Weakly supervised segmentation models as explainable radiological classifiers for lung tumour detection on CT images

Insights into Imaging

Table 1 Clinical characteristics of the study population. Stage represents the clinical stage in the training data and the pathological stage in test data

Variable	Value	N (train)	N (test)
Institution	Maastricht University Medical Centre	421	0
	Stanford University Medical School Hospital	0	57
	Veterans Association Palo Alto	0	85
Age (years)
[Range]	20–40	1	0
	41–60	84	17
	61–80	273	112
	81–100	41	13
Sex	Female	132	36
Sex	Male	289	106
T-stage	1	93	68
	2	155	48
	3	53	16
	4	117	5
	Unknown	3	5
N-stage	0	170	113
	1	23	12
	2	140	17
	3	85	0
	Unknown	3	0
M-stage	0	416	138
	1	1	4
	Unknown	4	0
Histology	Adenocarcinoma	51	111
	Large cell carcinoma	114	0
	Squamous cell carcinoma	151	28
	Other	63	3
Scanner manufacturer	GE Medical Systems	0	116
	Philips	0	2
	Siemens	421	14
	Toshiba	0	1