Fig. 2

Graphical overview of the eight image texture features: a glrlm_SRLGLE_90_2.5, b glrlm_SRLGLE_90_40, c glcm_dissimilarity_0_2.5, d Kurtosis_10, e glrlm_LRHGLE_90_2.5, f glrlm_SRE_0_40, g glcm_ASM_0_2.5, h Percentile_10th_40. For each texture feature, a violin plot shows the distribution of the feature quantified in all patient samples. Furthermore, tissue image patches corresponding to high and low values and the correlated texture feature map are shown. Two tissue image patches are taken from high-risk and low-risk patients. The color coding (i.e., red, orange, green, and blue) of the feature map corresponds to the gray level (from low to high) of the image. The P-value is computed by the independent samples t-test. A significant difference in feature distribution between the high-risk and low-risk groups is observed for each texture feature. ***Represents P < 0.001, **represents P < 0.01, and *represents P < 0.05