PA-Net: A phase attention network fusing venous and arterial phase features of CT images for liver tumor segmentation
Abstract: Highlights•A voxel-level attention module was designed to fuse arterial and venous phase CT image features for liver tumor segmentation.•Inter-phase spatial misalignment due to respiratory motion and registration errors is alleviated.•An effective learning strategy is designed for multi-phase liver tumor segmentation of thick layer CT images.•Multi-phase CT images from 362 patients were collected and labeled to validate our method.
Loading