中文摘要：

目的：运用支持向量机（support vector machine，SVM）实现尺、桡骨远端骨骺发育分级的自动化评估。方法收集我国140例11～19周岁青少年左侧腕关节X线正位片作为训练样本。将尺、桡骨远端骨骺分为五个发育分级，每个分级均包含28例样本。另选35例作为独立校验样本。建立尺、桡骨远端骨骺五个发育分级的SVM分类模型，用留一交叉验证法（leave one out cross validation，LOOCV）进行模型内部交叉验证以及梯度方向直方图（histogram of oriented gradient，HOG）进行模型外部验证，分别计算其准确率（PA）。结果桡骨远端骨骺分级SVM建模、LOOCV和HOG的PA分别为100.0%、78.6%和82.8%。尺骨远端骨骺分级SVM建模、LOOCV和HOG的PA分别为100.0%、80.0%和88.6%。结论运用SVM建立的尺、桡骨远端骨骺发育分级的自动化模型具有一定的可行性，为法医学骨龄评估软件的开发奠定基础。

英文摘要：

Objective To realize the autom ated assessm ent of the levels of epiphysis of distal radius and ulna by support vector m achine （SVM）. Methods The X-ray film s of the leftwrist jointswere taken from 140 teenagers aged from 11 to 19 years old as training sam ples. The levels of epiphysis of distal radius and ulnawere divided into five developm ental levels. Each level contained 28 sam ples. A nother 35 cas-eswere selected as independent verifying sam ples. SVM classification m odels of the five developm ental levels of epiphysis of distal radius and ulnawere established. The internal cross validationwas m ade by leave one out cross validation （LOOCV ）,while the external validationwas m ade by histogram of orient-ed gradient （HOG）, and then the accuracy （PA ） of testing resultswas calculated, respectively. Results The PA of SVM, LOOCV and HOG of distal radius epiphyseal levelwere 100%, 78.6%, and 82.8%, respec-tively;whereas the PA of SVM, LOOCV and HOGof distal ulna epiphyseal levelwere 100.0%, 80.0%and 88.6%, respectively. Conclusion The SVM -based autom atic m odels of the growth stage of distal ra-dius and ulna appear to have certain feasibility, and m ay provide a foundation for software developm ent of bone age assessm ent by forensic medicine.