Schweiger M, Arridge S R
Department of Medical Physics and Bioengineering, University College London, UK.
Phys Med Biol. 1999 Jul;44(7):1699-717. doi: 10.1088/0031-9155/44/7/310.
This paper addresses the problem of selecting measurement data types suitable for image reconstruction in optical tomography. The principal problem is the necessity to simultaneously reconstruct both the absorption and scatter distribution within an object. This can lead to ambiguities between these two parameters, and result in cross-talk and image artefact. A single-perturbation test problem is used to show that the ability to separate the absorption and scatter maps depends critically on the measurement data types used in the reconstruction, and that significant improvement can be achieved by a combination of suitable data sets. These results are verified by performing reconstructions of a simulated test object with a complex internal structure of embedded inhomogeneities, from which a variety of time-domain forward data sets are calculated. It is shown that dc data are insufficient to separate absorption and scatter features in the images, whereas a combination of moments and Laplace transforms of the temporal measurement signal provides the best reconstruction results of the data types investigated.
本文探讨了光学层析成像中选择适合图像重建的测量数据类型的问题。主要问题在于需要同时重建物体内部的吸收和散射分布。这可能导致这两个参数之间的模糊性，并产生串扰和图像伪影。一个单扰动测试问题被用于表明分离吸收和散射图的能力关键取决于重建中使用的测量数据类型，并且通过合适数据集的组合可以实现显著改进。通过对具有嵌入不均匀性复杂内部结构的模拟测试物体进行重建来验证这些结果，从该物体计算出各种时域正向数据集。结果表明，直流数据不足以在图像中分离吸收和散射特征，而时间测量信号的矩和拉普拉斯变换的组合在所研究的数据类型中提供了最佳的重建结果。
