Fig. 3

Calinski Harabasz scoring plot (A) to determine the optimal number of subareas (habitats). These values are averaged over 100 replicates. (B) Habitat clustering, demonstrating the three habitats defined by the normalized. (C-D) Selection of Tuning Parameters (Lambda) for the Least Absolute Shrinkage and Selection Operator (LASSO) Regression Model. (E) Thirty-one Radiomic Features and their Weights Selected by the LASSO Regression Algorithm