Langer Max, Cloetens Peter, Guigay Jean-Pierre, Peyrin Françoise
CREATIS-LRMN INSERM U630, F-69621, Lyon, France.
Med Phys. 2008 Oct;35(10):4556-66. doi: 10.1118/1.2975224.
A well-known problem in x-ray microcomputed tomography is low sensitivity. Phase contrast imaging offers an increase of sensitivity of up to a factor of 10(3) in the hard x-ray region, which makes it possible to image soft tissue and small density variations. If a sufficiently coherent x-ray beam, such as that obtained from a third generation synchrotron, is used, phase contrast can be obtained by simply moving the detector downstream of the imaged object. This setup is known as in-line or propagation based phase contrast imaging. A quantitative relationship exists between the phase shift induced by the object and the recorded intensity and inversion of this relationship is called phase retrieval. Since the phase shift is proportional to projections through the three-dimensional refractive index distribution in the object, once the phase is retrieved, the refractive index can be reconstructed by using the phase as input to a tomographic reconstruction algorithm. A comparison between four phase retrieval algorithms is presented. The algorithms are based on the transport of intensity equation (TIE), transport of intensity equation for weak absorption, the contrast transfer function (CTF), and a mixed approach between the CTF and TIE, respectively. The compared methods all rely on linearization of the relationship between phase shift and recorded intensity to yield fast phase retrieval algorithms. The phase retrieval algorithms are compared using both simulated and experimental data, acquired at the European Synchrotron Radiation Facility third generation synchrotron light source. The algorithms are evaluated in terms of two different reconstruction error metrics. While being slightly less computationally effective, the mixed approach shows the best performance in terms of the chosen criteria.
X射线微型计算机断层扫描中一个众所周知的问题是灵敏度较低。相衬成像在硬X射线区域可将灵敏度提高多达10³倍，这使得对软组织和小密度变化进行成像成为可能。如果使用足够相干的X射线束，例如从第三代同步加速器获得的X射线束，通过简单地将探测器移到成像物体下游就可以获得相衬。这种设置被称为在线或基于传播的相衬成像。物体引起的相移与记录强度之间存在定量关系，这种关系的反演称为相位恢复。由于相移与通过物体中三维折射率分布的投影成正比，一旦恢复了相位，就可以通过将相位作为断层重建算法的输入来重建折射率。本文给出了四种相位恢复算法的比较。这些算法分别基于强度传输方程（TIE）、弱吸收强度传输方程、对比度传递函数（CTF）以及CTF和TIE之间的混合方法。所比较的方法都依赖于相移与记录强度之间关系的线性化，以产生快速相位恢复算法。使用在欧洲同步辐射装置第三代同步加速器光源处获取的模拟数据和实验数据对相位恢复算法进行了比较。根据两种不同的重建误差指标对算法进行了评估。虽然混合方法的计算效率略低，但在所选择的标准方面表现出最佳性能。
