Table 4.

Segmentation performance (in %) of all UNet models using CCA database DB2X.

Model	Acc	Sens	Spec	Prec	MCC	Dice	Jaccard
UNet	98.97 ± 0.53	81.58 ± 8.58	99.72 ± 0.27	92.20 ± 7.39	85.97 ± 5.77	86.07 ± 6.17	76.01 ± 8.59
UNet++	99.05 ± 0.38	85.82 ± 7.21	99.62 ± 0.31	90.04 ± 8.09	87.23 ± 5.43	87.48 ± 5.57	78.14 ± 8.01
UNet3P	98.93 ± 0.56	80.39 ± 9.09	99.73 ± 0.26	92.42 ± 7.25	85.43 ± 6.24	85.49 ± 6.68	75.18 ± 9.00
Fractal-UNet	98.91 ± 0.54	80.08 ± 8.67	99.72 ± 0.27	85.17 ± 6.56	91.96 ± 7.61	85.06 ± 6.29	74.68 ± 8.92
Squeeze-UNet	98.90 ± 0.63	79.67 ± 10.45	99.73 ± 0.27	84.94 ± 7.56	92.33 ± 7.09	84.94 ± 7.60	74.48 ± 9.98
Attention-UNet	99.01 ± 0.42	81.75 ± 7.94	99.74 ± 0.26	92.71 ± 7.20	86.38 ± 5.70	86.50 ± 5.94	76.65 ± 8.36