P2Echo: Unified Text-Prompted Multi-View Echocardiography Segmentation

·2026.11.12 19:16

Key point

The framework uses a Dynamic Image-Text Alignment module to adaptively condition decoder features on requested views and anatomical structures.

P2Echo is a unified framework for text-prompted segmentation of multiple cardiac structures across standard echocardiographic views using a single model.
The system employs structured, view-aware text prompts to control segmentation, with a Dynamic Image-Text Alignment (DyITA) module that adaptively conditions decoder features on the requested view and anatomical structure.
Official pretrained weights are available for research in medical image analysis and echocardiographic segmentation.
The model has not been validated for clinical decision-making.

This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.

Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.