Automatic lung segmentation in routine imaging is primarily a data diversity problem, not a methodology problem

European Radiology Experimental

Table 4 Comparison to public systems

	Test datasets DSC for full volumes											DSC ± SD		HD95(mm) ± SD	MSD (mm) ± SD
	Public			Routine								DSC ± SD		HD95(mm) ± SD	MSD (mm) ± SD
Architecture	LTRC	LCTSC	VESS12	RRT	Atel	Emph	Fibr	Mass	PnTh	Trau	Norm	All(L)*	All	All	All
U-net(R-231)	0.99	0.94	0.98	0.97	0.97	0.99	0.97	0.98	0.99	0.97	0.97	0.98 ± 0.03	0.98 ± 0.03	3.14 ± 7.4	0.62 ± 0.93
ResU-net(R-231)	0.99	0.94	0.98	0.97	0.97	0.99	0.97	0.98	0.99	0.97	0.97	0.98 ± 0.03	0.98 ± 0.03	3.19 ± 7.35	0.64 ± 0.88
DRN(R-231)	0.98	0.94	0.98	0.95	0.96	0.99	0.97	0.98	0.98	0.96	0.97	0.97 ± 0.04	0.97 ± 0.06	6.22 ± 18.95	1.1 ± 2.54
Deeplab v3+ (R-231)	0.99	0.94	0.98	0.97	0.97	0.99	0.97	0.98	0.99	0.97	0.97	0.98 ± 0.03	0.98 ± 0.03	3.28 ± 7.52	0.65 ± 0.91
P-HNN	0.98	0.94	0.99	0.88	0.95	0.98	0.95	0.98	0.96	0.88	0.97	0.96 ± 0.09	0.94 ± 0.12	16.8 ± 36.57	2.59 ± 5.96
U-net(R-231)**	0.99	0.95	0.99	0.99	0.97	0.99	0.97	0.98	0.99	0.97	0.98	0.98 ± 0.01	0.98 ± 0.01	1.44 ± 1.09	0.35 ± 0.19
CIP**	0.99	0.94	0.99	0.96	0.90	0.99	0.92	0.98	0.99	0.86	0.99	0.98 ± 0.03	0.96 ± 0.05	4.65 ± 6.45	0.91 ± 1.09
CIP\#Cases**	96/105	19/24	17/20	13/20	2/2	2/2	4/4	2/2	2/2	1/3	1/7

A comparison to the segmentation algorithm of the chest imaging platform (CIP) and the trained P-HNN model is given. The results are expressed in mean and mean ± standard deviation for the Dice similarity coefficient (DSC), Robust Hausdorff distance (HD95), and mean surface distance (MSD)
*The LCTSC ground truth masks do not include high-density diseases, and the high number of LTRC test cases dominates the averaged results. Thus, “All(L)” (n = 167) is the mean over all cases that included LCTSC and LTRC, while “All” (n = 62) does not include the LCTSC and LTRC cases
**For these rows, only cases on which the CIP algorithm did not fail, and where the DSC was larger than 0 were considered (#Cases). For abbreviations, see Tables 1 and 2