Registration of Endoscopic Optical Information to Volumetric Imaging

背景：在医学实践中，放射学和内窥镜/光学成像各自提供了补充和独特的临床信息。放射成像测量在3D空间坐标中提供的大块组织属性，该属性与物理对象的属性精确匹配。然而，它没有足够的对比度和分辨率来识别临床上重要的粘膜信息。内窥镜光学成像在疾病诊断和监测黏膜解剖的治疗反应方面具有非常重要的价值。然而，这样的2D光学图像在空间上与对象的3D空间坐标/解剖分离，或者它们的空间关系可以由观察者间接推断。从临床角度来看，缺少的是内窥镜信息与放射成像提供的大块组织属性之间的解剖背景和关系。对于许多中空器官的疾病，如头颈部、胃肠道和肺部，将内窥镜成像与体显像进行空间精确映射将有助于提高诊断和治疗水平。使用跟踪技术和计算机视觉实现这种配准的尝试是不准确的，或者很难在中空器官中实现。中空器官的真正3D内窥镜成像仍然难以捉摸；特别是注册到标准放射成像的版本，其中包括地下信息。我们的长期愿景是开发注册为标准放射成像的“完整”3D光学/内窥镜成像组件。该系统将专门为使用内窥镜成像的中空器官设计，目标是将表面拓扑的2D光学图像和从光学断层扫描获得的亚表面信息配准到3D解剖成像。该提案的主要目标是核心技术的技术开发和测试，而不是临床验证和部署。为了实现这一目标，我们提出了两个总体目标。第一个目标是3D内窥镜检查，其中2D内窥镜图像被映射到所观察对象的3D表面上。我们的目标是将曲面拓扑和CT图像之间的配准误差控制在80%。就其本身而言，建议的目标是独一无二的，在每个领域都取得了重大进展：任何形式的空心器官3D内窥镜检查都将具有实质性的临床益处，而临床上更多地采用光学断层扫描需要提高这里提出的重建精度。光学和放射成像的集成将极大地改进中空器官的疾病诊断和治疗计划。

Background: In medical practice, radiological and endosopic/optical imaging each supply complimentary and unique clinical information. Radiological imaging measures bulk tissue properties supplied in 3D spatial coordinates that accurately match those of the physical subject. However, it has insufficient contrast and resolution to identify clinically important mucosal information. Endoscopic optical imaging is invaluable for disease diagnosis, and monitoring treatment response of mucosal anatomy. However, such 2D optical images are spatially disconnected from the subject's 3D spatial coordinates/anatomy or their spatial relationship may be inferred indirectly by the viewer. From the clinical perspective, what is missing is the anatomical context and relationship of the endoscopy information with respect to the bulk tissue properties supplied by the radiological imaging. For many diseases in hollow organs such as the head and neck, gastrointestinal tract and lung, spatially accurate mapping of endoscopic imaging to bulk radiological imaging would improve diagnosis and treatment. Attempts to achieve such registration using tracking technologies and computer vision are inaccurate or difficult to implement in hollow organs. True 3D endoscopic imaging for hollow organs remains elusive; especially a version registered to standard radiological imaging that includes subsurface information. Our long-term vision is the development of a "full" 3D optical/endoscopic imaging package registered to standard radiological imaging. The system will be designed specifically for hollow organs using endoscopic imaging, with the goal of registering both 2D optical images of the surface topology and subsurface information derived from optical tomography to 3D anatomical imaging. The primary objective in this proposal is the technology development and testing of the core technologies, not clinical verification and deployment. We propose two general objectives to achieve this goal. The first objective is 3D endoscopy in which 2D endoscopic images are mapped onto a 3D surface of the object in view. Our goal is registration error between the surface topology and CT images 80%. On their own, the proposed objectives are unique, offering significant advances in each field: any form of 3D endoscopy for hollow organs will have substantial clinical benefit while increased clinical adoption of optical tomography requires the improved reconstruction accuracy proposed here. Integration of optical and radiological imaging will substantially improve disease diagnosis and treatment planning in hollow organs.