非等中心锥束CT成像系统标定及三维重建

发布时间：2018-06-25 14:58

本文选题：体层摄影术 + X线计算机　；参考：《中国医学影像技术》2014年11期

【摘要】：目的研究非等中心锥束CT系统(CBCT)几何参数标定及投影图像重排,提出新的方法以提高参数标定精度及三维(3D)重建质量,以期突破现有算法要求等中心系统应用环境的限制。方法针对非等中心CBCT系统建立坐标系,推导标定模型中特定标志点空间坐标与其在投影图像中平面坐标之间的几何关系,建立虚拟等中心CBCT系统,提出面向非等中心CBCT系统的解析+迭代混合标定方法,对投影图像进行重排,在此基础上进行3D重建。将标定模型置于非等中心CBCT系统中,获取各个旋转角度下的投影图像;提取投影图像中特定标志点坐标,基于这些坐标数据,采用所提方法计算虚拟等中心CBCT系统几何参数;并求取虚拟等中心锥束投影图像;最后采用FDK算法进行3D重建。结果与文献[10]算法相比,所提算法对转轴在成像板上的投影坐标(u0'和v0')标定精度相当,而射线源到成像板的垂直距离(D')和射线源到转轴的距离(R')标定精度明显较优。结论本研究混合标定算法可提高D'和R'的标定精度,并可突破文献[10]算法等中心系统应用环境的限制。
[Abstract]:Objective to study the geometric parameter calibration and projection image rearrangement of non-isocentric cone-beam CT system (CBCT), and to propose a new method to improve the accuracy of parameter calibration and the quality of three-dimensional (3D) reconstruction, in order to break through the limitations of the existing algorithms. Methods the coordinate system was established for the non-isocentric CBCT system, and the geometric relationship between the space coordinates of the specific mark points and the plane coordinates in the projection image was deduced, and the virtual isocentric CBCT system was established. An analytical iterative hybrid calibration method for nonequicentrically oriented CBCT systems is proposed, which rearranges the projected images and reconstructs the images in 3D. The calibration model is placed in the non-isocentric CBCT system to obtain the projection image of each rotation angle, and the coordinate of the specific mark point in the projection image is extracted. Based on these coordinate data, the geometric parameters of the virtual isocentric CBCT system are calculated by the proposed method. The virtual isocentric cone-beam projection image is obtained. Finally, the FDK algorithm is used for 3D reconstruction. Results compared with the reference [10] algorithm, the calibration accuracy of the projection coordinates (u _ 0 'and v _ 0') of the rotating axis on the imaging plate is similar, but the calibration accuracy of the vertical distance (D') from the ray source to the imaging plate and the distance from the ray source to the axis (R') are obviously better. Conclusion this hybrid calibration algorithm can improve the calibration accuracy of D' and R', and break through the limitations of the central system application environment such as [10] algorithm.
【作者单位】：上海理工大学医疗器械与食品学院;
【基金】：国家自然科学基金(61003119)
【分类号】：R814.4

【参考文献】