Systems and methods for quantifying uncertainty of segmentation masks produced by machine learning models

By generating multiple enhanced images and quantifying the uncertainty of segmentation masks, the problem of confidence differentiation in medical image segmentation by machine learning models is solved, thereby improving the efficiency and accuracy of the medical imaging workflow.

CN116071378BActive Publication Date: 2026-06-09GE PRECISION HEALTHCARE LLC
View PDF 1 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
GE PRECISION HEALTHCARE LLC
Filing Date
2022-10-25
Publication Date
2026-06-09

AI Technical Summary

Technical Problem

Existing machine learning models cannot effectively distinguish between segmentation masks with high and low confidence in medical image segmentation, making it difficult for users to determine the certainty of automatically determined anatomical landmark locations and reducing the efficiency of medical imaging workflows.

Method used

By generating multiple enhanced images, using a trained machine learning model to produce multiple segmentation masks, quantifying the uncertainty of the segmentation masks, and prompting users to confirm or edit the position of the diameter gauge through uncertainty mapping and visual indicators, the uncertainty of the diameter gauge placement is automatically estimated.

Benefits of technology

It improves the efficiency of medical imaging workflows, enabling users to more accurately assess the confidence level of automatically determined landmark locations and simplifies anatomical measurement workflows.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN116071378B_ABST
    Figure CN116071378B_ABST
Patent Text Reader

Abstract

Systems and methods are provided for quantifying uncertainty of segmentation mask predictions made by machine learning models, where the uncertainty can be used to simplify anatomic measurement workflows by automatically identifying less certain caliper placements. In one example, the present disclosure teaches receiving an image including a region of interest, determining a segmentation mask for the region of interest using a trained machine learning model, placing a caliper at a location within the image based on the segmentation mask, determining an uncertainty for the location of the caliper, and responding to the uncertainty for the location of the caliper being greater than a predetermined threshold by displaying a visual indication of the location of the caliper via a display device and prompting a user to confirm or edit the location of the caliper.
Need to check novelty before this filing date? Find Prior Art