High precision localization of pulmonary nodules on chest CT utilizing axial slice number labels

BMC Medical Imaging

Table 1 NLST test set patient demographics

Variable	Missed/incorrect/unintended nodule predictions^a n (%)	Correct nodule predictions^a n (%)	p	Totals n (%)
Patients	15 (100%)	32 (100%)	–	47 (100%)
Age			0.586
(mean ± SD)	61.07 ± 4.74	61.91 ± 5.12		61.64 ± 4.97
Sex			0.806
Female	5 (33%)	8 (25%)		13 (28%)
Male	10 (67%)	24 (75%)		34 (68%)
Race			0.180
White	12 (80%)	31 (97%)		43 (91%)
Black	1 (7%)	0 (0%)		1 (2%)
Asian	1 (7%)	1 (3%)		2 (4%)
> 1 Race	1 (7%)	0 (0%)		1 (2%)
BMI			0.256
(mean ± SD)	29.72 ± 6.15	27.56 ± 5.35		28.25 ± 5.64
Smoker at start of NLST			0.468
Yes	6 (40%)	18 (56%)		24 (51%)
No (former smoker)	9 (60%)	14 (44%)		23 (49%)
Cigarettes/day			0.251
(mean ± SD)	32.67 ± 20.08	26.22 ± 9.00		28.28 ± 13.65
Smoking total years			0.978
(mean ± SD)	39.87 ± 8.68	39.94 ± 7.03		39.76 ± 7.41
Smoking pack years			0.320
(mean ± SD)	63.53 ± 54.91	51.62 ± 17.72		56.37 ± 34.33

Adjusted nodule performance with the highest recall score at a 20 mm distance threshold was used to split missed/incorrect/unintended and correct nodule predictions
SD standard deviation
^aFor patients with > 1 nodule, if at least one nodule was correctly identified for that patient, this patient was classified as a correct prediction

ISSN: 1471-2342