Table 1 Performance comparison of the proposed method with state-of-the-art models on the control dataset14,15,16. The results include the average Dice Similarity Coefficient (DSC) and the average 95th percentile of Hausdorff Distance (HD95) for both cortical (C) and trabecular (T) compartments.

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

Models

Avg DSC

DSC C

DSC T

Avg HD95

HD95 C

HD95 T

UNet

0.9007

0.9049

0.8965

0.4129

0.4440

0.3818

Attention UNet

0.9507

0.9629

0.9384

0.1931

0.2013

0.1850

UNETR

0.9662

0.9832

0.9492

0.1131

0.1269

0.0992

SwinUNETR

0.9735

0.9902

0.9569

0.0498

0.0153

0.0843

DBAHNet

0.9841

0.9913

0.9769

0.0095

0.0080

0.0110

  1. Best performances are in bold.