Development of a modified 3D region proposal network for lung nodule detection in computed tomography scans: a secondary analysis of lung nodule datasets

Table 3 Performance comparison of the modified 3D RPN trained on three datasets^†

Number of false positives per scan	Sensitivity							CPM
Number of false positives per scan	0.125	0.25	0.5	1	2	4	8	CPM
Data source
LUNA16	77.0%±2.9%	84.3%±2.6%	89.4%±2.4%	92.5%±1.9%	94.6%±1.5%	96.0%±1.0%	96.6%±0.7%	90.1%±1.4%
LNOP	45.9%±5.4%	58.3%±4.1%	67.4%±3.3%	76.9%±3.4%	84.8%±3.5%	91.0%±2.5%	94.6%±2.5%	74.1%±3.1%
LNHE	46.9%±3.4%	54.6%±5.5%	63.1%±4.3%	71.7%±3.0%	79.7%±3.2%	85.7%±2.2%	89.7%±2.9%	70.2%±2.1%

^†Values represent the pulmonary nodule detection sensitivities (%) according to the number of false positives per CT scan
The results were expressed as mean ± SD.

ISSN: 1470-7330