Table 4 Quantitative evaluation of DBAHNet segmentation performance with and without post-processing across the high-dose groups in Dataset 114, Dataset 315 and Dataset 216. The metrics include the average Dice Similarity Coefficient (DSC) and the average 95th percentile of Hausdorff Distance (HD95).

From: A robust deep learning approach for segmenting cortical and trabecular bone from 3D high resolution µCT scans of mouse bone

 

Dataset 114

Dataset 215

Dataset 316

PTH40

PTH80

Ris150

Ris150 + ML

4N

8N

DSC

HD95

DSC

HD95

DSC

HD95

DSC

HD95

DSC

HD95

DSC

HD95

No post-proc.

0.9440

0.0531

0.9498

0.0424

0.9797

0.0153

0.9682

0.0735

0.9733

0.0408

0.9723

0.0501

Post-proc.

0.9489

0.0430

0.9671

0.0250

0.9821

0.0080

0.9838

0.0093

0.9791

0.0099

0.9794

0.0473

  1. Best performances are in bold.