Table 5.

Comparative status of the proposed method with current state-of-the-art methods.

References	Dataset	Proposed method	Classification accuracy
Wu et al. (20)	2,00,000	A deep convolutional network with 100 layers.	0.825 on Four views
Ciritsis et al. (19)	20,578	A deep convolutional network with 11 layers and performed analysis separately on CC and MLO views.	0.897 On CC views and 0.866 on MLO views.
Kaiser et al. (24)	8,150	A multichannel architecture with transfer learning by VGG-Net.	0.88 on all four views
Shi et al. (22)	322	A light-weight deep learning architecture with 3 convolutional layers.	0.836 On MLO views.
Deng et al. (36)	18,157	A single channel architecture with transfer learning by Dense Net 121 combined with SE-Attention network.	0.9179 on all Four views
Proposed method	800	A multichannel architecture with transfer learning with Dense Net 121	0.90 on Four views