Skip to main content
EJNMMI Reports

Table 5 Univariate and multivariate analyses of overall survival

From: Prognostic analysis of curatively resected pancreatic cancer using harmonized positron emission tomography radiomic features

 

Univariate analysis

Multivariate analysis

Hazard ratio (95% CI)

p value

Hazard ratio (95% CI)

p value

Neural invasion

0.05–0.95

0.043

0.03–0.73

0.019

SUVmax

0.08–0.98

0.047

0.08–7.06

0.81

TLG

0.23–1.13

0.095

0.15–3.11

0.63

Shape sphericity

0.09–0.99

0.049

0.03–0.94

0.042

Histogram kurtosis

0.12–0.92

0.034

0.08–3.54

0.50

GLCM contrast

0.16–0.99

0.048

0.01–5.93

0.35

GLCM homogeneity

0.16–1.10

0.078

0.07–21.16

0.89

GLCM dissimilarity

0.16–0.99

0.048

NA*

NA*

GLRLM HGRE

0.15–1.10

0.076

0.05–79.60

0.74

GLRLM SRHGE

0.15–1.10

0.076

NA*

NA*

GLRLM GLNU

0.22–1.10

0.082

0.62–7.51

0.23

GLZLM HGZE

0.12–0.97

0.043

0.02–30.49

0.91

GLZLM SZHGE

0.12–0.97

0.043

NA*

NA*

GLZLM LZLGE

0.23–1.10

0.085

0.05–0.63

0.0076

GLZLM GLNU

0.20–1.01

0.054

0.09–2.07

0.29

  1. CI confidence interval, SUVmax maximum standardized uptake value, TLG total lesion glycolysis, GLCM grey-level co-occurrence matrix, GLRLM grey-level run length matrix, HGRE high grey-level run emphasis, SRHGE short-run high grey-level emphasis, (GLRLM) GLNU grey-level nonuniformity for run, LZLGE long-zone low grey-level emphasis, GLZLM grey-level zone length matrix, HGZE high grey-level zone emphasis, SZHGE short-zone high grey-level emphasis, LZLGE long-zone low grey-level emphasis, (GLZLM) GLNU grey-level nonuniformity for zone, NA* not available due to multicollinearity between (1) GLCM contrast and GLCM dissimilarity (Pearson correlation coefficient r = 0.95 in pre-harmonization and 0.97 in post-harmonization), (2) GLRLM HGRE and GLRLM SRHGE (r = 0.9996 in pre-harmonization and 0.9993 in post-harmonization), and (3) GLZLM HGZE and GLZLM SZHGE (r = 0.99 in pre-harmonization and 0.99 in post-harmonization)
Back to article page