DCAU-Net: dense convolutional attention U-Net for segmentation of intracranial aneurysm images

Visual Computing for Industry, Biomedicine, and Art

Table 3 Comparison of the performances of key components on the MICCAI 2020 ADAM testing set

Component	Dice	Sensitivity	Specificity
Ours	0.7455	0.7881	0.9996
UNet	0.4629	0.6077	0.9992
UNet + Dense	0.5424	0.6068	0.9995
UNet + CAM	0.6783	0.6254	0.9998
UNet + SAM	0.6426	0.7850	0.9997
UNet + CBAM	0.6813	0.7398	0.9997
UNet + Dense + CBAM	0.7292	0.7378	0.9996
Use BN	0.6612	0.6829	0.9996
Use ReLU activation	0.5731	0.6560	0.9993