Evaluating the performance of convolutional neural networks with direct acyclic graph architectures in automatic segmentation of breast lesion in US images

Table 5 Metrics values for cross-validation with 5 folders, for dataset B, using original images resized for 160 × 160 pixels

Folder	Global Accuracy	Accuracy	IoU	Weighted IoU	BF Score	Dice Coefficient
1	0.982	0.915	0.806	0.969	0.669	0.692
2	0.987	0.900	0.844	0.976	0.730	0.758
3	0.977	0.833	0.744	0.960	0.597	0.574
4	0.970	0.820	0.738	0.947	0.630	0.608
5	0.983	0.933	0.837	0.969	0.693	0.714
Mean	0.979	0.880	0.794	0.964	0.664	0.669
Standard Deviation	0.007	0.051	0.050	0.011	0.052	0.076

ISSN: 1471-2342