Figure 2.

Standardized dosiomics features (LASSO-routine, LET_d, D_LEM, D_MKM from top to bottom) as stratified by r-Cox according to the risk (low-risk in blue, high-risk in red) of LR. Features highlighted in bold showed statistically significant differences between the two classes (Mann–Whitney U test (α = 0.05)) and the obtained p-values are reported in brackets. Each label in the boxplot is structured as feature type (i.e., ‘shape’ for shape features, ‘firstorder’ for first order features and matrix name (e.g., ‘glcm’) for texture features) followed by the specific feature name according to PyRadiomics convention. Refer to supplementary materials Section S2 for more details.