Tumor image segmentation methods, devices, and electronic equipment

CN118212407BActive Publication Date: 2026-05-26PEKING UNIV +2
View PDF 2 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
PEKING UNIV
Filing Date
2024-03-11
Publication Date
2026-05-26

AI Technical Summary

Technical Problem

Existing tumor segmentation models are unable to perform accurate segmentation in complex medical tumor image segmentation scenarios and are difficult to incorporate human expertise.

Method used

A tumor segmentation model, including an encoder, decoder, and multi-scale attention module, is adopted. Feature fusion processing is performed by combining interaction points sampled from the original tumor image of the target object. The segmentation model includes a multi-scale attention module for feature fusion and upsampling to generate the target tumor image.

Benefits of technology

It achieves accurate segmentation of tumor regions in complex medical image segmentation scenarios, improving the automation and accuracy of tumor segmentation.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118212407B_ABST
    Figure CN118212407B_ABST
Patent Text Reader

Abstract

This application discloses a method, apparatus, and electronic device for tumor image segmentation. The method includes: acquiring an original tumor image to be segmented, wherein the original tumor image is an image obtained after a CT scan of the target organ of the target object; segmenting the original tumor image using a tumor segmentation model to obtain a target tumor image corresponding to the tumor region; wherein the tumor segmentation model includes an encoder, a decoder, and a multi-scale attention module, the multi-scale attention module performing feature fusion processing on the original tumor image by combining interaction points sampled from the target object in the original tumor image; and displaying the target tumor image. This application solves the technical problem of poor accuracy in tumor segmentation in related technologies.
Need to check novelty before this filing date? Find Prior Art

Citation Information

Patent Citations

  • Image multi-target extraction method and system based on deep circulation attention

    CN114092422A

  • Multi-modal interactive medical image segmentation method and device

    CN116862822A