人工智能辅助DDH诊断的应用:进展和挑战

Current challenges and recent advances in application of artificial intelligence for assisting in diagnosing developmental dysplasia of the hip

儿童骨盆正位X线片是股骨头骨化中心出现以后(4~6个月龄以上)儿童发育性髋关节发育不良的首选检查方法。但是在临床工作中,由于多种原因,依然存在很多对骨盆正位X线片判断不清,导致诊治困难的情况发生。随着人工智能技术的发展,该技术已经广泛应用于医疗领域。人工智能技术应用于儿童骨盆正位X线片辅助诊断中,可以提升阅片效率,提高儿童发育性髋关节脱位的诊断效率;提高诊疗水平,降低因为临床经验不足的原因导致的发育性髋关节脱位的误诊误治疗风险。然而,人工智能模型应用于骨盆正位X线片辅助诊断依然存在一些亟待解决的问题。本综述拟对人工智能和图像识别模型在儿童骨盆正位X线片辅助诊断中的应用进展进行回顾,并分析目前面临的技术挑战和未来该技术的发展方向,为发育性髋关节发育不良的智能化、自动化诊断提供研究思路。

Anteroposterior(AP)pelvic radiograph is a primary choice for diagnosing developmental dysplasia of the hip(DDH)in children aged over(4-6)months.However,many cases persist with uncertain diagnosis or inappropriate treatment due to a misinterpretation of AP pelvic radiographs.As artificial intelligence technology develops rapidly,it has been widely applied in clinical practices.Artificial intelligence technology has been applied for an auxiliary diagnosis of AP pelvic radiographs so as to boost the accurate and efficient interpretations of radiographs to enhance the diagnostics of DDH.Meanwhile,it is also capable of improving the diagnosis and therapy in DDH and lowering the risk of its misdiagnosis and mistreatment due to insufficient clinical experience.However,nowadays there are still several urgent issues to be solved in clinical application of artificial intelligence radiographic model for an auxiliary diagnosis of DDH.This review aims to review the progress of application in the artificial intelligence and image recognition technology for clinical auxiliary diagnosis of DDH based upon AP pelvic radiographs.This review summarized current technical challenges and future directions of this technology so as to provide research rationales for achieving an intelligent and automatic diagnosis of DDH.