Proficiency evaluation of shape and WPT radiomics based on machine learning for CT lung cancer prognosis

Egyptian Journal of Radiology and Nuclear Medicine

Table 4 Classification results attained for nine different State-of-art classifiers using two FTFS techniques

Feature selection algorithm	Classifiers	(Malignant vs Benign)
Feature selection algorithm	Classifiers	AUROC	Accuracy	Sensitivity/Recall/TPR	Precision/PPV	Specificity/TNR
ANOVA	FGSVM	0.883	0.837	0.978	0.83	0.453
	MGSVM	0.904	0.864	0.923	0.895	0.706
	CGSVM	0.895	0.827	0.917	0.856	0.579
	Decision Trees	0.878	0.856	0.91	0.895	0.709
	Ensemble Boosted Trees (BOCET)	0.905	0.866	0.917	0.902	0.727
	Ensemble Bagged Trees (BACET)	0.910	0.866	0.905	0.912	0.761
	Ensemble RUSBoosted Trees (RUSBOCET)	0.914	0.852	0.861	0.932	0.829
	Ensemble Subspace Discriminant	0.902	0.843	0.948	0.854	0.559
	Ensemble Subspace KNN	0.909	0.869	0.929	0.896	0.706
Chi-Square test	FGSVM	0.911	0.865	0.957	0.872	0.615
	MGSVM	0.910	0.904	0.93	0.941	0.824
	CGSVM	0.910	0.869	0.922	0.901	0.723
	Decision Tree	0.883	0.857	0.926	0.885	0.67
	Ensemble Boosted Trees (BOCET)	0.923	0.877	0.923	0.91	0.75
	Ensemble Bagged Trees (BACET)	0.929	0.891	0.935	0.918	0.772
	Ensemble RUSBoosted Trees (RUSBOCET)	0.925	0.861	0.869	0.937	0.84
	Ensemble Subspace Discriminant	0.911	0.86	0.948	0.872	0.621
	Ensemble Subspace KNN	0.896	0.865	0.93	0.891	0.689