基于深度学习的CT室放射防护状况智能检测技术研究  

Research on intelligent detection technology of radiation protection in CT room based on deep learning

作　　者：韩壮 李世蛟 杨恺 马彪 于航 王耀东[2] HAN Zhuang;LI Shijiao;YANG Kai;MA Biao;YU Hang;WANG Yaodong(Beijing To First Technology Co.,Ltd,Beijing 100085,China;School of Mechanical and Electronic Control Engineering,Beijing Jiaotong University)

机构地区：[1]北京争上游科技有限公司,北京100085 [2]北京交通大学机械与电子控制工程学院

出　　处：《中国数字医学》2024年第10期96-101,共6页China Digital Medicine

摘　　要：针对医院CT室人员的复杂性以及患者在进行放射治疗过程中潜在的安全风险,设计了放射治疗过程中操作规范性判别的智能视频检测系统。该系统以医院CT室内的真实视频监控影像数据为基础,构建了多目标样本标注图像数据集,并采用了YOLOv5网络来检测设备状态、医护人员与患者防护装备的穿戴情况。通过对目标特征进行智能分析判断,从而实现放射治疗过程中人员操作行为规范性的实时监测。实验结果显示,该方法能够有效分析实时采集到的视频信息,实现CT室内多种目标的准确识别以及对患者防护服穿戴情况的有效监控。In view of the complexity of the personnel in hospital CT room and the potential safety risks of patients during the process of radiotherapy,this paper designed an intelligent video detection system for judging the operation regularity in the process of radiotherapy.Based on the real video surveillance image data in the CT room of the hospital,the system constructed a multi-target sample annotated image dataset,and adopted YOLOv5 network to detect equipment status,and the wearing of protective equipment of medical staff and patients.By intelligently analyzing and judging the target features,the real-time monitoring of the standardization of personnel's operation behavior during radiotherapy was realized.The experimental results show that this method can effectively analyze the video information collected in real time,and realize the accurate identification of multiple objects in the CT room as well as the effective monitoring of the patient's wearing of protective clothing.

关 键 词：深度学习 视频检测 目标检测 放射防护 YOLO 

分 类 号：R197.3[医药卫生—卫生事业管理] R319[医药卫生—公共卫生与预防医学]