Table 1 The specifications of the experimental datasets (“–” means none).

From: Deep supervision and atrous inception-based U-Net combining CRF for automatic liver segmentation from CT

Datasets

Training

Test

Size

In-plane resolution (mm)

Inter-slice resolution (mm)

Slice num

LiTS17

130

70

512 × 512

0.55–1.0

0.45–6.0

42–1026

3DIRCADb

20

512 × 512

0.56–0.81

1.25–4

74–225

SLiver07

20

10

512 × 512

0.56–0.8

1–3

64–394