Histopathological Tissue Segmentation of Lung Cancer with Bilinear CNN and Soft Attention

Table 4

The classification F1 score of tissue types on colorectal cancer dataset.

Model	TUM	STR	LYM	MUC	MUS	NOR	BAC	DEB	ADI
VGG19 [30]	0.9870	0.9266	0.9773	0.9748	0.9662	0.9720	0.9586	0.9642	0.9580
DeepTissue Net [14]	0.9806	0.9413	0.9722	0.9784	0.9742	0.9806	0.9988	0.9711	0.9958
EfficientNet [31]	0.9814	0.9569	0.9827	0.9665	0.9671	0.9818	0.9982	0.9718	0.9942
ResNet50 [29]	0.9756	0.9243	0.9673	0.9852	0.9568	0.9866	0.9978	0.9729	0.9972
ResNet50+bilinear pooling module	0.9787	0.9502	0.9808	0.9614	0.9602	0.9869	0.9978	0.9771	0.9940
ResNet50+attention module	0.9803	0.9554	0.9810	0.9667	0.9826	0.9851	0.9974	0.9673	0.9940
ResNet50+bilinear pooling module+attention module	0.9860	0.9591	0.9845	0.9763	0.9693	0.9879	0.9982	0.9830	0.9966