Table 2 Multi-modal liver segmentation accuracy using pre-trained CT model for Anatomy3 CECT, 3DIrcad and DECT datasets acquired with different scan protocols, and cross-validation on CHAOS CT-MR dataset.
From: Deep action learning enables robust 3D segmentation of body organs in various CT and MRI images
Dataset | No. | Our Method | 3D U-Net |
|---|---|---|---|
DICE [%] | DICE [%] | ||
AVG ± STD | AVG ± STD | ||
Anatomy3 CECT | 20 | 82.6 ± 7.6 | 72.4 ± 9.9 |
3DIrcad | 20 | 82.3 ± 9.0 | 59.4 ± 11.1 |
Dual Energy CT | 45 | 86.5 ± 3.5 | 60.5 ± 10.5 |
