Table 3.

Comparison of classification results of glioma molecular markers with multiple labeling status between between different attention mechanisms (unit: %)

Model		DADC	CBAM [26]	CA [27]	SE [28]	SCSE [29]	GAM [30]
ACC (%)	IDH1	97.32	95.63	95.98	96.10	96.51	96.10
	Ki67	98.31	98.14	97.79	97.15	97.15	97.32
	MGMT	96.62	96.51	95.75	95.69	95.63	96.04
	P53	98.20	97.15	96.56	96.45	96.68	97.44
Pre (%)	IDH1	97.44	95.55	95.93	95.81	96.29	95.99
	Ki67	97.89	97.77	97.31	96.09	95.88	96.64
	MGMT	96.80	96.44	95.59	95.66	95.38	95.86
	P53	98.24	97.07	96.52	96.69	96.85	97.34
Recall (%)	IDH1	96.77	95.00	95.39	95.78	96.16	95.57
	Ki67	96.68	96.23	95.56	94.71	94.94	94.70
	MGMT	96.00	96.09	95.33	95.13	95.30	95.69
	P53	97.37	95.98	95.10	94.67	95.07	96.43
F1-score (%)	IDH1	97.10	95.26	95.64	95.79	96.23	95.77
	Ki67	97.27	96.98	96.40	95.39	95.40	95.64
	MGMT	96.40	96.26	95.46	95.38	95.34	95.77
	P53	97.79	96.50	95.78	95.60	95.90	96.87
AUROC (%)	IDH1	97.09	95.26	95.64	95.79	96.23	95.77
	Ki67	97.27	96.98	96.40	95.38	95.40	95.64
	MGMT	96.37	96.26	95.46	95.38	95.34	95.77
	P53	97.79	96.50	95.78	95.60	95.91	96.87

Comparison of the joint prediction results of multiple models for multi-label status of glioma molecular markers

Bold values indicate the optimal performance