No abstract

No abstract

No abstract

Hepatic veins segmentation is a fundamental task for liver preoperative planning. The segmentation target distribution in the hepatic vein scene is extremely unbalanced. To solve this problem, we consider the 3D spatial distribution and density awareness (SDA) of hepatic veins and propose an automatic segmentation network based on 3D U-Net which includes a multi-axial squeeze and excitation module and a distribution correction module. The multi-axial squeeze and excitation module restrict the activation area to the area with hepatic veins. The distribution correction module improves the awareness of the sparse spatial distribution of the hepatic veins. To obtain global axial information and spatial information at the same time, we study the effect of different training strategies on hepatic vein segmentation. Experimental results on the medical segmentation decathlon dataset and our private dataset show that our method has advantages over the state-of-the-art methods in terms of various performance indicators.

No abstract