近红外光早期宫颈癌诊断的频域逆蒙特卡洛光学参量重构研究  

Frequency-domain Inverse Monte Carlo Simulation for the Diagnosis of the Early Cervical Cancer Based on NIR Diffuse Measurement

作　　者：张顺起[1] 侯少华[1] 赵会娟[1] 周晓青[1] 高峰[1] 

机构地区：[1]天津大学精密仪器与光电子工程学院,天津300072

出　　处：《光子学报》2009年第7期1800-1805,共6页Acta Photonica Sinica

基　　金：国家自然科学基金(60578008);973计划(2006CB705700)资助

摘　　要：研究了基于时域蒙特卡洛模拟的频域信息提取方法、光学参量反演方法和提高反演速度的措施.提出了利用离散傅立叶变换(传统法)和改进法两种提取频域信号幅值相位的方法.通过建立一定散射系数范围内的正向MC模拟数据库、采用拟合方法以及利用朗伯-比尔定理,解决了光学参量反演中快速获得MC模拟结果的问题.模拟实验结果表明:改进法无论在反演准确度还是计算时间上均优于传统法,当其中一个参量不变时,散射系数的反演相对误差小于±6%,吸收系数的反演相对误差小于±10%;当吸收和散射系数同时变化时(30<μs<100cm-1,0.2<μa<0.50cm-1),两个参量反演相对误差在±10%以内.采用快速反演技术重构一组光学参量所需要的计算时间小于1分钟.Measurement of near infrared diffuse light on frequency domain is a new technology for the diagnosis of early cervical cancer. Concerning the structure of the cervix,Monte-Carlo simulation （MC） has to be adopted for describing the photon migration in tissue. This article deals with the optical property reconstruction based on MC,especially with the extraction of frequency domain information from the timedomain MC and the retrenchment of the reconstruction time. In extracting the frequency domain information, in addition to the traditional discrete Fourier Transform method, an improved method is developed for reducing the reconstruction error. In order to shorten the computation time, MC simulation databases under a range of scattering coefficients were pre-built and combined with Lagrange interpolation and Lambert-Beer＇s Law for rapidly obtaining the MC simulation in any optical properties. Reconstruction results from simulated data showed that the improved method has the advantages over the traditional one in both the reconstruction accuracy and computation time. The relative error in reconstruction of scattering coefficient and absorption coefficient is less than ± 10%. With the rapid reconstruction strategy developed in this article the computation time for reconstructing one set of the optical properties is less than lmin.

关 键 词：近红外扩散光 频域 宫颈组织 逆蒙特卡洛模拟 LM优化算法 

分 类 号：Q682[生物学—生物物理学] R318.5[医药卫生—生物医学工程]