基于深度学习OCR的医疗设备质控检测原始记录表智能识别系统的设计与应用  

Design and Application of Intelligent Recognition System for Medical Equipment Quality Control Testing Original Record Forms Based on Deep Learning OCR Technology

作　　者：林艺文 LIN Yiwen(Department of Equipment and Materials,Zhongshan Hospital Xiamen University,Xiamen Fujian 361004,China)

机构地区：[1]厦门大学附属中山医院设备物资部,福建厦门361004

出　　处：《中国医疗设备》2024年第9期54-61,共8页China Medical Devices

基　　金：福建省卫生健康科技计划(2021RKB010)。

摘　　要：目的为了提高纸质医疗设备质控检测原始记录表手写数据的电子化录入效率,替代传统手工录入方式,实现手写检测数据的批量化自动录入。方法基于Python语言,开发一套基于深度学习光学字符识别(Optical Character Recognition,OCR)的医疗设备质控检测原始数据记录表智能识别系统。深度学习OCR技术采用百度智能云OCR云服务,实现批量识别质控检测记录表电子图片,获取结构化的检测数据识别结果,并将识别结果以电子表格的形式导出。结果该系统已实现8种常用医疗设备质控检测原始记录表的智能化识别,经实验测试,8种质控检测记录表平均识别耗时为5.45 s,平均识别正确率为95.94%。系统应用后,医疗设备质控检测原始记录表手写数据电子化录入用时显著低于传统手工录入方式,且差异有统计学意义(P<0.001)。结论该系统识别速度快,识别正确率高,实现了医疗设备质控检测原始记录表批量化、智能化、电子化自动录入,节省了大量人力,提高了质控检测数据整理效率,为质控检测数据的深度分析打下坚实基础。Objective In order to improve the electronic input efficiency of the handwritten data in paper-based medical equipment quality control testing original record forms,to replace the traditional manual input method,and realize batch automatic entry of handwritten testing data.Methods Based on Python language,an intelligent recognition system for medical equipment quality control testing original record forms based on deep learning optical character recognition(OCR)technology was developed.The deep learning OCR technology used Baidu AI Cloud OCR cloud service.The system could batch recognize electronic images of quality control testing record forms,obtain structured testing data recognition results,and export them to spreadsheets.Results The system has achieved intelligent recognition of 8 common medical equipment quality control testing original record forms.Through experimental tests,the average recognition time of 8 kinds of quality control testing records was 5.45 s,and the average recognition accuracy was 95.94%.After the application of the system,the electronic input time of handwritten data in the original medical equipment quality control testing record form was significantly lower than the traditional manual input method,and the difference was statistically significant(P<0.001).Conclusion The system has fast recognition speed,high recognition accuracy,and achieves batch,intelligent,and electronic data input of medical equipment quality control testing original record forms,which saves a lot of manpower,improves the efficiency of quality control testing data collection and lays a good foundation for quality control testing data analysis.

关 键 词：医疗设备质控 表格识别 光学字符识别 深度学习 质控记录表 

分 类 号：R197.324[医药卫生—卫生事业管理]