A deep learning-based multi-modal medical image artifact intelligent detection method and system

CN121767355BActive Publication Date: 2026-06-19FANTASTIC BIOIMAGING CO LTD
View PDF 2 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
FANTASTIC BIOIMAGING CO LTD
Filing Date
2026-03-02
Publication Date
2026-06-19

AI Technical Summary

Technical Problem

Existing technologies rely on manual assessment of artifacts, which is inefficient and highly subjective. They cannot adapt to the differentiated characteristics of multimodal medical images and cannot provide a unified quantitative assessment, leading to increased diagnostic difficulty and the risk of misdiagnosis.

Method used

A deep learning-based multimodal medical image artifact detection method is adopted. Standardized images are generated through DICOM parsing and preprocessing, and artifact detection is performed using convolutional neural networks and multi-classification networks. Structured reports are generated by combining U-Net pixel-level segmentation and interpretable heatmaps.

Benefits of technology

It enables accurate detection and quantitative evaluation of multimodal image artifacts, provides standardized detection reports, improves detection efficiency and accuracy, and reduces the risk of misdiagnosis.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN121767355B_ABST
    Figure CN121767355B_ABST
Patent Text Reader

Abstract

This application relates to a deep learning-based intelligent detection method and system for multimodal medical image artifacts, comprising: receiving multimodal raw medical images and preprocessing the raw medical images to generate standardized target medical images; performing global detection based on the target medical images to obtain artifact confidence; when the artifact confidence exceeds a set threshold, identifying the corresponding artifact type according to a preset multi-classification network, and performing pixel-level segmentation and localization of the target medical images based on the artifact type to extract the main artifact regions and locations, and calculating a severity score based on the main artifact regions; generating an interpretable heatmap based on the target medical images, and extracting key features from the interpretable heatmap to generate a structured detection report, thereby solving the problems of reliance on manual work and strong modality specificity in the prior art.
Need to check novelty before this filing date? Find Prior Art