中国人逐拍标注心电图数据库的构建

发布时间：2018-03-01 20:36

本文关键词： 心电图数据库标注　出处：《北京协和医学院》2012年硕士论文　论文类型：学位论文

【摘要】：越来越多的心电图机配备了用于识别心拍类型的心电自动诊断系统,在其上市前,相关质检部门必须利用由医学专家人工标注的心电图数据库对其进行性能检测。国际电工委员会在其针对心电设备检测的标准中明确规定,用于验证心电图自动诊断算法的心电图数据库,必须从该心电图机预期使用的特定人群中采集。本工作正是要构建这样一个用于检验自动诊断算法的中国人逐拍标注心电图数据库。本工作利用来自北京协和医院从1995年至2000年之间的1200条Holter心电记录,结合对临床信息数据库的查询和心电图相关医学知识,选取出了具有代表性的待标注心电数据段。这些心电图数据采用在心电信息领域被广泛地应用的MIT-BIH数据格式,心电图机的质量检测人员或者技术研发人员利用该数据格式及相关开源软件,可以进行心电波形的查看,数据分析和处理,以及不同数据格式之间的相互转换。为了帮助临床医生顺利完成对心电波形的逐拍标注工作,本工作还配置出了一套基于移动存储设备的医生用标注工具。该标注工具的硬件部分可以是一个移动硬盘或者一个U盘,在这个移动存储设备上安装有开源的Ubuntu操作系统和标注软件。临床医生可以在任何一台个人台式电脑或者笔记本电脑上使用该标注工具,标注过程所需要的软件、待标注心电图数据及其预标注文件都只需存进该移动存储设备中,医生完成标注后的标注文件及其中间文件都会直接存入移动存储设备中。这样一种标注工具既方便了临床医生的使用,同时也保证了心电图数据及其标注文件的安全性。最后,在对医生标注结果进行了比对、分析和处理之后,就形成了中国人逐拍标注心电图数据库,其中共包括65条时长为30分钟10秒的心电数据段。这些标注文件所涉及的心拍类型完全覆盖了IEC标准中的正常心拍类型“N”、心房早搏类型“S”、心室早搏类型“V”、室性心拍和正常心拍的融合类型“F”、起搏心拍类型“Q”和不可读心拍类型“U”。因此,利用该数据库,可以按照IEC标准实现对心电图机自动诊断功能的性能检测工作。
[Abstract]:An increasing number of electrocardiograph machines are equipped with automatic ECG diagnostic systems for identifying cardiac beat types, and before their launch, Relevant quality inspection departments must use the ECG database manually marked by medical experts to test their performance. The International Electrotechnical Commission has specified in its standards for the testing of ECG equipment, An ECG database used to verify an automatic ECG diagnosis algorithm, It must be collected from the specific population expected to be used by the electrocardiograph machine. This work is to construct such a database of Chinese electrocardiogram (ECG) which is used to test the automatic diagnosis algorithm. This work uses 1, 200 Holter ECG records from Beijing Union Hospital from 1995 to 2000, combined with the inquiry of clinical information database and the medical knowledge of electrocardiogram. The representative ECG data segments are selected. These ECG data are in the MIT-BIH data format which is widely used in the field of ECG information. Using the data format and the related open source software, the quality inspectors or technical researchers of the ECG machine can view the ECG waveform, analyze and process the data, and convert the different data formats to each other. In order to help clinicians to complete the work of ECG waveform tagging, This work also has a set of doctor's marking tools based on removable storage device. The hardware part of the tool can be a portable hard disk or a U disk. The removable storage device has an open source Ubuntu operating system and tagging software that clinicians can use on any desktop or laptop computer to mark the software needed for the process. The ECG data to be tagged and its prelabeled files only need to be stored in the removable storage device. The annotated documents and their intermediate files are stored directly in the mobile storage device. This tool not only facilitates the use of clinicians, but also ensures the safety of ECG data and its tagging documents. Finally, after comparing, analyzing and processing the results of the doctor's labeling, a Chinese ECG database was formed. There are 65 ECG data segments with 30 minutes 10 seconds duration. The type of beat involved in these tagging documents completely covers the normal beat type "N", atrial premature beat type "S", ventricular premature beat type "V" and ventricular beat type in IEC standard. "F", "Q" and "U", respectively. By using this database, the function of ECG automatic diagnosis can be tested according to IEC standard.
【学位授予单位】：北京协和医学院
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：R318.0

【相似文献】