EURASIP Journal on Image and Video Processing

Table 3 Performance analysis of all algorithms on STARE databases with respect to the measuring metrics

From: Retinal vessel segmentation with constrained-based nonnegative matrix factorization and 3D modified attention U-Net

Methods	Accuracy	Specificity	Sensitivity	Precision
U-Net [1]	0.9561	0.9765	0.7758	0.7885
AG-UNet [35]	0.9636	0.9778	0.8437	0.8188
IterNet [26]	0.9517	0.9575	0.9027	0.7157
DenseNet [36]	0.9495	0.9503	0.9424	0.6925
V-GAN [37]	0.9482	0.9475	0.9540	0.6832
Ours	0.9703	0.9801	0.8881	0.8408

Back to article page