以時域有限差分法模擬分析CPML光學靶之效能

Abstract

在大多數生物組織中，散射會限制光學成像技術的適用性。近年來，利用有限差分時域法（FDTD）在巨觀散射模擬中解出準確電磁場的分佈成為一種常見的方法。在這個研究中，我們提出光學靶，應用在有限差分時域法光學散射模擬中。為了建構出光傳遞通過巨觀混濁介質到目標位置，必須設置具有消掉任何入射光性質的光學吸收體，摺積式完美匹配層(CPML)吸收邊界條件被用來產生局部的圓形結構的光學吸收體，此光學吸收體可以吸收來自所有方向的光。為了驗證此方法的可行性，我們利用二維FDTD演算法來產生一個柱狀的光學吸收體，並且探討在全方向入射的光條件下的情況。藉由改變CPML相關變因，我們計算其吸收效率在不同應用中的可行性，並且分析數值光學靶在光的巨觀散射模擬中表現的影響因素。

High scattering in most biological tissues limits the applicability of optical imaging techniques: Focusing depth and resolution depend not only on the absorption loss, but also on the successive scattering through the medium. Due to the complexity of the light propagation in tissue, accurate and robust simulations are necessary. In recent years, computations use the finite-difference time-domain (FDTD) method to exactly solve the electromagnetic field distribution in scattering through macroscopic media. In this research, we propose implementing an optical target for FDTD light scattering simulations. To model light propagation through a macroscopic turbid medium to a target position, an absorber is required to eliminate impinging light. To construct a tool that absorbs incident light from all incident directions, we modify the convolutional perfectly matched layers (CPML) absorbing boundary condition into a localized, round-shaped optical target. The cylindrical target is then validated using two-dimensional FDTD simulations under omnidirectional light incidence. Varying the different CPML parameters, we compute the absorption efficiency for its characterization in a wide range of applications, and we analyze the factors affecting its performance as a numerical optical target in macroscopic light scattering simulations. To demonstrate the applicability of the presented model, three examples are given in which we report: 1) Effective light elimination and isolation of electromagnetic fields within the problem region. 2) Detection of an ideal absorber within random media. And, 3) shaping the CPML absorbing boundary condition for reduction in computational time and memory resources of an FDTD simulation.