Fine-Grained Lung Cancer Classification from PET and CT Images Based on Multidimensional Attention Mechanism

Table 1

Evaluation of different modality datasets.

Network architecture	Data	AUC				Balanced accuracy
		Score of each category		Average score	Variance
3D DenseNet	CT	LUSC	0.667	0.621	0.08	0.52
		LUAD	0.711
		SCLC	0.485
3D DenseNet	PET	LUSC	0.722	0.678	0.12	0.53
		LUAD	0.768
		SCLC	0.544
MF-DenseNet	CT + PET	LUSC	0.863	0.810	0.05	0.57
		LUAD	0.877
		SCLC	0.690