摘要
超声诊断技术用于婴儿发育性髋关节发育不良(developmental dysplasia of the hip,DDH)早期筛查,大大降低了DDH的晚期发病率及手术率。各地髋关节超声早期筛查模式不一,现有的检查方法自动化程度低,彻底推广工作量大。人工智能(artificial intelligence,AI)因其有基于机器学习与深度神经网络的图像特征提取能力,可以进行图像特征提取与识别,借助机械臂的应用能实现标准图像自动获取,AI在髋关节超声筛查中的机遇与挑战并存,为DDH超声筛查开辟新的途径。现有DDH图像的AI研究都是建立在髋关节未发生脱位的标准图像诊断方面,未解决动态图像中自动识别并捕获,也未实现脱位的髋关节图像特征判定与诊断。实现智能化的髋关节超声检查,需要解决机械臂的人工智能、图像识别与图像诊断的人工智能,研发髋关节超声检查机器人系统,建立同质化的筛查体系。本文就DDH早期超声诊断中的图像特征性判定、AI技术在髋关节超声图像中识别和测量及存在问题与关键技术突破解决方案进行分析探讨,以期为临床实践提供一定的指导。
Ultrasonic screening technology has been applied for early screening of developmental dysplasia of the hip(DDH).It greatly reduces the incidence rate and operative rate of DDH.Modes of early hip ultrasonic screening differ greatly in regions.The existing examinations have a low degree of automation and a large workload of thorough promotion.Artificial intelligence(AI)is capable of extracting and recognizing image features because of its functions of machine learning and deep neural network.With the application of mechanical arm,it can automatically obtain standard images.AI offers both ample opportunities and great challenges in ultrasonic screening of hip.It opens up a new pathway for DDH ultrasonic screening.AI research of existing DDH images is based upon the standard image diagnosis of non-dislocation of hip joint.It does not resolve the problem of automatic recognition and capture in dynamic images,nor does it realize the feature judgment and diagnosis of dislocation hip joint image.To realize intelligent hip ultrasonic examination,clinicians need to solve the AI of manipulator,image recognition and image diagnosis,develop hip ultrasonic examination robot system and establish homogeneous screening system.This review focused upon image characteristic judgment in the early ultrasonic diagnosis of DDH,the recognition and measurement of AI technology in ultrasonic image of hip joint,outstanding problems and breakthrough solutions of key technologies to provide guidance for clinical practices.
作者
王加宽
李德宇
崔建伟
李煊鹏
Wang Jiakuan;Li Deyu;Cui Jianwei;Li Xuanpeng(Department of Pediatric Orthopedics,Yangzhou Maternal&Child Healthcare Hospital,Affiliated Hospital,Medical College of Yangzhou University,Yangzhou 225002,China;School of Instrument Science&Engineering,Southeast University,Nanjing 211499,China)
出处
《中华小儿外科杂志》
CSCD
北大核心
2023年第9期861-864,F0003,共5页
Chinese Journal of Pediatric Surgery
基金
扬州市十三五领军人才(创新团队)(LJRC201823)
江苏省妇幼健康科研项目(F202114)。
关键词
髋脱位
发育性髋关节发育不良
超声检查
人工智能
Hip dislocation
Developmental dysplasia of the hip
Ultrasonography
Artificial intelligence