数据清洗技术在DICOM格式医学图像质控中的应用  

Study on Data Cleaning Technology in DICOM format Medical Image Quality Control

作　　者：郝烨[1] 唐桥红 李佳戈[1] 王浩[1] 孟祥峰[1] 任海萍[1] HAO Ye;TANG Qiaohong;LI Jiage;WANG Hao;MENG Xiangfeng;REN Haiping(Division of Active Medical Device and Medical Optics, National Institutes for Food and Drug Control, Beijing 100050, China)

机构地区：[1]中国食品药品检定研究院光机电室,北京100050

出　　处：《中国医疗设备》2018年第12期10-13,共4页China Medical Devices

基　　金：国家重点研发计划项目(2016YFC0107100);体育总局重点课题联合中国红十字基金会燎原基金项目(2015B101)

摘　　要：随着信息技术和互联网行业的发展,全球进入大数据时代,数据的开发、挖掘和分析应用越来越广泛,对数据的质量要求也越来越高。目前,国内外的专家学者对医疗领域人工智能产品都进行了很多研发,人工智能产品的研发需要依托海量的医学临床数据。为了保证这类产品的质量,必须从源头进行必要的筛选和清洗,以保障数据质量,支持后续的产品研发与验证过程。本文对DICOM格式的数据清洗问题进行分析,开发了对原始数据进行清洗和审核的流程,在实践中进行了测试,证明能够有效地发现数据缺陷,为今后开展医学人工智能专用数据集的质控工作起到借鉴作用。With the development of information technology and Internet industry,the world has entered the era of big data.The development,mining and analysis of data are more and more widely applied,and high-quality data is increasingly demanded.At present,experts and scholars have done a lot of research and development on artificial intelligence products in the medical field.The research and development of AI products rely on massive medical clinical data.In order to ensure the quality of such products,the necessary screening and cleaning from the source must be carried out to control data quality and support the follow-up product development and verification process.We analyzed the data cleaning problem in DICOM format,developed the process of original data cleaning and auditing,and tested it in practice.It proves that the data defects can be found effectively,which can be used for reference in the future quality control of medical AI data sets.

关 键 词：数据清洗 数字图像通讯协议 医学图像 质量控制 

分 类 号：TN915.12[电子电信—通信与信息系统]