TABLE 5.

Comparison of our method (InceptionV3 + Fast BCNNs) and other state-of-the-art methods on BreaKHis dataset.

				Accuracy (%)
				Magnification Factor
Method	Input	Resolution	Experiment	40×	100×	200×	400×
Cascianelli et al., 2017	Image	224 × 224	MSBC	87.00	85.20	85.00	81.30
Benhammou et al., 2018	Image	Unspecified	MSBC	86.50	83.20	85.40	80.30
Song et al., 2017	Image	Unspecified	MSBC	87.70	87.60	86.50	83.90
Gupta and Bhavsar, 2018	Image	224 × 224	MSBC	94.71	95.90	96.76	89.11
Cascianelli et al., 2017	Image	256 × 256	MSBC	95.80	96.90	96.70	94.90
Wei et al., 2017	Image	224 × 224	MSBC	97.02	97.23	97.89	97.50
Gandomkar et al., 2018	Patch	224 × 224	MSBC	98.52	97.90	98.33	97.66
Our method	Image	224 × 224	MSBC	99.33	99.04	98.84	98.53
Zhang et al., 2018	Image	224 × 224	MIBC	81.20
Gupta and Bhavsar, 2017	Image	Unspecified	MIBC	88.09
Karthiga and Narasimhan, 2018	Image	Unspecified	MIBC	93.30
Our method	Image	224 × 224	MIBC	99.24
Bardou et al., 2018	Image	Unspecified	MSFC	88.23	84.64	83.31	83.98
Li et al., 2018	Image	Unspecified	MSFC	94.80	94.03	93.85	90.71
Han et al., 2017	Image	256 × 256	MSFC	92.80	93.90	93.70	92.90
Gandomkar et al., 2018	Patch	224 × 224	MSFC	95.60	94.89	95.70	94.63
Our method	Image	224 × 224	MSFC	95.99	95.84	94.70	94.51
Our method	Image	224 × 224	MIFC	95.95