Gao F, Poulet P, Yamada Y
Appl Opt. 2000 Nov 1;39(31):5898-910. doi: 10.1364/ao.39.005898.
A Newton-Raphson inversion algorithm has been extended for simultaneous absorption and scattering reconstruction of fully three-dimensional (3D) diffuse optical tomographic imaging from time-resolved measurements. The proposed algorithm is derived from the efficient computation of the Jacobian matrix of the forward model and uses either the algebraic reconstruction technique or truncated singular-value decomposition as the linear inversion tool. Its validation was examined with numerically simulated data from 3-D finite-element discretization models of tissuelike phantoms, with several combinations of geometric and optical properties, as well as two commonly used source-detector configurations. Our results show that the fully 3-D image reconstruction of an object can be achieved with reasonable quality when volumetric light propagation in tissues is considered, and temporal information from the measurements can be effectively employed. Also, we investigated the conditions under which 3-D issues could be approximately addressed with two-dimensional reconstruction algorithms and further demonstrated that these conditions are seldom predictable or attainable in practice. Thus the application of 3-D algorithms to realistic situations is necessary.
一种牛顿-拉夫逊反演算法已得到扩展,用于从时间分辨测量中对全三维(3D)漫射光学层析成像进行吸收和散射同时重建。所提出的算法源自正向模型雅可比矩阵的高效计算,并使用代数重建技术或截断奇异值分解作为线性反演工具。通过具有多种几何和光学特性组合的组织样体模三维有限元离散模型的数值模拟数据,以及两种常用的源-探测器配置,对其有效性进行了检验。我们的结果表明,当考虑组织中的体光传播时,可以以合理的质量实现物体的全三维图像重建,并且测量中的时间信息可以得到有效利用。此外,我们研究了二维重建算法可近似解决三维问题的条件,并进一步证明这些条件在实践中很少是可预测或可实现的。因此,将三维算法应用于实际情况是必要的。