3D X-Ray Source Deblurring in High Cone-Angle Micro-CT
3D X-Ray Source Deblurring in High Cone-Angle Micro-CT
复制标题
高锥角 Micro-CT 中的 3D X 射线源去模糊
DOI：
--
复制
发表时间：
2015
影响因子：
1.8
通讯作者：
A. Sheppard
中科院分区：
文献类型：
--
作者：
H. Li;A. Kingston;G. Myers;B. Recur;A. Sheppard
高几何倍率x射线微计算机断层扫描（μCT）用于研究昆虫、细胞、骨骼、复合材料和矿物材料的许多高分辨率特征。基于实验室的μCT在精细聚焦几何结构下的分辨率受限于在照明辐射的空间相干长度以下发生的模糊：分辨率不能小于x射线源光斑的大小。在无法减小源光斑尺寸的情况下（例如，由于信噪比、时间或成本考虑），需要对这种模糊进行建模和校正。在澳大利亚国立大学ct实验室，我们使用高锥角和高几何放大倍率，透射x射线源光斑大小高达三个体素，这在投影中产生了模糊。这项工作采用模拟这种源光斑大小的模拟方法，并比较了水平锥角（通常称为风扇角）为0.06度，14.36度和60度的系统。我们的目标是在重建过程中消除这种模糊。此外，在高锥角几何中，使用仅对每个投影图像进行反卷积的重建方法会导致重建体积中的分辨率不均匀。另外，完全模拟非点源并避免此类伪影的迭代方法在计算上是昂贵的。我们提出了一种混合方法，通过更好地建模物理来校正非点源的影响，而不仅仅是对每个投影图像进行反卷积，因此得到的结果更接近迭代完全建模方法，同时计算成本更低。
High geometric magnification X-ray micro-computed tomography (μCT) is used to study many high-resolution features in insects, cellular, bones, composite and mineral materials. The resolution of lab-based μCT in a fine-focus geometry is limited by blurring that occurs below the spatial coherence length of the illuminating radiation: resolution can be no smaller than the size of the X-ray source spot. In cases where the source spot size cannot be reduced (e.g. due to signal-to-noise, time or cost considerations) there is a need to model and correct for this blurring. In ANU CT-lab, we use a high cone angle and high geometric magnification with transmission x-ray source spot size up to three voxels, this creates blurring in the projection. This work takes a simulation approach mimicking such source spot size, and compares systems with horizontal cone-angles (often referred to as the fan angle) of 0.06, 14.36 and 60 degrees. We aim to eliminate this blurring in the reconstruction process. Furthermore, in a high cone-angle geometry, using a reconstruction method that only deconvolves each projection image leads to non-uniform resolution in the reconstruction volume. Alternatively, iterative methods that fully model the non-point source and avoid such artefacts are computationally expensive. We propose a hybrid method that corrects the effect of the non-point source by better modelling the physics rather than just deconvolving each projection image, therefore obtains results closer to the iterative full modelling method, and while being computationally much cheaper.
影响因子：
3.5
作者：
D. F. Yu;J. Fessler
通讯作者：
D. F. Yu;J. Fessler