A method for training an EVT prognostic prediction model, an EVT prognostic prediction method, equipment, media, and products.

By extracting radiomics features of the region of interest of the infarct lesion in the posterior circulation partition-weighted imaging of patients with acute basilar artery occlusion and training a classification network, the problems of strong subjectivity, insufficient information and timeliness in the existing technology are solved, and an objective and accurate assessment of the prognosis of EVT is achieved.

CN122091231APending Publication Date: 2026-05-26CAPITAL UNIVERSITY OF MEDICAL SCIENCES
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
CAPITAL UNIVERSITY OF MEDICAL SCIENCES
Filing Date
2026-04-02
Publication Date
2026-05-26

Smart Images

  • Figure CN122091231A_ABST
    Figure CN122091231A_ABST
Patent Text Reader

Abstract

This application discloses a method, device, medium, and product for training an EVT prognostic prediction model, relating to the field of prognostic prediction technology. The method for training an EVT prognostic prediction model includes: acquiring a patient's preoperative diffusion-weighted imaging image and functional outcome labels; identifying the region of interest (ROI) of the infarct on the preoperative diffusion-weighted imaging image; assigning weights to each brain region in the preoperative diffusion-weighted imaging image according to posterior circulation partitioning rules, and generating a posterior circulation partitioned weighted ROI based on the distribution of the ROI in each brain region; extracting radiomics features from the posterior circulation partitioned weighted ROI; and training a classification network using the radiomics features and corresponding functional outcome labels to obtain the EVT prognostic prediction model. This application achieves objective and accurate prognostic prediction through the fusion of key region weighting and radiomics.
Need to check novelty before this filing date? Find Prior Art