TABLE 1.

Comparison of AJI of different methods applied to the TCGA test set.

Organ	Bladder	Colorectal	Stomach	Breast	Kidney	Liver	Prostate	Overall
FCN-8 (Long et al., 2015)	0.5376	0.4018	0.5279	0.5598	0.5267	0.5045	0.5709	0.5171
Mask R-CNN (He et al., 2015)	0.5011	0.3814	0.6151	0.4913	0.5182	0.4622	0.5322	0.5002
U-Net (Ronneberger et al., 2015)	0.5403	0.4061	0.6529	0.4681	0.5426	0.4284	0.5888	0.5182
CNN3 (Kumar et al., 2017)	0.5217	0.5292	0.4458	0.5385	0.5732	0.5162	0.4338	0.5083
DIST (Naylor et al., 2019)	0.5971	0.4362	0.6479	0.5609	0.5534	0.4949	0.6284	0.5598
Stacked U-Net	0.6138	0.5188	0.5845	0.5605	0.5647	0.4594	0.5300	0.5474
U-Net (DLA)	0.6215	0.5322	0.5938	0.5747	0.5624	0.4642	0.5602	0.5584
A two-stage U-Net (Mahbod et al., 2019)	0.5706	0.4891	0.6545	0.5613	0.5755	0.4989	0.6316	0.5687
Two-stage learning U-Net (DLA) (Kang et al., 2019)	0.6285	0.5376	0.6620	0.6096	0.6024	0.5142	0.5720	0.5895
Ours	0.5926	0.5586	0.6541	0.5907	0.5926	0.5346	0.6521	0.5965

The best results are shown in bold.