Table 7 Results of all four experiments done using nnU-Net.

From: Deep learning for 3D vascular segmentation in hierarchical phase contrast tomography: a case study on kidney

Expt.

Train data

Test data

DSC

CLD

NSD (t = 1)

NSD (t = 0)

ASSD

1

Kidney 1,2

Kidney 3

0.9410

0.8886

0.9651

0.7631

4.300

2

Kidney 1,3

Kidney 2

0.9523

0.8533

0.9518

0.7120

0.8639

3

Kidney 2,3

Kidney 1

0.8585

0.8228

0.8968

0.7132

2.9270

4

Half of kidney 1

Other half of kidney 1

0.9513

0.8631

0.9404

0.8549

2.1561

  1. SD surface distance, NSD normalised surface DSC, CLD centerline DSC, t tolerance (in voxels), ASSD average symmetric surface distance
Back to article page