基于人工智能的数字X线机自助检查系统的设计

Design of digital X-ray machine self examination system based on Artificial Intelligence

摘要:

为了实现数字X线机检查的体位识别智能化与摆位自助化，采用自主研发的基于堆叠沙漏网络的体位识别装置、应用动作相关关系模型的体位判断装置和加入语音提示功能的数字X线机“自助检查”装置，解决了如今影像摄片体位精准度不高、影像科技术人员数量不足、传统摄片耗时较长等问题。经多次实验结果表明，系统将原本人均将近6分钟的摄影时间缩短到大约三分半，将甲级片出片率的不足40%提升到将近80%，进而将误诊率减少30%，同时能够有效地减少医患接触，减低院内感染的风险，系统实现了缩短摄片时间、提高摄影质量、降低院内感染的目标。

Abstract:

In order to realize the intelligent position recognition and self-help positioning of digital X-ray machine examination, the self-developed position recognition device based on stacked hourglass network, position judgment device applying action correlation model and "self-service examination" device of digital X-ray machine with voice prompt function are adopted to solve the problems of low position accuracy of image photography, the number of technicians in the imaging department is insufficient, and the traditional photography takes a long time. The results of many experiments show that the system reduces the original photography time of nearly 6 minutes to about one-third and a half, increases the production rate of class a films from less than 40% to nearly 80%, and then reduces the misdiagnosis rate by 30%. At the same time, it can effectively reduce the contact between technician and patients and reduce the risk of nosocomial infection. The system can shorten the photography time, improve the photography quality and achieve the goal of reducing nosocomial infection.