Detection of pulmonary ground-glass opacity based on deep learning computer artificial intelligence

Table 2 Classification results of GGOs and non-GGOs by ResNet and pre-trained ResNet

	Threshold	TP	FN	FP	TN	TPR	Precision	F-score
Pre-trained ResNet	0.1	262	31	43	257	0.894198	0.8590164	0.876254
	0.2	255	38	38	262	0.870307	0.8703072	0.870307
	0.3	255	38	33	267	0.870307	0.8854167	0.877797
	0.4	255	38	33	267	0.870307	0.8854167	0.877797
	0.5	253	40	31	269	0.863481	0.8908451	0.87695
	0.6	252	41	29	271	0.86007	0.8968	0.87805
	0.7	250	43	28	272	0.853242	0.8992806	0.875657
	0.8	246	47	25	275	0.83959	0.9077491	0.87234
	0.9	244	49	23	277	0.832765	0.9138577	0.871429
ResNet	0.1	270	23	79	221	0.921502	0.773639	0.841121
	0.2	269	24	72	228	0.918089	0.7888563	0.84858
	0.3	267	26	66	234	0.911263	0.8018018	0.853035
	0.4	265	28	63	237	0.904437	0.8079268	0.853462
	0.5	263	30	59	241	0.89761	0.81677	0.85528
	0.6	260	33	56	244	0.887372	0.8227848	0.853859
	0.7	258	35	53	247	0.880546	0.829582	0.854305
	0.8	256	37	51	249	0.87372	0.8338762	0.853333
	0.9	250	43	46	254	0.853242	0.8445946	0.848896

FN false negative, FP false positive, TN true negative, TP true positive, TPR true positive rate

ISSN: 1475-925X