DBT Masses Automatic Segmentation Using U-Net Neural Networks

Table 5

Comparisons of selected studies in the detection of masses in the DBT images.

Method	Classifier	DBT dataset size	Sen	Acc	AUC
Kim et al. [31]	LDA	36	0.90	—	—
Shamsolmoali et al. [28]	SVM	160	—	—	0.847
Sajjad et al. [29]	DCNN	344	0.89	0.864	—
Glorot and Bengio et al. [30]	DCNN	324	0.80	—	0.80
Palma et al. [17]	SVM	101	0.90	—	—
Chan et al. [16]	Neural network	752	0.80	—	—
Reiser et al. [14]	LDA	100	0.80	—	—
Proposed	U-net	87	0.869	0.871	0.859