Preoperative CT-based deep learning radiomics model to predict lymph node metastasis and patient prognosis in bladder cancer: a two-center study

Insights into Imaging

Table 1 Baseline information of the patients

		Training set (n = 185)	External test set (n = 54)	p
Age (years)		68.04 ± 10.00	64.37 ± 12.53	0.052
Gender	Male	155	48	0.356
Gender	Female	30	6
Location	Bladder triangle	52	12	0.390
Location	Other part of the bladder	133	42
Shape	Cauliflower-like	95	31	< 0.001
	Papillary	31	19
	Mound-like	59	4
Size		4.44 ± 2.17	2.58 ± 1.12	< 0.001
Calcification	No	156	46	0.878
Calcification	Yes	29	8
Cystic necrosis	No	153	49	0.151
Cystic necrosis	Yes	32	5
Tumor boundary	Clear	115	49	< 0.001
Tumor boundary	Unclear	70	5
Number	Solitary	149	43	0.882
Number	Multiple	36	11
Stalk	absent	148	27	< 0.001
Stalk	present	37	27
CT reported T stage	Ta-T2	138	45	0.182
CT reported T stage	T3-T4	47	9
CT reported LN status	N0	161	47	0.998
CT reported LN status	N1-3	24	7
CT value in corticomedullary phase		64.17 ± 24.87	75.87 ± 21.52	0.002
CT value in nephrographic phase		75.86 ± 20.93	81.52 ± 17.11	0.071
CT value in excretory phase		74.75 ± 17.98	90.35 ± 50.67	0.030