Table 5. Performance benchmarking with... | Emerald Publishing

Table 5.

Performance benchmarking with weakly and fully supervised methods in the CoNSeP data set

Method	AJI	Dice	PQ
Weakly supervised
Pseudoedgenet (Yoo et al., 2019)	0.221	0.331	0.153
BoNuS (Lin et al., 2024)	0.354	0.651	0.380
Partial points (Qu et al., 2019)	0.366	0.646	0.391
Point annotations (Tian et al., 2020)	0.464	0.749	0.398
DAWN (Zhang et al., 2024)	0.509	0.805	0.477
Fully supervised
U-Net (Ronneberger et al., 2015)	0.499	0.761	0.434
HoVer-Net (Graham et al., 2019)	0.513	0.837	0.492
CDNet (He et al., 2021)	0.541	0.835	0.514
Mulvernet (Vo and Kim, 2023)	0.515	0.833	0.482
LG-NuSegHop (baseline)	0.422	0.654	0.407
LG-NuSegHop (dom. Adapted)	0.461	0.691	0.427

Method	AJI	Dice	PQ
Weakly supervised
Pseudoedgenet (Yoo et al., 2019)	0.221	0.331	0.153
BoNuS (Lin et al., 2024)	0.354	0.651	0.380
Partial points (Qu et al., 2019)	0.366	0.646	0.391
Point annotations (Tian et al., 2020)	0.464	0.749	0.398
DAWN (Zhang et al., 2024)	0.509	0.805	0.477
Fully supervised
U-Net (Ronneberger et al., 2015)	0.499	0.761	0.434
HoVer-Net (Graham et al., 2019)	0.513	0.837	0.492
CDNet (He et al., 2021)	0.541	0.835	0.514
Mulvernet (Vo and Kim, 2023)	0.515	0.833	0.482
LG-NuSegHop (baseline)	0.422	0.654	0.407
LG-NuSegHop (dom. Adapted)	0.461	0.691	0.427