SERR-U-Net: Squeeze-and-Excitation Residual and Recurrent Block-Based U-Net for Automatic Vessel Segmentation in Retinal Image

Table 4

Comparing results of different ablation experiments on the 10 STARE images.

Methods	AUC	F1-score
U-Net	0.9793	0.8360
Recurrent+U-Net	0.9799	0.8383
SE-ResNet+U-Net	0.9834	0.8372
Recurrent+ResNet+U-Net (R2-U-Net)	0.9856	0.8474
Ours (SERR-U-Net)	0.9859	0.8478