High-Resolution Histopathological Image Classification Model Based on Fused Heterogeneous Networks with Self-Supervised Feature Representation

Table 2

Classification performance comparison of different models on TCGA dataset.

	Model	Acc	Precision	Recall	F1 score	AccImp
H	VGG	0.790	0.791	0.795	0.789	−
	GoogleNet	0.780	0.812	0.799	0.779	−
	ResNet	0.820	0.827	0.830	0.820	−
	DenseNet	0.820	0.830	0.830	0.820	−
J	Model 1	0.860	0.858	0.863	0.859	4.9%
	Model 2	0.840	0.840	0.845	0.839	2.4%
	Model 3	0.850	0.848	0.847	0.847	3.7%
	Model 4	0.880	0.883	0.873	0.877	7.3%
	Model 5	0.890	0.889	0.887	0.888	8.5%