Enhancing brain metastasis prediction in non-small cell lung cancer: a deep learning-based segmentation and CT radiomics-based ensemble learning model

Table 3 Comparisons of the performance of different prediction models in the training and validation cohorts

Model	Dataset	ACC (%)	SEN (%)	SPE (%)	PPV (%)	NPV (%)	OR	F1 Score	F1_Weighted Score	MCC
Radiomics Feature Model	TC	80.59	81.73	79.17	82.93	77.78	17	0.82	0.81	0.61
	VC1	79.5	76.83	82.28	81.82	77.38	15.39	0.79	0.79	0.59
	VC2	76.92	84.21	73.91	57.14	91.89	15.11	0.68	0.78	0.53
Clinical Feature Model	TC	67.02	57.69	78.57	76.92	60	5	0.66	0.67	0.37
	VC1	67.08	52.44	82.28	75.44	62.5	5.12	0.62	0.66	0.36
	VC2	58.46	57.89	58.70	36.67	77.14	1.95	0.45	0.60	0.15
Fusion Feature Model	TC	81.91	84.13	79.17	83.33	80.12	20.15	0.84	0.82	0.63
	VC1	83.23	82.93	83.54	83.95	82.5	24.66	0.83	0.83	0.66
	VC2	80.00	84.21	78.26	61.53	92.31	19.2	0.71	0.81	0.58

ISSN: 1470-7330