Endometrial cancer infiltration measurement method based on semantic segmentation

By automatically determining the measurement plane and resampling to obtain the boundary line through a deep learning model, the problem of measurement inconsistency and error accumulation caused by manual selection in the existing technology is solved, and the measurement of endometrial cancer invasion is automated and consistent.

CN122199418APending Publication Date: 2026-06-12CHANGZHOU TUMOR HOSPITAL (CHANGZHOU FOURTH PEOPLES HOSPITAL)
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
CHANGZHOU TUMOR HOSPITAL (CHANGZHOU FOURTH PEOPLES HOSPITAL)
Filing Date
2026-03-02
Publication Date
2026-06-12

AI Technical Summary

⚠Technical Problem

In existing technologies, methods for measuring endometrial cancer invasion rely on manual selection of the deepest invading section, resulting in inconsistent measurement planes and problems such as strong subjectivity, insufficient repeatability, and error accumulation.

⚗Method used

A semantic segmentation method based on deep learning is adopted. After preprocessing the three-dimensional magnetic resonance image, it is input into the deep learning model and outputs the semantic segmentation mask of the tumor, endometrium and myometrium and the heat map of the deepest invasion key point. The measurement plane is automatically determined and the boundary line is obtained by resampling. The myometrial invasion depth and proportion are calculated.

🎯Benefits of technology

The measurement of endometrial cancer invasion has been automated and standardized, reducing measurement bias and improving measurement consistency and efficiency.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122199418A_ABST
    Figure CN122199418A_ABST
Patent Text Reader

Abstract

The application discloses a kind of endometrial carcinoma infiltration measurement methods based on semantic segmentation, to solve the problems that the deepest infiltration section and the optimal measurement plane are difficult to be automatically determined in the prior art, and the myometrial infiltration depth and myometrial infiltration ratio measurement are not standardized, the application inputs depth learning model after pre-processing three-dimensional magnetic resonance image, outputs semantic segmentation result of tumor, endometrium and myometrium, the deepest infiltration key point heat map and measurement plane parameter, determines the deepest infiltration point and constructs the measurement plane through the point, resamples along the measurement plane to image and mask, extracts endometrial and myometrial junction line and uterine serosa boundary line and carries out geometric measurement, realizes the technical effect of automatically outputting myometrial infiltration depth and myometrial infiltration ratio.
Need to check novelty before this filing date? Find Prior Art