Table 1.

Study characteristics

Study [Ref.]	Year	Country	Design	Nodule size, mm	Real-time diagnosis	Deep learning	Training set size	Validation set size
								M	B
Zhu et al. [18]	2013	China	R	4−52	No	No	464	148	77
Song et al. [14]	2015	China	R	NR	No	No	155	21	20
Wu et al. [15]	2016	China	R	NR	No	No	485	260	225
Yu et al. [16]	2017	China	P	>2	No	No	610	17	33
Thomas et al. [19]a	2017	USA	R	NR	No	No	410	11	61
Zhang et al. [17]	2019	China	R	≤25	No	No	1,238	118	708
Choi et al. [25]	2017	South Korea	P	≥5	Yes	Yes	−	43	59
Jeong et al. [27]	2019	South Korea	P	≥10	Yes	Yes	−	44	56
Yoo et al. [28]	2018	South Korea	P	≥5	Yes	Yes	−	50	67
Gitto et al. [26]	2019	Italy	R	18±7	Yes	Yes	−	14	48
Kim et al. [31]	2019	South Korea	R	1.2±0.9 (B)	Yes	Yes	−	86	132
				1.2±0.9 (M)
Gao et al. [20]	2018	China	R	17±14 (B)	No	Yes	3,700	239	103
				10±7 (M)
Song et al. [23]	2019	China	R	NR	No	Yes	6,228	180	187
Li et al. [7]	2019	China	R	NR	No	Yes	312,399	118	156
Ko et al. [21]	2019	China	R	10−20	No	Yes	594	100	50
Song et al. [22]	2019	South Korea	P	NR	No	Yes	1,358	50	50
Wang et al. [24]	2019	China	R	NR	No	Yes	5,007	242	109
Luo et al. [29]a	2018	China	NR	NR	No	Yes	NR	292	208
Guan et al. [30]	2019	China	R	NR	No	Yes	2,437	209	190

B, benign; M, malignant; NR, not reported; P, prospective; R, retrospective.

^aConference abstract.