Multiparametric Magnetic Resonance Imaging Information Fusion Using Graph Convolutional Network for Glioma Grading

Table 2

Demonstration of the context information fusion method based on graph convolution and 3D convolution.

Dataset	Method	T1CE		T1		T2		FLAIR
		AUC	ACC	AUC	ACC	AUC	ACC	AUC	ACC
BraTS2020	2D-ResNet	0.932	0.907	0.896	0.840	0.849	0.840	0.894	0.853
	3D-ResNet	0.939	0.920	0.905	0.853	0.860	0.880	0.853	0.867
	G-2D-ResNet	0.946	0.920	0.906	0.853	0.873	0.880	0.908	0.867
GliomaHPPH2018	2D-ResNet	0.842	0.787	0.921	0.894	0.729	0.745	0.931	0.872
	3D-ResNet	0.844	0.809	0.960	0.894	0.750	0.766	0.923	0.872
	G-2D-ResNet	0.900	0.830	0.962	0.915	0.813	0.787	0.942	0.894