Coupled Forward-Adjoint Monte Carlo Simulations of Radiative Transport for the Study of Optical Probe Design in Heterogeneous Tissues

Abstract

Abstract. We introduce a novel Monte Carlo method for the analysis of optical probe design that couples a forward and an adjoint simulation to produce spatial-angular maps of the detected light field within the tissue under investigation. Our technique utilizes a generalized reciprocity theory for radiative transport and is often more efficient than using either forward or adjoint simulations alone. For a given probe configuration, the technique produces rigorous, transport-based estimates of the joint probability that photons will both visit any specified target subvolume and be detected. This approach enables the entire tissue region to be subdivided into a collection of target subvolumes to provide a phase-space map of joint probabilities. Such maps are generated efficiently using only one forward and one adjoint simulation for a given probe configuration. These maps are used to identify those probe configurations that best interrogate targeted subvolumes. Inverse solutions in a layered tissue model serve to illustrate and reinforce our analysis.

In Vivo Multiphoton NADH Fluorescence Reveals Depth-Dependent Keratinocyte Metabolism in Human Skin

“…The photons were launched isotropically from the fluorescent source, and propagated within a conventional Monte Carlo simulation (30) using discrete absorption weighting, which decrements the photon weight at each collision by the ratio of the scattering coefficient to the total attenuation coefficient (m s /m t ). We generated Monte Carlo interrogation maps, subdividing the tissue into three-dimensional voxels, to understand how the light interrogates various regions in the tissue before detection (31). For those photons that exit the tissue and enter the detector, we rewound the photon's trajectory back to the source, determined which voxels the photon visited before detection, and added the photon's detected weight into the voxel.…”

Comparison of Monte Carlo methods for fluorescence molecular tomography-computational efficiency

“…5 However, because the Jacobian is the product of the source field and the detection field, this method can suffer from high variance on the boundaries. 13 The forwardadjoint Monte Carlo method is readily expanded to the time-domain method by introducing a time variable t. 14 According to the transport theory, the fluorescence intensity measured at r d and time t for an impulsive excitation at r s and t 0 ¼ 0 can be written as:…”

A nonlinear moment model for radiative transfer equation in slab geometry

“…The RTE is a high-dimensional integro-differential kinetic equation, so how to develop effective numerical methods for RTE is an important issue. The common numerical methods can be classified into two categories: the probabilistic methods like the direct simulation Monte Carlo (DSMC) methods [2,27,16], and the deterministic schemes [3,34,46,29,15,18,42,1,20,22], such as the discrete ordinates method (S N ) [3,34,46], the moment methods [29,15,18,1,22] and etc.…”