MRI在华东地区汉族男性青少年手腕骨骺发育的量化及骨龄评估中的应用  

MRI Application in Quantification of Epiphyseal Development in the Wrist and Bone Age Estimation of Han Male Adolescents in East China

作　　者：周智露 张东飞 陈捷敏 王亚辉 郝虹霞 刘太昂 何宇亨 龙定念 刘瑞珏 万雷 ZHOU Zhi-lu;ZHANG Dong-fei;CHEN Jie-min;WANG Ya-hui;HAO Hong-xia;LIU Tai-ang;HE Yu-heng;LONG Ding-nian;LIU Rui-jue;WAN Lei(School of Forensic Medicine,Guizhou Medical University,Guiyang 550009,China;Anhui Tianheng Forensic Judicial Identification Institute,Fuyang 236000,Anhui Province,China;Shanghai Key Labora-tory of Forensic Medicine,Key Laboratory of Forensic Science,Ministry of Justice,Shanghai Forensic Ser-vice Platform,Academy of Forensic Science,Shanghai 200063,China;Shanghai Shuzhiwei Information Technology Co.,Ltd,Shanghai 200444;Guangxi Jingui Judicial Expertise Center,Nanning 530000,China)

机构地区：[1]贵州医科大学法医学院,贵州贵阳550009 [2]安徽天衡司法鉴定所,安徽阜阳236000 [3]司法鉴定科学研究院、上海市法医学重点实验室司法部司法鉴定重点实验室、上海市司法鉴定专业技术服务平台,上海200063 [4]上海数之微信息科技有限公司,上海200444 [5]广西金桂司法鉴定中心,南宁广西530000

出　　处：《法医学杂志》2024年第6期589-596,607,共9页Journal of Forensic Medicine

摘　　要：目的探索上海、浙江、江苏地区男性青少年手腕部MRI在骨龄评估中的价值。方法选取124例上海、浙江、江苏地区6.0~18.0岁汉族男性作为研究对象,测量身高、体质量,对其手腕部行T1WI、T2WI序列扫描。获取手腕部MRI图像后,选择桡骨、尺骨、第一至五掌骨干骺端及对应骨骺作为观察指标,由一名副高级影像学专家对各指标的发育情况进行分级(0~2级),另一名专家测量各指标的最大宽度。将身高、体质量、各指标分级及最大宽度测量值等作为输入变量,年龄作为目标变量,利用SPSSModeler软件建立支持向量机、随机森林、当前现实树、线性回归模型评估骨龄,并选取准确率最高的模型。结果身高、体质量、手腕骨骨骺发育分级、各骨干骺端及对应骨骺的最大宽度均与年龄存在相关性(P<0.05)。骨龄与实际生活年龄差值在1.0岁与1.5岁以内准确率最高的模型均为支持向量机(88.7%、96.0%)。结论MRI图像用于骨龄评估具有可行性,量化骨骺及对应骨骼干骺端最大宽度并结合MRI图像分级法可有效减少预测误差。Objective To investigate the value of wrist MRI in bone age estimation for male adolescents in Shanghai,Zhejiang and Jiangsu.Methods A total of 124 Han male adolescents aged 6.0 to 18.0 years from Shanghai,Zhejiang and Jiangsu were selected as subjects.Their weight and height were measured,and T1WI and T2WI sequences of the wrist were scanned.The distal ends of the radius and ulna,and the first to five metacarpal epiphyses and corresponding metaphyses were selected as observational indexes after MRI images of the wrist were obtained.The development of each index was classified(0-2 grades)by a deputy senior imaging expert,then the maximum width of each index was measured by another deputy senior expert.Height,weight,classification and maximum width of indexes were used as input variables,and age was used as the target variable.Support vector machine,random forest,current reality tree,and linear regression models were established to estimate the bone age,and the model with the highest accuracy was selected.Results The height,weight,classification of wrist bone epiphysis development,maximum width of each bone metaphysis and epiphysis were all correlated with age(P<0.05).The accuracies of the support vector machine were the highest when the differences between bone age and actual chronological age were within 1.0 and 1.5 years(88.7%and 96.0%,respectively).Conclusion It is feasible to estimate bone age by using MRI images.Quantifying the maximum width of the epiphysis and corresponding metaphysis of bone and combining it with MRI image classification can effectively reduce the estimation error.

关 键 词：法医人类学 年龄推断 磁共振成像 手腕部 骨骺 骨发育分级 机器学习 青少年 

分 类 号：R89[医药卫生—法医学] DF795.6[医药卫生—临床医学] D919.6[政治法律—诉讼法学]