BMC Medical Informatics and Decision Making

Table 1 Represents the model comparison of each component on the test set, and k denotes the convolutional filters are used to extract features

From: MSAL-Net: improve accurate segmentation of nuclei in histopathology images by multiscale attention learning network

Method	MoNuseg test set
Method	K	DC	IOU	PR	RC	Training Time
Unet [23] + Res-18 [35]	–	0.781	0.642	0.774	0.792	211.9
Backbone + Res-34 [35]	–	0.793	0.652	0.785	0.801	–
Backbone + DDC	–	0.810	0.682	0.783	0.845	–
Backbone + Res-34 + DDC	–	0.817	0.692	0.790	0.837	–
Backbone + ECA	3	0.809	0.681	0.778	0.829	–
Backbone + ECA + BR	3	0.815	0.688	0.805	0.827	–
Backbone + DDC + ECA	3	0.820	0.696	0.809	0.848	–
Backbone + Res-18 + DDC + ECA + BR	3	0.829	0.699	0.813	0.846	181.3
MSAL-Net + Res-34	3	0.839	0.706	0.821	0.853	194.5
MSAL-Net + Res-50	3	0.847	0.713	0.816	0.881	317.7
MSAL-Net + Rest-101	3	0.830	0.695	0.801	0.862	429.1

Back to article page

ISSN: 1472-6947

Contact us

General enquiries: journalsubmissions@springernature.com