Bilenca A, Desjardins A, Bouma B, Tearney G
Opt Express. 2005 Nov 28;13(24):9822-33. doi: 10.1364/opex.13.009822.
Monte-Carlo simulation is an important tool in the field of biomedical optics, but suffers from significant computational expense. In this paper, we present the multicanonical Monte-Carlo (MMC) method for improving the efficiency of classical Monte Carlo simulations of light propagation in biological media. The MMC is an adaptive importance sampling technique that iteratively equilibrates at the optimal importance distribution with little (if any) a priori knowledge of how to choose and bias the importance proposal distribution. We illustrate the efficiency of this method by evaluating the probability density function (pdf) for the radial distance of photons exiting from a semi-infinite homogeneous tissue as well as the pdf for the maximum penetration depth of photons propagating in an inhomogeneous tissue. The results agree very well with diffusion theory as well as classical Monte-Carlo simulations. A six to sevenfold improvement in computational time is achieved by the MMC algorithm in calculating pdf values as low as 10(-8). This result suggests that the MMC method can be useful in efficiently studying numerous applications of light propagation in complex biological media where the remitted photon yield is low.
蒙特卡罗模拟是生物医学光学领域的一种重要工具，但存在显著的计算成本。在本文中，我们提出了多正则蒙特卡罗（MMC）方法，以提高经典蒙特卡罗模拟生物介质中光传播的效率。MMC是一种自适应重要性抽样技术，它在几乎没有（如果有的话）关于如何选择和偏向重要性提议分布的先验知识的情况下，以最优重要性分布迭代地达到平衡。我们通过评估从半无限均匀组织出射的光子的径向距离的概率密度函数（pdf）以及在非均匀组织中传播的光子的最大穿透深度的pdf来说明该方法的效率。结果与扩散理论以及经典蒙特卡罗模拟非常吻合。在计算低至10^(-8)的pdf值时，MMC算法实现了六到七倍的计算时间改进。这一结果表明，MMC方法在有效研究复杂生物介质中光传播的众多应用方面可能是有用的，在这些应用中发射光子产率较低。
