Skip to main content

Table 2 Multivariable models for the prediction of tumor growth speed

From: Multi-window CT based Radiomic signatures in differentiating indolent versus aggressive lung cancers in the National Lung Screening Trial: a retrospective study

Covariate

Radiomics features

Radiomics features combined with demographics

mOR (95% CI)

Bootstrap mOR

(95% CI)

Bootstrap P value

Covariate

mOR (95% CI)

Bootstrap mOR

(95% CI)

Bootstrap P value

Lung Window Mask

LW_F8

1.00 (1.00–1.01)

1.00 (1.00–1.01)

.004

Radio_LL

2.22 (1.51–3.28)

2.40 (1.38–4.16)

<.001

LW_F87

0.77 (0.67–0.88)

0.77 (0.67–0.88)

<.001

Sex

2.37 (0.99–5.68)

2.38 (0.91–6.22)

.073

LW_F311

1903.06 (4.87–7.44 × 105)

1903.06 (4.87–7.44 × 105)

.016

COPD

5.78 (0.68–49.04)

1.16 × 1016 (0–5.20 × 1056)

.018

AUC

0.800 (0.717–0.883)

0.799 (0.717–0.883)

 

AUC

0.813 (0.735–0.890)

0.808 (0.727–0.888)

 

Accuracy

81.33%

81.33%

 

Accuracy

84.00%

84.00%

 

Specificity

66.67%

66.67%

Specificity

58.97%

58.97%

Sensitivity

86.49%

86.49%

Sensitivity

92.79%

92.79%

Difference Region Mask

DR_F8

1.01 (1.00–1.02)

1.01 (1.00–1.02)

<.001

Radio_DR

2.27 (1.56–3.31)

2.42 (1.45–4.03)

<.001

DR_F87

0.80 (0.70–0.92)

0.78 (0.64–0.96)

<.001

Sex

2.47 (1.00–6.08)

2.54 (0.95–6.81)

.055

DR_F286

0.89 (0.80–0.99)

0.89 (0.78–1.00)

.042

COPD

6.47 (0.72–58.03)

2.41 × 1016 (0–1.17 × 1057)

.006

DA_F311

836.55 (1.75–3.99 × 106)

1499.14 (1.34–1.68 × 106)

.038

    

AUC

0.820 (0.743–0.896)

0.819 (0.742–0.896)

 

AUC

0.837 (0.767–0.907)

0.828 (0.753–0.903)

 

Accuracy

73.33%

82.00%

 

Accuracy

84.00%

84.67%

Specificity

79.49%

61.54%

Specificity

69.23%

69.23%

Sensitivity

71.17%

89.19%

Sensitivity

89.19%

90.09%

Combined Windows Mask

LW_F44

66,379.49 (0.64–6.89 × 109)

5.14 × 105 (1.11–2.37 × 105)

.020

Radio_Comb

2.31 (1.62–3.29)

2.49 (1.46–4.22)

<.001

LW_F311

1406.04 (2.53–7.83 × 105)

2903.14 (1.66–5.06 × 106)

.030

Sex

2.17 (0.88–5.35)

2.23 (0.85–5.84)

.008

DR_F8

1.01 (1.00–1.01)

1.01 (1.00–1.01)

.003

COPD

6.90 (0.76–62.40)

6.53 × 1016 (0–6.87 × 1057)

.017

DR_F87

0.80 (0.70–0.93)

0.78 (0.63–0.98)

.005

    

DR_F296

0.85 (0.75–0.96)

0.83 (0.70–0.99)

.017

    

AUC

0.845 (0.770 to 0.920)

0.846 (0.772 to 0.920)

 

AUC

0.861 (0.791 to 0.931)

0.855 (0.781 to 0.929)

 

Accuracy

83.33%

84.00%

 

Accuracy

84.67%

84.67%

 

Specificity

79.49%

79.49%

 

Specificity

76.92%

76.92%

 

Sensitivity

84.68%

85.59%

Sensitivity

87.39%

87.39%

  1. Note. F8 means statistical 10th percentile, F19 means root mean square, F44 means volume at intensity fraction 90, F87 means weighted CoM_z, F296 means 3D Wavelet P1L2C10 feature, F311 means voxel dimension x, Radio_LL means radiomics score calculated by the selected lung window features; Radio_DR means radomics score calculated from the selected difference region features; Radio_comb means radiomics score calculated from the selected lung window and difference region features
  2. NA, not available. These variables were eliminated in the multivariable logistic regression mode, so the OR and P values were not available