香港科技大学研究人员利用中国最大的mpMRI数据集开发出与放射科医生在乳腺癌诊断上相匹配的AI模型-动脉网

HKUST Researchers Develop AI Model Matching Radiologists in Breast Cancer Diagnosis Using China’s Largest mpMRI Dataset

GeneOnline 等信源发布 2025-06-05 01:10



可切换为仅中文







by Mark Chiang

马克·蒋

Share To

分享到

Researchers at the Hong Kong University of Science and Technology (HKUST) have developed an advanced artificial intelligence model, named Mixture of Modality Experts (MOME), designed to assist in non-invasive breast cancer diagnosis. The AI system, trained using China’s largest multiparametric MRI (mpMRI) breast cancer dataset, has demonstrated expert-level accuracy in classifying tumor malignancy.

香港科技大学（HKUST）的研究人员开发了一种先进的人工智能模型，名为“多模态专家混合”（MOME），旨在协助非侵入性乳腺癌诊断。该人工智能系统使用中国最大的多参数磁共振成像（mpMRI）乳腺癌数据集进行训练，在肿瘤恶性分类方面展现了专家级别的准确性。

Its performance matches that of radiologists with over five years of professional experience..

其性能与具有五年以上专业经验的放射科医生相当。

The MOME model utilizes a large-scale dataset to analyze mpMRI scans and identify whether tumors are malignant or benign without requiring invasive procedures. By leveraging multiple imaging modalities, the AI system processes complex data to deliver precise diagnostic results. Researchers report that MOME’s accuracy aligns closely with the expertise of seasoned radiologists, highlighting its potential as a reliable tool for medical professionals in diagnosing breast cancer..

MOME模型利用大规模数据集分析mpMRI扫描，判断肿瘤是恶性还是良性，且无需侵入性操作。通过整合多种成像模式，该人工智能系统处理复杂数据并提供精确的诊断结果。研究人员报告称，MOME的准确性与经验丰富的放射科医生的专业水平高度一致，突显了其作为乳腺癌诊断可靠工具的潜力。

Newsflash | Powered by GeneOnline AI

新闻快讯 | 由GeneOnline AI提供支持

Source: GO-AI-ne1

源：GO-AI-ne1

Date: June 5, 2025

日期：2025年6月5日

相关内容：

DrugPatentWatch: Biosimilar Success Hinges on FDA Engagement, BPCIA Understanding

DrugPatentWatch：生物仿制药的成功取决于FDA的参与和对BPCIA的理解

DrugPatentWatch Research: Neglecting APIs Leads to Wasted Resources in Key Starting Material Strategies

DrugPatentWatch研究：忽视API会导致关键起始物料策略中的资源浪费