高峰^1,2, 吴雨桐¹, 贾梦宇¹, 王兵元¹, 赵会娟^1,2
AuthorsHTML:高峰^1,2, 吴雨桐¹, 贾梦宇¹, 王兵元¹, 赵会娟^1,2
AuthorsListE:Gao Feng^1,2, Wu Yutong¹, Jia Mengyu¹, Wang Bingyuan¹, Zhao Huijuan^1,2
AuthorsHTMLE:Gao Feng^1,2, Wu Yutong¹, Jia Mengyu¹, Wang Bingyuan¹, Zhao Huijuan^1,2
Unit:1. 天津大学精密仪器与光电子工程学院，天津 300072；2. 天津市生物医学检测技术与仪器重点实验室，天津 300072
Unit_EngLish:1．College of PrecisionInstruments and of Optoelectronics Engineering, Tianjin University, Tianjin 300072, China
2．Tianjin Key Laboratory of Biomedical Detecting Techniques and Instruments, Tianjin 300072, China
Abstract_Chinese:采用蒙特卡罗方法(Monte-Carlo, MC)建立多波段拉曼光子输运精确模型是发展扩散拉曼层析成像技术(diffuse Raman tomography, DRT)的重要基础步骤．传统拉曼蒙特卡罗(Raman Monte-Carlo, RMC)方法虽模拟精确但耗时严重, 在DRT技术中应用受限, 为此, 设计了基于并行化逆向追踪技术的拉曼光子蒙特卡罗建模方法, 根据格林互易性质, 调换模型中给定散射光源和探测器的位置, 在表面探测点处发射散射光源, 对大量光子的输运过程进行基于GPU并行化的逆向追踪并积分求和, 该方法可有效简化传统RMC方法中的的逐点模拟过程, 从而提高计算效率．模拟结果表明, 采用此新方法获取89、178和356个波段下模拟数据的计算耗时分别为50、85和180 min, 与传统方法相比有明显提升, 因而该方法可作为浅层生物组织DRT技术适用的快速正向建模方法．
Abstract_English:An accurate and fast multi-wavelength Raman forward model of photon-transport is greatly significant for the development of diffuse Raman tomography(DRT). Traditional Raman Monte-Carlo(RMC)modeling approach could precisely describe the photon-transportation of Raman scattering in complex turbid medium. However，this method is computationally expensive and restricted in DRT application. To cope with this issue，this work proposes a parallelized reverse Raman Monte-Carlo approach. The approach firstly develops a reverse Monte-Carlo approach to simulating Raman photon migration，where the inner source and the boundary detector are exchanged according to the Green’s Reciprocity Principle，and then further accelerates the process using a GPU-based parallelization. As a result，this approach effectively simplifies the element-oriented simulation process of traditional method for a large number of scattering sources at each inner node. Experimental results show that the proposed GPU-accelerated reverse RMC approach could obtain more accurate Raman spectrums within significantly reduced time，thus lending itself to a new effective method for the DRT implementation.
Keyword_Chinese:扩散拉曼层析成像; 拉曼光子输运模型; 蒙特卡罗模型; 格林互易性质; 并行化处理技术
Keywords_English:diffuse Raman tomography; Raman photon-transportation model; Monte-Carlo model; Green’s reciprocity principle; parallel processing technology

PDF全文下载地址:http://xbzrb.tju.edu.cn/#/digest?ArticleID=6040