Skip to main content

Table 3 Comparisons of radiomics features between the training, testing and validation cohorts

From: CT-based radiomics models may predict the early efficacy of microwave ablation in malignant lung tumors

Different models

Training cohort (n = 72)

Testing cohort (n = 32)

Validation cohort (n = 26)

Effective treatment cohort

Ineffective treatment cohort

p value

Effective treatment cohort

Ineffective treatment cohort

p value

Effective treatment cohort

Ineffective treatment cohort

p value

T-RO

stdDeviation

-0.166 (-1.096, 0.616)

0.584 ± 1.089

0.017*

0.272 ± 1.127

1.357 ± 0.854

0.035*

0.054 ± 1.068

-0.227 ± 0.844

0.590

Inertia_AllDirection_offset7_SD

-0.266 (-0.289, -0.243)

-0.251 (-0.266,0.949)

0.037*

-0.212 (-0.298, 0.315)

-0.009 (-0.266, 0.971)

0.356

-0.314 (-0.314, -0.233)

-0.114 (-0.332, 2.415)

0.477

P-RO

Inertia_angle0_offset1

-0.389 (-0.797, 0.315)

0.731 ± 1.344

0.011*

-0.330 ± 1.063

0.747 ± 1.173

0.036*

-0.816 ± 0.949

0.343 ± 1.345

0.414

LongRunLowGreyLevelEmphasis_angle45_offset4

-0.307 (-0.513, -0.142)

-0.208(-0.396,1.065)

0.039*

-0.320 (-0.505, -0.121)

-0.374 (-0.461, 0.753)

0.772

-0.253 (-0.918, 0.314)

0.192 ± 0.887

0.454

  1. *p < 0.05 (two-sided) was considered statistically significant