基于膀胱壁组织/弹性模型的虚拟膀胱镜辅助检测方法研究

This project studies the mixed bladder tissue model based on multi-scale imaging information by combinating the Magnetic Resonance Image and Optical Coherence Tomography, and establishes elasticity model of bladder wall by imaging the filling bladder multiple times. Furthermore, this study introduces mixed tissue model and elasticity modelinto virtual cystoscope system, which assists physicians to detect bladder tumors. On the one hand, features sensitive to bladder tumors can be extracted for the computer automatic detection of tumors, and the quantitative features for tumor diagnosis can be provided to the physicians as well. On the other hand, through the study on the virtual palpation technology based on virtual cystoscope, information of the soft tissue elasticity, visual feedback and force feedback is provided to physicians to make empirical diagnosis, which provides a new method for the non-invasive detection of bladder tumors. This study gets rid of the traditional research mode of emphasis on the visualization of virtual cystoscope environment and the interactive operation, and greatly promotes the virtual cystoscope technology both on the establishment of new theory models and the exploration of new manners of computer aided detection, which makes the virtual cystocope technology as an effective complement and an expansion for the optical cystoscope, the gold standard of clinical practice. Our group's previous research lays a solid foundation for the further development of this project.

本项目将核磁共振成像和光学相干层析成像相结合，研究基于多尺度影像信息的膀胱混合组织模型；并通过对充盈中的膀胱进行多次成像，建立膀胱壁软组织的弹性模型。进一步，将混合组织模型和组织弹性模型引入到虚拟膀胱镜系统中，辅助医生进行膀胱肿瘤检测。一方面，从模型中提取对肿瘤敏感的特征，实现计算机自动肿瘤检测，并向医生提供肿瘤的量化特征；另一方面，将模型中的弹性属性以视觉和力觉反馈的方式提供给医生，研究基于虚拟膀胱镜的虚拟触诊技术，供医生进行经验性判断。为膀胱肿瘤的无创检测提供一种新方式。本研究摆脱了传统虚拟内窥镜着重虚拟内窥环境可视化和交互操作的研究模式，在建立理论模型和探索新的辅助检测方式方面都对虚拟内窥技术有较大的推动，使之成为临床金标准-光学膀胱镜真正的有效补充和拓展。项目组前期的研究为本项目的开展提供了扎实的基础。

本项目使用核磁共振(Magnetic Resonance Image, MRI)和光学相干层析(Optical CoherenceTomography, OCT)方法对膀胱肿瘤进行成像，对多模态影像下的膀胱肿瘤表达进行研究，为虚拟膀胱镜辅助检测和辅助诊断打下基础。主要完成了如下五项研究内容：一、研究适合于膀胱壁成像的T1加权的MRI核磁序列，并初步获得具有较好膀胱壁对比度以及时间、空间分辨率的成像序列及参数。二、研究基于EM-MAP的双水平集方法，对MRI图像中的膀胱壁进行更精准的分割。三、基于分割结果，对膀胱壁厚度进行精确计算，使用膀胱壁厚度作为膀胱肿瘤出现的指标之一。四、对分割后的膀胱壁进行三维表达和显示，并初步构造弹性模型，用于对重建的三维膀胱壁进行按压、拉伸、切割等操作。五、初步研究了OCT图像中，膀胱肿瘤和正常组织的影像表达差异。该项目工作已经发表SCI论文3篇，均为JCR 1区和2区文章，会议论文1篇，公开中国发明专利2项。