Table 8. The quantitative results showing percentages improvements of the model after adding additional components to UNET (Ronneberger, Fischer & Brox, 2015) structure.

D-UNET denotes dense networks with UNET structure, DID-UNET represents dense networks and improved dilation convolution to the structure of UNET, and ADID-UNET refers to proposed model with dense networks improved dilation convolution and attention gate modules to the UNET structure. ↑ indicates that the performance index is higher than that of UNET structure, ↓ indicates that the performance index is lower than that of UNET structure.

Method	ACC	DC	Sen	Sp	Pc	AUC	F1	Sm	Eα	MAE
UNET	0.9696	0.7998	0.8052	0.9957	0.8247	0.9347	0.8154	0.8400	0.9390	0.0088
D-UNET	0.9700	0.8011	0.8096	0.9966	0.8596	0.9492	0.8184	0.8528	0.9394	0.0083
DID-UNET	0.9700	0.8023	0.7987	0.9964	0.8425	0.9549	0.8241	0.8447	0.9374	0.0084
ADID-UNET	0.9701	0.8031	0.7973	0.9966	0.8476	0.9551	0.8200	0.8509	0.9449	0.0082
Improvement of D-UNET	↑0.04%	↑0.13%	↑0.44%	↑0.09%	↑3.49%	↑1.45%	↑0.30%	↑1.28%	↑0.04%	↓0.05%
Improvement of DID-UNET	↑0.04%	↑0.25%	↓0.65%	↑0.07%	↑1.78%	↑2.02%	↑0.87%	↑0.47%	↓0.16%	↓0.04%
Improvement of ADID-UNET	↑0.05%	↑0.33%	↓0.79%	↑0.09%	↑2.29%	↑2.04%	↑0.46%	↑1.09%	↑0.59%	↓0.06%