表 2.

不同模态组合间的比较

Comparison between different modal combinations

Models	Accuracy (%)	Precision (%)	Recall (%)	F1-score (%)
OM(No Seg) model: The model trained on optical microscope data without segmenting glomerular regions. OM model: The model trained on optical microscope datawith segmenting glomerular regions. IM model: The model trained on immunofluorescence microscope data. TEM model: The model trained on transmission electron microscope data. OM&IM model: The fusion model trained on optical microscope and immunofluorescence microscope data. The naming convention for the other models was the same.
OM(No Seg) model	72.90±3.67	72.80±3.62	72.87±3.49	72.60±3.66
OM model	77.50±3.06	78.13±3.91	76.93±3.17	76.93±2.98
IM model	87.08±4.25	87.53±3.62	87.27±4.12	86.87±4.04
TEM model	38.08±3.78	36.73±3.61	38.20±3.44	35.60±3.71
OM&IM model	87.92±3.06	88.60±3.81	88.00±3.00	87.87±3.44
IM&TEM model	87.09±2.43	87.00±2.20	86.93±2.72	86.67±2.39
OM&TEM model	75.83±2.12	76.73±3.19	75.87±2.07	75.60±2.12
OM&IM&TEM model	88.34±2.12	88.73±2.40	88.27±2.23	88.13±2.08