A multifeature nuclear morphometric score (MFSp) accurately discriminates cancer vs. benign cell populations: AUC results for two one-step logistic regression models.

*<p>Model A: Five features selected by backwards elimination: (FeretY_ave, MaxDiameter_ave, Elongation_ave, Slope_ave, ODKurtosis_ave).</p>**<p>Model B: Five features (SumOD_sd, MaxDiameter_sd, TSD_sd, TEntropy_sd, No.MedDensityObjects_sd), selected after competition among all models with ≤5 covariates based on leave-one-out AUC.