TABLE 2.

The results of comparable experiments and ablation studies on OD segmentation of premature infants.

Methods	Dsc (%)	Sen (%)	Parameters (M)
FCN (Long et al., 2015)	89.42	86.95	18.64
DeepLabV3 (Chen et al., 2017)	85.65	90.07	58.16
DANet (Fu et al., 2020)	89.49	91.26	49.48
GCN (Peng et al., 2017)	88.81	89.53	23.62
PSPNet (Zhao et al., 2017)	91.29	91.68	27.76
U-Net (Ronneberger et al., 2015)	89.87	93.56	7.76
Att-UNet (Oktay et al., 2018)	90.91	95.48	8.73
CE-Net (Gu et al., 2019)	91.22	93.74	29.00
CPFNet (Feng et al., 2020)	91.40	89.34	43.27
UNet++ (Zhou et al., 2018)	90.92	92.51	9.16
CS2Net (Mou et al., 2021)	90.64	90.24	8.93
HRSeNet (Wang et al., 2019)	90.88	93.03	1.63
UNet + DsSE + MsFF	0.9130	0.9257	8.09
TransUNet (Chen et al., 2021)	0.9012	0.9314	105.28
Baseline	91.20	91.70	21.66
Baseline + DsSE	92.58	92.35	21.93
Baseline + MsFF	92.75	93.34	21.91
AFENet	93.12	93.22	22.18

Bold values are indicate the best performance.