A breakthrough approach that enables natural language prompts to guide segmentation of anatomical structures across medical image sequences (endoscopy, ultrasound, CT, MRI).

• Introduces a novel architecture that tracks objects of interest throughout image sequences using text descriptions
• Delivers more accurate segmentation compared to traditional 2D and 3D segmentation models
• Provides a user-friendly interface for clinicians through natural language interaction
• Demonstrates clinical potential across multiple medical imaging modalities

This innovation represents a significant advancement in making medical imaging interpretation more accessible and efficient for healthcare professionals, potentially improving diagnostic accuracy and workflow.

Text-Promptable Propagation for Referring Medical Image Sequence Segmentation