人工智能软件在儿童发育相关疾病骨龄评估中的临床应用  

Clinical Application of Artificial Intelligence Software in Bone Age Assessment ofDevelopmental Related Diseases in Children

作　　者：邓成清 孙安林 唐宇轩[2] 张萍[1] 唐爱群[2] DENG Cheng-qing;SUN An-lin;TANG Yu-xuan;ZHANG Ping;TANG Ai-qun(Department of Radiology,Sheyang County People's Hospital,Yancheng City,Yancheng Jiangsu 224300,China;Department of Pediatrics,Sheyang County,People's Hospital,Yancheng City,Yancheng Jiangsu 224300,China)

机构地区：[1]盐城市射阳县人民医院放射科,江苏盐城224300 [2]盐城市射阳县人民医院儿科,江苏盐城224300

出　　处：《中华养生保健》2023年第22期54-57,共4页CHINESE HEALTH CARE

摘　　要：目的探讨与分析人工智能软件在儿童发育相关疾病骨龄评估中的临床应用价值。方法选取2020年5月—2022年12月在江苏省盐城市射阳县人民医院诊治的84例发育相关疾病儿童作为研究对象,患儿都给予左手数字化X线检查,采用杭州深睿博联科技有限公司人工智能影像分析软件对数字化X线摄片进行骨龄评估,同时也由影像科住院医师与影像科主任医师进行阅片,以影像科主任医师的判断结果作为金标准,判定评估价值。结果人工智能阅片时间显著少于住院医师阅片与金标准阅片,差异有统计学意义(P<0.05)。住院医师阅片与金标准阅片时间比较,差异无统计学意义(P>0.05)。人工智能阅片误差与金标准阅片比较,差异无统计学意义(P>0.05),住院医师阅片误差与金标准阅片比较,差异无统计学意义(P<0.05)。住院医师阅片、人工智能阅片、金标准阅片准确率分别为88.10%、97.62%、98.81%,人工智能阅片、金标准阅片的准确率显著高于住院医师阅片,差异有统计学意义(P<0.05)。人工智能阅片、金标准阅片的准确率比较,差异无统计学意义(P>0.05)。结论人工智能软件在儿童发育相关疾病骨龄评估中的临床应用能缩短阅片时间,提高对于骨龄判定的准确性,缩小阅片的误差。Objective To explore and analyse the clinical application values of artificial intelligence software in bone age assessment of developmental related diseases in children.Methods 84 cases of children with developmental related diseases diagnosed and treated at a certain hospital from May 2020 to December 2022 were selected as the research subjects.All children were underwent left hand digital X-ray examination,and bone age assessment were conducted using artificial intelligence imaging analysis software from Hangzhou Shenrui Bolian Technology Co.,Ltd.The digital X-ray films were also reviewed by resident imaging physicians and chief imaging physicians.Used the chief physician of the imaging department as the gold standard to determine the evaluation value.Results AI reading time was signifi cantly less than that of residents and gold standard(P<0.05),and there was no signifi cant diff erence between residents and gold standard(P>0.05).There was no signifi cant diff erence between the AI reading error(P>0.05),and the resident and gold standard reading error(P<0.05).The accuracy of resident reading,artificial intelligence reading and gold standard reading was 88.10%,97.62%and 98.81%respectively.The accuracy of AI reading and gold standard reading was signifi cantly higher than that of resident reading(P<0.05),and there was no significant difference in the accuracy of AI reading and gold standard reading(P>0.05).Conclusion The clinical application of artifi cial intelligence software in the assessment of bone age in children with developmental related diseases can shorten the reading time,improve the accuracy of bone age determination,and reduce the error of reading.

关 键 词：人工智能软件 儿童发育相关疾病 骨龄 阅片时间 准确性 误差 

分 类 号：R179[医药卫生—妇幼卫生保健]