面向智能诊断的数字病理标注管理系统的设计  

Design of the digital pathology labeling management system for intelligent diagnosis

作　　者：丁偕[1] 刘鸣 张传国 Ding Xie;Liu Ming;Zhang Chuanguo(Wonders Information Co.,Ltd.,Shanghai 201112,China)

机构地区：[1]万达信息股份有限公司,上海201112

出　　处：《中国数字医学》2022年第1期83-88,共6页China Digital Medicine

基　　金：长三角全数字智能病理远程诊断平台(沪经信智〔2019〕1014号)。

摘　　要：目的:通过构建面向智能诊断的数字病理标注管理系统,满足病理智能诊断中模型训练对标注数据的要求,从而加速图像识别技术、AI技术在医学诊断,尤其在病理诊断方面的应用,提高诊断的效率与准确率。方法:系统通过OpenSlide实现将不同格式数字病理切片转换成为DZI格式,应用端使用OpenSeadragon实现大尺寸、高分辨率图像展示,标注绘制功能基于Paper.js插件提供铅笔、多边形等多种工具,服务端使用Springboot框架实现对业务流程的控制。结果:实现了系统设计的总体功能,系统能够满足对大尺寸、高分辨率病理图像的标注与管理工作。结论:系统解决现有标注数据采集系统存在的问题,有效提升获取大尺寸、高分辨率病理图像上标注数据的效率,并帮助管理者进行标注数据质量的维护。Objective By constructing the digital pathological annotation management system for intelligent diagnosis,to meet the requirements of model training for annotation data in intelligent diagnosis of pathology,so as to accelerate the application of image recognition technology and AI technology in medical diagnosis and especially pathological diagnosis,and improve the efficiency and accuracy of diagnosis.Methods The system converts digital pathological slices in different formats into DZI format through OpenSlide.The application side uses OpenSeadragon to display images with a big size and high resolution.Based on Paper.js plug-in,the annotation and drawing function provides pencil,polygon and other tools.On the service side,the Springboot framework is used to control the business process.Results The overall function of the system design is realized.The system can complete the annotation and management work of pathological images with a big size and high resolution.Conclusion The system solves the problems in the existing annotation data acquisition system,effectively improves the efficiency of obtaining annotation data on pathological images with a big size and high resolution,and helps managers to maintain the quality of annotation data.

关 键 词：在线标注 高通量数字图像 DeepZoom OpenSlide Paper.js 

分 类 号：R319[医药卫生—基础医学]