临床骨科杂志2025，Vol.28Issue(1)：20-24,5.DOI:10.3969/j.issn.1008-0287.2025.01.007
改进后的U-net神经网络新模型在青少年特发性脊柱侧弯患者Cobb角测量中的应用价值
Application value of improved U-net neural network new model in Cobb angle measurement of the adolescent idiopathic scoliosis patients
摘要
Abstract
Objective To investigate the application value of the improved U-net neural network new model(new model method)based on artificial intelligence(AI)deep learning technique for Cobb angle measurement of adoles-cent idiopathic scoliosis(AIS)patients.Methods The full-length and positive-lateral X-ray films of 46 AIS Lenke type-1 patients were selected,and Cobb angles were measured by using the new model method,Surgimap method and Photoshop method.The Cobb angles and taken time were compared among different measurement methods.The consis-tency tests of Cobb angle measurements were performed on the different groups of measurers using Surgimap method and Photoshop method,as well as for the same group of measurers.Drawing a Bland-Altman plot was used to evaluate the consistency of data measured by different methods,in order to further validate the evaluation of intraclass correla-tion coefficient(ICC).Results There was no statistical significance in the Cobb angles measured by the three meas-urement methods(P>0.05);compared with Surgimap and Photoshop methods,the new model method had the shortest time in the Cobb angle meausrement,and the differences were statistically significant(P<0.05).The Surgimap meth-od and Photoshop method had high reliability and good repeatability,excellent consistency for Cobb angle measure-ment by different groups of measurers and the same group of measurers.There were slight variations in Cobb angles measurement and selection of upper and lower vertebrae among different measurers using the three measurement meth-ods,but overall consistency was good.Conclusions The improved U-net neural network new model based on AI deep learning technique has the advantages of high reliability and fast speed for measuring Cobb angles in AIS patients,and its accuracy has reached the level of senior specialist physicians.关键词
青少年特发性脊柱侧弯/Cobb角/人工智能/U-netKey words
adolescent idiopathic scoliosis/Cobb angle/artificial intelligence/U-net分类
医药卫生引用本文复制引用
李成,柯宝毅,胡巍,马文宇,王子民,覃军焱..改进后的U-net神经网络新模型在青少年特发性脊柱侧弯患者Cobb角测量中的应用价值[J].临床骨科杂志,2025,28(1):20-24,5.基金项目
广西壮族自治区桂林市科技局计划项目(编号:20210227-2) （编号:20210227-2）
广西壮族自治区卫生健康委员会自筹课题(编号:Z20210639) （编号:Z20210639）
