同种异体骨移植手术中应力分析与三维特征匹配研究

At present, allograft bone transplantation is the mainstream approach to repair bone defects, where the designed transplantation plan and transplantation accuracy greatly impact the surgery effect. Whether clinicians can achieve the accurate allograft bone transplantation or not mainly depends on the following key problems: (1) how to design the cutting plane which meets the biomechanical demand; (2) how to utilize the three dimensional information of bones to efficiently and effectively select the best matched allograft bone; (3) how to scientifically and objectively evaluate the patient’s postoperative performance. (4) how to model the relationship between the bone density and effect of bone reconstruction. In allusion to the above problems, this project plans to utilize stress analysis and three dimensional feature matching to build an accurate scheme for allograft bone transplantation as follows: Firstly, this project will propose a cutting plane design method based on image analysis and finite element analysis, to plan the cutting plane which could meet the biomechanical demand. Secondly, this project will propose an allograft bone matching method based on feature learning by convolutional neural network to automatically select the best matched allograft bone. Then, this project will propose an error analysis and finite element analysis based postoperative evaluation scheme, to precisely assess the accuracy of allograft bone transplantation and also the biomechanical circumstance around the operation site. Finally, this project will model the relationship between the bone density and the reconstruction effect through extensive experiments and statistical analysis, to optimize the cutting plane designing and allograft bone matching. The research results of this project will not only improve the success rate of the allograft transplantation surgeries, but also promote the development of precision medicine in other related branches.

同种异体骨移植是目前修复骨缺损的主流方法，其手术方案及精度极大影响着手术效果。手术能否合理、精准实施的关键在于：1）如何设计符合生物力学要求的切割平面；2）如何利用骨骼三维信息高效筛选异体骨；3）如何科学评估手术效果；4）如何建立异体骨骨质密度分布与骨骼重建效果之间的关系模型。针对以上问题，本项目拟利用应力分析、三维特征匹配等关键技术打造精准异体骨移植手术方案：首先，提出基于应力分析的切面设计方法，规划能够营造良好生物力学环境的切面设计方案；其次，提出基于卷积神经网络的三维特征匹配算法，实现最佳异体骨的自动筛选；然后，提出基于几何误差分析和应力分析的术后评估方法，精确评估手术完成质量及植骨部位生物力学环境；最后，通过实验设计与统计分析构建异体骨骨质密度分布与骨骼重建效果之间的关系模型，优化切面设计及异体骨筛选方法。研究成果不仅可以提高异体骨移植手术成功率，而且可以推进相关学科的医疗精准化。

影像导航系统的介入可以使术者完成对骨肿瘤外科边界的精确切除，保留更多的骨骼结构，为骨切除后定制异体骨等复杂重建创造条件。然而，目前骨肿瘤治疗中的术前切面设计、同种异体骨筛选和术中切除重建等步骤仍依赖手工操作，可能引入较大外科误差进而导致肿瘤复发、转移等，并且现有的术后评估方法尚不完善，不利于医生掌握手术完成精度。因此，研究精准的术前切面规划、异体骨筛选及客观的术后评估对骨肿瘤精准治疗具有重要的理论研究意义和临床使用价值。针对影像导航下骨肿瘤手术中手工操作易引入外科误差的问题，本项目提出了一系列算法用于解决骨肿瘤精准治疗涉及的三个关键科学问题，即术前切面规划、异体骨筛选和术后外科误差分析。本项目提出的三个方法分别在从北京积水潭医院收集的临床数据上进行了大量实验验证。结果表明，三个方法相较于传统方法可以显著提升骨肿瘤治疗精度。此外，本项目提出的部分算法已应用于治疗骨肿瘤的临床实践中，推进了精确医疗的发展。