An uncertain boundary region-aware network for multi-scale liver tumor segmentation

doi:10.1016/j.eswa.2025.130732

An uncertain boundary region-aware network for multi-scale liver tumor segmentation

2025-12-09

0

PRE

AI

DOI：10.1016/j.eswa.2025.130732

Original

Original request for help

Save

Abstract

En 中文

• Multi-scale UMD-Net is proposed for accurate liver tumor segmentation. • Uncertain Boundary Reinforcement (UBR) enhances boundary region discriminability. • Key Feature Preservation (KFP) module mitigates local feature loss from the encoder. • Multi-scale Dynamic Awareness (MDA) module adaptively extracts and fuses features.

An uncertain boundary region-aware network for multi-scale liver tumor segmentation

Abstract

Journal

Organization