 

0.78

0.57

0.84

− 0.20

− 0.12

− 0.35

− 0.15

 

Mean Absolute Deviation

 

0.47

0.61

0.76

(a)

0.66

− 0.15

0.13

 

Mean

 

0.99

1.00

1.00

− 0.21

− 0.13

− 0.17

− 0.17

 

Median

 

1.00

1.00

1.00

− 0.20

− 0.13

− 0.17

− 0.18

 

Minimum

 

0.97

0.98

− 0.24

− 0.16

 

Range

 

0.59

0.57

0.80

(a)

0.42

− 0.35

− 0.15

 

Robust Mean Absolute Deviation

 

0.49

0.29

0.52

0.15

0.71

− 0.26

0.28

 

Root mean squared

 

0.99

1.00

1.00

− 0.21

− 0.13

− 0.16

− 0.17

 

Skewness

(a)

0.64

0.92

0.92

− 0.49

− 0.58

(a)

0.13

 

Total energy

 

1.00

1.00

1.00

0.51

0.54

− 0.45

− 0.45

 

Uniformity

 

0.55

0.61

0.74

− 0.16

− 0.59

− 0.23

− 0.45

 

Variance

 

0.44

0.94

0.99

(a)

0.60

(a)

0.08

glcm

Autocorrelation

 

0.51

0.40

0.55

0.31

0.36

0.21

− 0.31

 

Cluster Prominence

 

0.39

0.71

0.53

− 0.16

0.35

0.18

0.06

 

Cluster Shade

 

0.40

0.57

0.50

− 0.37

(a)

− 0.15

− 0.06

 

Cluster tendency

 

0.50

0.68

0.70

(a)

0.48

0.21

0.14

 

Contrast

 

0.69

0.69

0.80

− 0.15

− 0.44

0.39

0.32

 

Correlation

 

0.63

0.77

0.73

0.34

0.66

− 0.35

− 0.26

 

Difference average

 

0.71

0.73

0.77

− 0.11

− 0.39

0.37

0.48

 

Difference entropy

 

0.71

0.75

0.79

− 0.12

− 0.39

0.35

0.50

 

Difference variance

 

0.68

0.76

0.85

− 0.21

− 0.51

0.40

0.30

 

Inverse Difference (ID)

(a)

0.72

0.72

0.78

(a)

0.36

− 0.32

− 0.47

 

Inverse Difference Moment (IDM)

(a)

0.72

0.72

0.78

(a)

0.36

− 0.31

− 0.47

 

Inverse Difference Moment Normalized (IDMN)

 

0.70

0.70

0.79

0.14

0.43

− 0.39

− 0.33

 

Inverse Difference Normalized (IDN)

 

0.71

0.74

0.77

(a)

0.38

− 0.36

− 0.49

 

Informational Measure of Correlation (IMC) 1

 

0.66

0.67

0.70

− 0.30

− 0.65

0.38

0.48

 

Informational Measure of Correlation (IMC) 2

 

0.65

0.70

0.68

0.32

0.67

− 0.31

− 0.12

 

Inverse Variance

(a)

0.72

(a)

0.78

(a)

0.35

− 0.31

− 0.10

 

Joint Average

 

0.47

0.42

0.58

0.32

0.29

0.21

− 0.29

 

Joint Energy

 

0.66

0.64

0.75

(a)

− 0.28

− 0.28

− 0.49

 

Joint Entropy

 

0.69

0.68

0.79

(a)

0.17

0.30

0.46

 

Maximal Correlation Coefficient (MCC)

 

0.48

0.73

0.75

(a)

0.60

− 0.43

− 0.61

 

Maximum Probability

(a)

0.60

0.61

0.75

− 0.12

− 0.36

− 0.30

− 0.49

 

Sum Average

 

0.47

0.42

0.58

0.32

0.29

0.21

− 0.29

 

Sum entropy

 

0.59

0.64

0.71

0.15

0.54

0.24

0.40

 

Sum squares

 

0.49

0.66

0.66

(a)

0.44

0.25

0.16

glszm

Gray Level Non Uniformity

 

0.88

0.89

0.92

0.72

0.50

− 0.56

− 0.62

 

Gray Level Non Uniformity Normalized

 

0.54

0.65

0.72

− 0.15

− 0.59

− 0.29

− 0.52

 

Gray Level Variance

 

