Table 4.

Summary of reviewed works on deep learning models.

Subcategory	Related works	Year	Technique	Filter	Database	Evaluation metric
DL	[160]	2015	CRF		INbreast and DDSM-BCRP	89.0% of Dice
DL	[157]	2018	Adversarial FCN-CRF		INbreast and DDSM-BCRP	97.0% accuracy
DL	[158]	2018	FrCN		INbreast	92.97 segmentation accuracy, 92.69% Dice and MCC of 85.93%
DL	[174]	2018	CRU-Net		INbreast and DDSM	93.66% of Dice for INbreast and 93.32% for DDSM
DL	[161]	2019	ResCU-Net and MS-ResCU-Net		INbreast	91.78% of Dice, 94.16% accuracy, and Jac of 85.12% based on MS-ResCU-Net
DL	[162]	2019	U-Net and AGS		DDSM	82.24% F-score, 77.89% sensitivity, and 78.38% accuracy
DL	[169]	2019	RU-Net	cLare filter	INbreast and DDSM-BCRP	98.0% of Dice, 94.0% of IOU, and 98.0% accuracy
DL	[170]	2019	U-Net	Laplace filter	DDSM	97.80% of Dice and 98.50% of Fi-score
DL	[171]	2019	AUNet		INbreast and DDSM	81.80% of Dice for DDSM and DI of 79.10% for INbreast
DL	[163]	2020	Mask RCNN		INbreast	88.0% of Dice
DL	[164]	2020	FrCN		INbreast	92.69% of Dice, 92.97% accuracy, and Jac of 86.37%
DL	[165]	2020	U-Net	Adaptive median	INbreast and DDSM	89.0% of Dice and mean IOU of 90.90%
DL	[159]	2020	DS-U-Net	cLare filter	INbreast and DDSM	82.7% of Dice, Jac of 99.7%, and accuracy of 83.0%
DL	[166]	2020	cGAN	Median filter	INbreast	88.0% of Dice, Jac of 78.0%, and 98.0% accuracy
DL	[167]	2020	cGAN	Morphological filter	DDSM	94.0% of Dice and IOU of 87.0%
DL	[168]	2020	Mask RCNN and DeepLab	Savitzky Golay filter	MIAS and DDSM	80.0% accuracy
DL	[175]	2020	Mask RCCN-FPN		DDSM	91.0% accuracy and 84.0% precision
DL	[176]	2020	U-Net		DDSM	79.39% of Dice, AUC of 86.40%, and 85.95% of accuracy
DL	[172]	2021	U-Net		DDSM	88.0% accuracy
DL	[173]	2021	U-Net		MIAS and DDSM	98.87% of Dice, AUC of 98.88%, and Fi-score of 97.99%