A High-Precision Classification Method of Mammary Cancer Based on Improved DenseNet Driven by an Attention Mechanism

Table 4

Classification accuracy of different models for images with different magnifications.

Network model	The experimental type	Accuracy of different magnification (%)
		40x	100x	200x	400x
DenseNet121	Binary classification	90.90	90.01	91.91	91.66
	Multiclass classification	82.24	79.19	81.23	83.30
MFSCNet A	Binary classification	99.51	99.46	99.89	99.05
	Multiclass classification	97.13	94.36	98.41	95.96
MFSCNet B	Binary classification	98.97	98.55	98.06	99.32
	Multiclass classification	93.01	93.73	94.05	91.76
MFSCNet C	Binary classification	99.36	99.05	99.00	98.81
	Multiclass classification	92.91	92.03	91.47	92.51