An adaptive windowing method and related apparatus for annotation interface based on deep semantic segmentation and multi-region priority fusion

By employing deep semantic segmentation and multi-region priority fusion, the problems of visual differentiation and multi-view consistency of anatomical regions in medical image annotation interfaces were solved, enabling automated differentiated window adjustment and personalized annotation, thereby improving annotation efficiency and consistency.

CN122308691APending Publication Date: 2026-06-30ZHUHAI HENGQIN ALL-STAR MEDICAL TECHNOLOGY CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
ZHUHAI HENGQIN ALL-STAR MEDICAL TECHNOLOGY CO LTD
Filing Date
2026-05-29
Publication Date
2026-06-30

AI Technical Summary

Technical Problem

Existing medical image annotation interfaces cannot provide differentiated visualization for different anatomical regions. The windowing parameters are fixed, lacking the ability to adjust local windows and integrate windowing results from multiple regions. They fail to process according to clinical priorities, lack the ability to learn the preferences of annotators, and exhibit inconsistencies when annotating from multiple perspectives.

Method used

A deep semantic segmentation network is used for instance-level anatomical region segmentation. Differentiated parameters are provided based on the anatomical category-window parameter mapping matrix. Dynamic window transformation is performed through three-parameter collaborative optimization. Multi-region weighted fusion is performed based on clinical priority weights. It supports interactive fine-tuning of windowing by annotators and adaptive learning, and realizes multi-view windowing linkage.

Benefits of technology

It enables automatic identification and differentiated windowing based on anatomical regions, provides priority visualization of key anatomical structures, improves annotation efficiency and consistency, adapts to the personalized needs of different annotators, and ensures consistency of multi-view windowing.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122308691A_ABST
    Figure CN122308691A_ABST
Patent Text Reader

Abstract

This application provides an adaptive windowing method and related apparatus for annotated interfaces based on deep semantic segmentation and multi-region priority fusion. The method includes: S1. acquiring intracavitary surgical images and a set of target anatomical categories; S2. performing deep semantic segmentation on the surgical images based on a deep semantic segmentation architecture to obtain a set of anatomical regions, including vascular, neural, muscular, fat, and fascial regions; S3. querying the differential windowing parameters corresponding to each anatomical region based on an anatomical category-windowing parameter mapping matrix; S4. obtaining multiple windowed sub-images based on a dynamic windowing transformation formula using contrast, brightness, and gamma correction parameters; S5. performing weighted fusion on the multiple windowed sub-images based on multi-region priority to obtain fused pixel values; S6. receiving interactive input and performing adaptive learning updates based on historical preferences. This application also provides related apparatus.
Need to check novelty before this filing date? Find Prior Art