Researchers have developed RF-HiT, a Rectified Flow Hierarchical Transformer designed for general medical image segmentation. This model addresses limitations in existing transformer and diffusion-based methods by integrating an hourglass transformer backbone with a multi-scale hierarchical encoder. RF-HiT achieves high efficiency with linear complexity and fast inference, requiring only a few discretization steps and minimal computational resources. AI
Summary written by gemini-2.5-flash-lite from 1 source. How we write summaries →
RANK_REASON This is a research paper detailing a new model for medical image segmentation.