A Technical Review of Convolutional Neural Network-Based Mammographic Breast Cancer Diagnosis

Table 1

A summary of CNN based MBCD methods.

	Year	Database	No. of images	AUC	ACC	SPE	SEN
[51]	2016	DDSM	600		0.967
[52]	2016	BCDR-F03	736	0.82 ± 0.03
[53]	2016	In-house	607	0.86
[54]	2017	In-house	3158	0.88	0.82	0.72	0.81
[55]	2017	In-house	2454	0.82 ± 0.02
[56]	2017	In-house	245	0.86 ± 0.01
[57]	2017	INbreast	115	0.91 ± 0.12	0.95 ± 0.05
[58]	2017	In-house	560	0.79 ± 0.02
[59]	2017	IRMA	2796	0.839	0.837	0.854	0.797
[60]	2018	In-house	78	0.81 ± 0.05
[61]	2018	In-house	3290	0.7274
[62]	2018	DDSM	600		0.974
		MIAS	120		0.967
[63]	2018	BCDR-F03	736	0.813
[64]	2018	DDSM	5316	0.98	0.9735
		BCDR-F03	600	0.96	0.9667
		INbreast	200	0.97	0.9550
[65]	2018	DDSM	600		0.97
[66]	2018	DREAM	82,000	0.85
		INbreast	115	0.95
[67]	2018	BCDR-F03	736	0.891	0.852
[68]	2018	BCDR-F03	736	0.88	0.81