Table 2 Results of the validation using dice similarity coefficient and hausdorff distance for the three networks
From: Development of a deep learning-based model for guiding a dissection during robotic breast surgery
DL vs. Expert1 | DL vs. Expert2 | ||
|---|---|---|---|
EfficientDet-b4 | Hausdorff distance (mm) | 18.84\(\:\pm\:\)30.39 | 19.14\(\:\pm\:\)3.58 |
Intersection overlay union (%) | 81.50\(\:\pm\:\)7.71 | 80.10\(\:\pm\:\)8.34 | |
YOLO v5 | Hausdorff distance (mm) | 10.80\(\:\pm\:\)2.39 | 11.18\(\:\pm\:\)1.81 |
Intersection overlay union (%) | 82.71\(\:\pm\:\)6.02 | 81.08\(\:\pm\:\)8.42 | |
RetinaNet | Hausdorff distance (mm) | 9.80\(\:\pm\:\)2.57 | 10.32\(\:\pm\:\)1.09 |
Intersection overlay union (%) | 82.89\(\:\pm\:\)5.28 | 81.88\(\:\pm\:\)6.96 |