0.46

0.64

0.72

(a)

0.46

0.39

0.43

 

High Gray Level Zone Emphasis

 

0.52

0.29

0.49

0.30

0.40

0.21

− 0.43

 

Large Area Emphasis

 

0.74

0.22

0.44

0.16

0.40

− 0.18

− 0.51

 

Large Area High Gray Level Emphasis

 

0.71

0.74

0.60

0.56

0.60

− 0.17

− 0.50

 

Large Area Low Gray Level Emphasis

 

0.43

0.88

0.96

− 0.15

0.16

(a)

− 0.17

 

Low Gray Level Zone Emphasis

 

0.25

0.72

0.65

− 0.35

(a)

− 0.13

− 0.04

 

Size Zone Non Uniformity

 

0.89

0.74

0.90

0.64

0.71

− 0.29

− 0.41

 

Size Zone Non Uniformity Normalized

 

0.75

− 0.38

0.78

− 0.18

− 0.39

0.17

− 0.49

 

Small Area Emphasis

 

0.75

− 0.37

0.78

− 0.18

− 0.39

0.17

− 0.49

 

Small Area High Gray Level Emphasis

 

0.52

0.18

0.50

0.25

0.33

0.22

− 0.47

 

Small Area Low Gray Level Emphasis

 

0.22

0.40

0.29

− 0.40

− 0.15

− 0.21

− 0.29

 

Zone Entropy

(a)

0.73

0.28

0.75

0.40

0.66

0.12

0.52

 

Zone Percentage

 

0.75

0.68

0.79

− 0.16

− 0.39

0.22

0.24

 

Zone Variance

 

0.74

0.20

0.43

0.15

0.40

− 0.17

− 0.51

glrlm

Gray Level Non Uniformity

 

0.89

0.88

0.89

0.66

0.50

− 0.61

− 0.62

 

Gray Level Non Uniformity Normalized

 

0.55

0.62

0.74

− 0.16

− 0.59

− 0.24

− 0.49

 

Gray Level Variance

 

0.49

0.53

0.66

(a)

0.47

0.36

0.28

 

High Gray Level Run Emphasis

 

0.52

0.37

0.52

0.31

0.39

0.21

− 0.36

 

Long Run Emphasis

 

0.74

0.63

0.75

0.15

0.39

− 0.26

− 0.33

 

Long Run High Gray Level Emphasis

(a)

0.50

0.68

0.84

0.36

0.47

(a)

− 0.32

 

Long Run Low Gray Level Emphasis

(a)

0.29

0.98

0.96

− 0.29

(a)

(a)

− 0.02

 

Low Gray Level Run Emphasis

 

0.24

0.78

0.75

− 0.34

(a)

0.43

0.33

 

Run Entropy

 

0.60

0.66

0.71

0.23

0.61

0.20

0.20

 

Run Length Non Uniformity

 

0.98

0.71

0.98

0.79

0.80

− 0.47

− 0.24

 

Run Length Non Uniformity Normalized

 

0.75

0.68

0.80

− 0.16

− 0.39

0.21

0.26

 

Run Percentage

 

0.75

0.69

0.79

− 0.16

− 0.39

0.21

0.31

 

Run Variance

 

0.74

0.55

0.73

0.15

0.39

− 0.26

− 0.34

 

Short Run Emphasis

 

0.75

0.68

0.79

− 0.16

− 0.39

0.21

0.24

 

Short Run High Gray Level Emphasis

 

0.52

0.18

0.48

0.29

0.37

0.23

− 0.02

 

Short Run Low Gray Level Emphasis

 

0.23

0.70

0.62

− 0.36

(a)

0.28

0.16

  1. The columns have the following meanings: w: results of paired Wilcoxon test between FOR-PROCESSING and FOR PRESENTATION: as regads dense-area non-significant (p > 0.05) test have been marked with “(a)”; as regads no-dense area all features were significantly different; r: Spearman’s correlation coefficient (they are all statistically significant, p < 0.05); CanCor: canonical correlation (see text for details); PD-FOR-PRES and PD-FOR-PROC: Spearman’s correlation coefficient between PD and features computed on FOR-PRES and FOR-PROC respectively, non-significant correlation (p > 0.05) have been marked with (a). For feature named ‘Minimum’ it was not possible to compute Spearman correlation on no-dense area because the value is always 0