A lung nodule early warning method and system based on multi-model fusion

By integrating time-series information and lung nodule features, and dynamically correcting the prior knowledge matrix, the lung nodule identification method of the multi-model fusion approach solves the problem of lack of intermediate annotation in the existing lung nodule identification methods, and achieves highly accurate lung nodule early warning and dynamic control.

CN120431044BActive Publication Date: 2026-06-09上海中域工业互联网研究院 +2
View PDF 2 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
上海中域工业互联网研究院
Filing Date
2025-04-23
Publication Date
2026-06-09

AI Technical Summary

Technical Problem

Existing methods for identifying lung nodules lack intermediate annotations, leading to decreased prediction accuracy. They also lack effective dynamic control mechanisms, making it impossible to effectively intervene in the intermediate stages of disease progression.

Method used

A multi-model fusion approach is adopted, including standardization, feature recognition, temporal probability prediction, learning supervision, and relabeling models, combined with a three-dimensional convolutional neural network, to quantify the risk of deterioration and trigger clinical early warning.

Benefits of technology

By integrating time-series information and lung nodule features through multi-model fusion, and dynamically correcting the prior knowledge matrix, the potential state of lung nodules can be predicted, supporting medical intervention and improving prediction accuracy and dynamic control capabilities.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN120431044B_ABST
    Figure CN120431044B_ABST
Patent Text Reader

Abstract

The present application relates to a kind of lung nodule early warning method and system based on multi-model fusion, belong to health management field.Therein, the method includes collecting lung nodule CT image and implementing standardization processing, generates lung nodule analysis image;The feature recognition of lung nodule analysis image is executed to generate the lung nodule label image with multi-dimensional pathological feature label, and the lung nodule time series data set is structured;Lung nodule label image and lung nodule time series data set are used to construct lung nodule fusion prediction model and output lung nodule relabeling pathological image, lung nodule fusion prediction model integrates time series probability prediction model, learning supervision model and relabeling model;Based on lung nodule relabeling pathological image, construct three-dimensional convolutional neural network early warning model, quantify deterioration risk coefficient, when deterioration risk coefficient breaks through preset alert threshold, trigger clinical early warning mechanism, the present application realizes the relabeling of patient lung nodule potential state by fusion model, completes lung nodule early warning.
Need to check novelty before this filing date? Find Prior Art