Fig. 2

Radiomics features selection using tenfold cross-test and LASSO regression. A Tuning parameter(λ) selection in the LASSO regression used tenfold cross-test based on the minimum criterion. Dotted vertical lines indicated the optimal values using the minimum criteria and the1-SE criteria. A λ value of 0.024(log(λ) = -3.715) was chosen (minimum criterion) according to tenfold cross-test. B A coefficient profile plot was produced against the log(λ) sequence. Dotted vertical lines indicated the value obtained by the above tenfold cross-test, which resulted in 13 radiomics features with nonzero coefficients. C After tenfold cross-test and LASSO regression, the name and coefficient of selected radiomics features showed by a bar diagram