TSHVNet: Simultaneous Nuclear Instance Segmentation and Classification in Histopathological Images Based on Multiattention Mechanisms

Table 5

_scores of various nuclear categories on different combinations of SimAM modules.

Method	Dataset metrics
	CoNSeP					PanNuke
	HoVer-Net	0.738	0.618	0.564	0.532	0.348	0.790	0.465	0.413	0.153	0.463	0.591
HoVer-Net+Res	0.761	0.655	0.622	0.579	0.393	0.809	0.495	0.440	0.169	0.470	0.599
HoVer-Net+Des	0.760	0.666	0.615	0.570	0.360	0.806	0.480	0.445	0.168	0.471	0.563
HoVer-Net+Res+Des	0.763	0.659	0.601	0.562	0.356	0.809	0.502	0.445	0.168	0.470	0.591
TSHVNet (Res)	0.757	0.647	0.598	0.553	0.362	0.815	0.528	0.455	0.176	0.530	0.620
TSHVNet (Des)	0.759	0.650	0.600	0.560	0.360	0.816	0.527	0.454	0.177	0.532	0.621
TSHVNet	0.763	0.669	0.621	0.583	0.358	0.820	0.531	0.460	0.179	0.536	0.623