Fig. 3

Radiomics feature selection using the Least Absolute Shrinkage and Selection Operator (LASSO) binary logistic regression model. (a) Lasso Coefficient Path: This plot illustrates the progression of LASSO coefficients as the regularization parameter (λ) increases. As λ becomes larger, the coefficients of less significant features are progressively shrunk towards zero, effectively performing feature selection. (b) Lasso Cross-Validation Curve: This curve displays the cross-validation error as a function of λ. The optimal λ is identified at the point where the cross-validation error is minimized (right dashed line). A feature set is selected based on one standard error from the minimum λ (left dashed line), ensuring a balance between model generalization and the prevention of overfitting.