多元影像信息智能化分析引导精准骨科植入的关键问题研究

The incidence of orthopaedic diseases and the demand for minimally invasive orthopaedic implant surgery are both increasing year by year in the world. At present, intraoperative X-ray imaging is widely used for clinical practice to guide minimally invasive orthopaedic implantion. However, this method lacks of personalized implant selection information and intraoperative soft tissue information, which results in high complication rates. To solve these bottleneck problems, this project will explore three scientific methods:1) Combining the advantages of magnetic resonance and computed tomography images, a dual-channel model of image reconstruction and bone structure segmentation is developed to establish a population database of bone morphology; 2) based on the population database of bone, a multi-view model matching method is proposed to realize automatic real-time measurement of bone morphological characteristics for the quantitative selection of implants preoperatively; 3) a fusion method of cascading location and gradient coding is developed to realize the correction of preoperative bone and soft tissue for the radiation-free ultrasound navigation of orthopaedic implantion. With complementary advantages of multivariate imaging information, precise orthopaedic implantion navigation method based on intelligent analysis of multivariate imaging information is realized, which can provide quantified three-dimensional disease and implant information to improve the accuracy of surgery and reduce the complications.

在全球范围内，骨科疾病的发病率越来越高，骨科微创植入手术的需求量也逐年攀升。目前，临床中主要采用术中X线影像来引导骨科微创植入手术，但单一X线影像引导的方式缺乏患者个性化的植入物定量选型信息和术中骨周围软组织信息的治疗指导，导致手术并发症居高不下。为解决该问题，本项目将提出针对骨植入手术导航的多元影像信息智能分析方法，包括：1）开发影像重建-分割双通道模型，结合核磁和CT影像优势，建立人群化的骨形态数据库； 2）基于人群化骨数据库，提出多视角模型匹配方法，实现基于骨形态特征自动实时测算的植入物术前定量选型；3）为获得术中软组织信息，开发了定位与图像梯度编码级联的融合方法，实现病人术前骨及软组织匹配校正，完成无辐射的骨科植入超声导航。通过多元影像信息的优势互补，实现基于多元影像信息智能分析的骨科精准植入导航方案，为医生提供量化的三维病症和植入物信息，提高手术精准性，减少并发症。

随着人口老龄化的加剧和运动健身文化的兴起，骨科微创植入手术需求量逐渐攀升。本项目以中国人群骨科微创植入手术的精准引导为例，研究多元影像信息智能化分析与结合方法，将人群化先验信息和术中软组织关节信息融入手术引导中，实现新型的骨科植入超声导航系统。具体开展了三维骨组织快速提取与重建、术中无辐射的超声分析与结构识别、三维骨结构形态特征自动测算、多元影像信息智能化分析与导航等研究工作。实验结果表明所提出的智能导航方法通过术中局部信息对术前组织结构进行更新矫正，保证了导航信息的准确性，能为医生提供全局精准的引导信息。基于以上研究，已发表医工交叉类期刊和会议论文18篇，申请了国家发表专利12项。这些科学性的解决方案为骨科病变的临床治疗引导提供精确导航的基础理论和技术支撑，可望推动骨外科手术理论及临床治疗的发展与革新。