Table 3 Final C-MAC mesh metrics evaluated against the ground-truth calcification segmentation

From: Robust automated calcification meshing for personalized cardiovascular biomechanics

Experiment condition

Non-empty target segmentation

Empty target

Initial segment

Post-process

Dice ↑

HD 95% (mm) ↓

CD (mm) ↓

Merged nodes ↑

False pos (vx) ↓

Ground truth

✓

0.683 ± 0.036

1.703 ± 1.023

0.525 ± 0.110

2899 ± 1774

N/A

GDL (ours)

✓

0.630 ± 0.074

4.560 ± 4.519

0.852 ± 0.423

2295 ± 998

7.5 ± 10.8

  1. These metrics are measuring the error induced by the entire C-MAC pipeline including the initial segmentation, segmentation post-processing, and meshing steps. Similar to Table 2, we used 10 repeated trials to obtain these metrics. All metrics are mean ± stdev across patients.
Back to article page