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

Insights into Imaging

Table 4 Diagnostic performance of the radiomics models in independent test CT scans

Feature selection	Model name	Evaluation indicators	1	2	3	4	5	6	7	8	9	10
		Accuracy	0.9310	0.9333	0.8000	0.7586	0.8966	0.8966	0.9333	0.9000	0.8333	1.0000
	SVM	FN	2	2	6	7	3	3	1	1	3	0
		FP	0	0	0	0	0	0	1	2	2	0
		Accuracy	0.7931	0.9000	0.7667	0.7586	0.8276	0.8276	0.9333	0.9000	0.8667	0.9667
No	Naive Bayes	FN	2	3	6	5	3	2	1	0	2	0
		FP	4	0	1	2	2	3	1	3	2	1
		Accuracy	0.9310	0.9333	0.8333	0.8276	0.8966	0.9310	0.9667	0.9667	0.8667	1.0000
	XGBoost	FN	2	2	5	5	3	2	1	1	3	0
		FP	0	0	0	0	0	0	0	0	1	0
		Accuracy	0.9310	0.8667	0.8000	0.7931	0.8966	0.8966	0.9667	0.9333	0.9000	1.0000
	SVM	FN	2	3	6	6	3	3	1	1	3	0
		FP	0	1	0	0	0	0	0	1	0	0
		Accuracy	0.8966	0.9000	0.8000	0.8276	0.8966	0.8621	0.9667	0.9000	0.8667	1.0000
Boruta	Naive Bayes	FN	2	3	6	5	2	2	1	0	2	0
		FP	1	0	0	0	1	2	0	3	2	0
		Accuracy	0.9310	0.9333	0.8000	0.8276	0.8966	0.8966	0.9333	0.9333	0.8667	1.0000
	XGBoost	FN	2	2	5	5	3	2	1	1	2	0
		FP	0	0	1	0	0	1	1	1	2	0
		Accuracy	0.9655	0.9333	0.8000	0.8621	0.8276	0.8966	0.9333	0.9333	0.8667	1.0000
	SVM	FN	0	2	6	4	4	3	1	1	3	0
		FP	1	0	0	0	1	0	1	1	1	0
		Accuracy	0.8966	0.9000	0.8333	0.8276	0.7931	0.7931	0.9667	0.9000	0.8667	1.0000
Random forest	Naive Bayes	FN	2	3	5	5	4	3	1	0	2	0
		FP	1	0	0	0	2	3	0	3	2	0
		Accuracy	0.8966	0.9333	0.8333	0.8621	0.8966	0.9310	0.9667	0.9000	0.9000	1.0000
	XGBoost	FN	1	2	5	4	2	1	1	1	1	0
		FP	2	0	0	0	1	1	0	2	2	0
		Accuracy	0.8966	0.9000	0.8000	0.7931	0.8276	0.8966	0.9333	0.8333	0.8333	1.0000
	SVM	FN	2	2	6	6	4	3	1	1	3	0
		FP	1	1	0	0	1	0	1	4	2	0
		Accuracy	0.8621	0.9000	0.7667	0.7931	0.8276	0.7931	0.9333	0.9333	0.8333	0.9333
Correlation coefficient	Naive Bayes	FN	2	2	6	5	4	3	1	1	3	0
		FP	2	1	1	1	1	3	1	1	2	2
		Accuracy	0.9655	0.9333	0.8333	0.8276	0.8966	0.8966	0.9000	0.9333	0.9000	1.0000
	XGBoost	FN	1	2	5	5	3	2	1	1	1	0
		FP	0	0	0	0	0	1	2	1	2	0