三维颅颌面手术设计系统关键技术研究

项目名称： 三维颅颌面手术设计系统关键技术研究

项目编号： No.61272225

项目类型： 面上项目

立项/批准年度： 2013

项目学科： 自动化技术、计算机技术

项目作者： 陈莉

作者单位： 清华大学

项目金额： 80万元

中文摘要： 颅颌面是人体功能、形象和情感的重要部位，高质量颅颌面手术设计软件对提高手术质量、减少病人痛苦具有重要意义。然而目前该类软件的研发存在一些关键研究问题尚未解决，对其进行研究在理论上和临床中均具有重要意义。本项目旨在探索基于样本模板、形变模型的自学习颅颌面骨块分割方法、基于骨性标志点自动提取的高效颅颌面数据库检索方法、可突出医生感兴趣组织结构的高质量智能体绘制方法、基于样例图像的高真实感绘制参数自学习方法、基于用户行为模式分析的三维体拾取方法，解决颅颌面手术设计软件存在的交互量太大、显示智能性不够、数据库难以很好地辅助手术设计等关键难点问题，进而通过研究手术流程可灵活定制的基于GPU的实时体绘制架构，突破目前商业医学软件普遍采用的以数据为中心所带来的功能和质量瓶颈，借助北京大学口腔医院拥有的目前国内最大颅颌面数据库，最终实现一个实时、高质量数据库辅助颅颌面手术设计原型系统。

中文关键词： 医学可视化；手术设计软件；三维体交互；三维体分割；GPU实时体绘制

英文摘要： As an essential area of human body, Craniomaxillofacial Complex handles the most important parts of anatomical functionalities and daily communication. Craniomaxillofacial surgery design under the guidance of software requires both the high precision and better usability, which poses a great challenge. However, some of the key issues remain unsolved, restricting the functionality of the surgical design software. Thus research on software guided craniomaxillofacial surgeries holds great significance both theoretically and clinically. This project aims at providing a better software solution for craniomaxillofacial surgeries by researching and developing a variety of new techniques, including sample templates and morphable-model based self-learning bone segmentation, bony-landmark-extraction based database retrieval, high-quality intelligent volume rendering highlighting the interested organs and structures, sample-image based self-learning optimization of high-quality rendering and user-behavior guided volume-pickup. By using a self-designed, feature-rich, and real-time GPU rendering framework, this project is to overcome the performance difficulties and quality bottlenecks faced by the current data-centered commercial medical software. With the largest domestic craniomaxillofacial database (up to 10000 cases) pr

英文关键词： Medical Visualization；Surgery CAD Software；3D Volume Interaction；3D Volume Segmentation；Realtime GPU Volume Rendering