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

Table 3 Metrics values for cross-validation with 5 folders for dataset A, using cropped images resized for 160 × 160 pixels

Folder	Global Accuracy	Accuracy	IoU	Weighted IoU	BF Score	Dice Coefficient
1	0.952	0.944	0.864	0.913	0.679	0.877
2	0.969	0.961	0.904	0.942	0.754	0.916
3	0.936	0.944	0.834	0.887	0.603	0.850
4	0.961	0.947	0.884	0.923	0.704	0.904
5	0.964	0.954	0.894	0.932	0.724	0.915
Mean	0.956	0.950	0.876	0.920	0.693	0.918
Standard Deviation	0.011	0.006	0.025	0.019	0.051	0.025

ISSN: 1471-2342