Classification of nasal polyps and inverted papillomas using CT-based radiomics

Insights into Imaging

Table 3 Average performance of various machine learning models

Feature selection	Model name	AUC	Accuracy	Sensitivity	Precision	Specificity	F1 score
No	SVM	0.8969	0.8883	0.8270	0.9644	0.9668	0.8842
	Naive Bayes	0.8604	0.8540	0.8543	0.8724	0.8666	0.8574
	XGBoost	0.9220	0.9153	0.8510	0.9929	0.9929	0.9143
Boruta	SVM	0.9068	0.8984	0.8260	0.9838	0.9876	0.8938
	Naive Bayes	0.9013	0.8916	0.8602	0.9368	0.9424	0.8901
	XGBoost	0.9078	0.9018	0.8573	0.9568	0.9582	0.9017
Random forest	SVM	0.9077	0.9018	0.8512	0.9644	0.9643	0.8995
	Naive Bayes	0.8872	0.8777	0.8466	0.9220	0.9278	0.8770
	XGBoost	0.9184	0.9120	0.8896	0.9443	0.9473	0.9122
Correlation coefficient	SVM	0.8803	0.8714	0.8263	0.9310	0.9342	0.8682
	Naive Bayes	0.8659	0.8576	0.8407	0.8842	0.8911	0.8567
	XGBoost	0.9141	0.9086	0.8707	0.9603	0.9575	0.9093