Fig. 2

Radiomics feature selection using least absolute shrinkage and selection operator (LASSO) logistic regression. a Tuning parameter (λ) selection by 10-fold cross-validation with minimum criteria. Binomial deviance (y-axis) was plotted against log(λ) (x-axis). The dotted vertical lines were drawn at the optimal value of λ, where the model provided the best fit to the data. The optimal value of λ was 0.039, and the corresponding value of log(λ) = − 3.244. b LASSO coefficient profiles of the whole features set. The dotted vertical line was plotted at the value selected with 10-fold cross-validation, where 22 optimal features with non-zero coefficients were indicated in the plot