基于人工智能的胸部CT影像智能化质控系统研究  

Research on AI-based intelligent quality control system for chest CT

作　　者：胡佳迎 黄宗浩[2] 王奕[1,3] Hu Jiaying;Huang Zonghao;Wang Yi(Department of Big Data Center,Fudan University Shanghai Cancer Center,Shanghai 200032,China;IT Center,Fudan University Shanghai Cancer Center,Shanghai 200032,China;Department of Oncology,Shanghai Medical College,Fudan University)

机构地区：[1]复旦大学附属肿瘤医院大数据中心,上海200032 [2]复旦大学附属肿瘤医院信息中心,上海200032 [3]复旦大学上海医学院肿瘤学系

出　　处：《中国医疗管理科学》2025年第2期82-87,共6页Chinese Journal Of Medical Management Sciences

摘　　要：目的解决胸部CT影像质控中存在的问题,如人工误差大、效率低下、质控标准不一致和技术限制,通过开发一套基于人工智能的胸部CT影像智能化质控系统,提高影像质控的效率和一致性,以确保医疗影像质量。方法系统设计包括模块化、定制化质控和基于因果的人工智能质控,以增强鲁棒性和可解释性。系统功能集成了基于U-Net和ResNet的神经网络模块,用于器官分割、骨骼分割分类、躯干提取、噪声分级、影像质量分析和伪影识别。研究构建了一个包含多种质量问题影像的数据集,并在多家医院进行了系统部署和测试。结果系统在识别胸部CT影像中的质控问题上表现出较高准确率,综合识别准确率达到80%以上。系统处理每张CT影像的时间平均在1~2 min,显著低于人工质控所需时间。实际应用中,系统提高了胸部CT影像的质控效率和一致性,减少了因影像质量问题导致的重复检查,提高了患者的就医满意度。结论基于人工智能的胸部CT影像智能化质控系统显著提升了影像质控的效率和一致性,减轻了医生的工作负担,提高了阅片效率和患者就诊体验。Objective To design an artificial intelligence(AI)-based intelligent quality control system for chest computed tomography(CT)images,thus solving existing issues in the quality control of chest CT(e.g.,large manual errors,low efficiency,inconsistent quality control standards,and technical limitations),improving the efficacy of imaging quality control and consistency,and ensuring the quality of medical imaging.Methods The system design included modular,customized quality control,and causal AI quality control to enhance robustness and interpretability.The system functions integrated neural network modules based on the U-Net and ResNet for organ segmentation,skeletal segmentation classification,trunk extraction,noise classification,image quality analysis,and artifacts identification.A dataset containing various quality issues was constructed,and the system was deployed and tested in multiple hospitals.Results The system demonstrated high accuracy in identifying quality issues in chest CT images,with an overall identification accuracy rate of over 80%.The system processed each CT image in an average of 1-2 minutes,which was significantly faster than the time required for manual quality control.In practical application,the system improved the efficiency and consistency of chest CT image quality control,reduced repeated examinations due to image quality issues,and enhanced patient satisfaction with medical services.Conclusion The AI-based intelligent quality control system for chest CT images developed in this study significantly improves the efficiency and consistency of image quality control,reduces the workload of medical staff,and improves the efficiency of film reading and patient experience.

关 键 词：人工智能 胸部CT影像 智能化质控 

分 类 号：R197.324[医药卫生—卫生事业管理]