人工智能在全髋关节置换术前规划的应用  

Application of artificial intelligence in preoperative planning for total hip arthroplasty

作　　者：郑勇强[1] 张金山[1] 林亮[1] 林振宇[1] 黄晓鹏 陈雪婷 刘晓峰 HENG Yong-qiang;ZHANG Jinshan;LIN Liang;LIN Zhen-yu;HUANG Xiao-peng;CHEN Xue-ting;LIU Xiao-feng(Department of Orthopedics,Jinjiang Hospital,Quanzhou 362000,China;Department of Acupuncture and Rehabilitation,Dehua Hospital of Traditional Chinese Medicine,Quanzhou 362500,China)

机构地区：[1]福建省晋江市医院骨科,福建泉州362000 [2]福建省德化县中医院针灸康复科,福建泉州362500

出　　处：《中国矫形外科杂志》2024年第15期1436-1440,共5页Orthopedic Journal of China

基　　金：福建省卫生健康中青年科研重大项目(编号:2022ZQNZD015);福建省中医药大学校管科研课题(编号:XB2022149;XB2023197);晋江市医院(上海市第六人民医院福建医院)科技计划项目(2023LC01)。

摘　　要：[目的]探讨人工智能(artificial intelligence,AI)在全髋关节置换术(total hip arthroplasty,THA)术前假体测量的准确性。[方法]选取2021年6月—2021年12月收治的24例髋关节残疾患者,按照随机数字表法分为AI组(12例)及常规组(12例)进行假体型号预测,分析AI三维规划的准确性及可行性。[结果]两组手术时间、总失血量、术后住院天数以及住院总费用的差异均无统计学意义(P>0.05)。AI组在髋臼杯一致性[(±0/±1/±2),(91.7%/8.3%/0%)vs(33.3%/41.7%/25.0%),P=0.013]和股骨假体型号一致性的准确度[(±0/±1/±2),(83.3%/8.3%/8.3%)vs(41.7%/33.3%/25.0%),P=0.017]显著高于常规组。随访时间平均(15.2±5.7)个月,与术前相比,末次随访时两组VAS评分均显著降低(P<0.05),髋伸曲ROM和Harris评分显著增加(P<0.05)。相应时间点,两组VAS评分、髋伸曲ROM和Harris评分的差异均无统计学意义(P>0.05)。[结论]AI三维规划较胶片模板测量在THA术前假体型号的选择具有更精准的预测价值。[Objective]To investigate the accuracy of artificial intelligence(AI)in the measurement of prostheses before total hip arthroplasty(THA).[Methods]A total of 24 patients who were undergoing THA for endstage hip arthropathies from June 2021 to December 2021 were divided into AI group(12 cases)and conventional group(12 cases)according to random number table method.The accuracy and feasibility of AI preoperative planning predicting prosthetic size were analyzed.[Results]There were no significant differences in operation time,total blood loss,postoperative hospitalization days and total hospitalization costs between the two groups(P>0.05).AI group proved significantly superior to the conventional group in terms of consistency of the acetabular component[(+0/+1/+2),(91.7%/8.3%/0%)vs(33.3%/41.7%/25.0%),P=0.013]and the femoral component[(+0/+1/+2),(83.3%/8.3%/8.3%)vs(41.7%/33.3%/25.0%),P=0.017].With time of follow-up period lasted for(15.2±5.7)months in a mean,the VAS scores significantly decreased(P<0.05),while hip flexion-extension ROM and Harris scores significantly increased in both groups(P<0.05),whereas which were not significantly different between the two groups at any corresponding time points(P>0.05).[Conclusion]This AI preoperative planning has more accurate predictive value than the film template measurement in the selection of prosthetic size before THA.

关 键 词：终末髋关节病 全髋关节置换 人工智能三维规划 胶片模板测量 

分 类 号：R687[医药卫生—骨科学]