Please use this identifier to cite or link to this item:
http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/47531
Title: | 利用高光譜影像系統量測組織仿體光學參數以及血紅素濃度 Quantification of the optical properties and hemoglobin of tissue phantom using a hyperspectral imaging-based system |
Authors: | Chun-Yu Chen 陳俊宇 |
Advisor: | 宋孔彬(Kung-Bin Sung) |
Keyword: | 蒙地卡羅,漫反射,光學組織參數,仿體,傅立葉轉換光譜儀, Monte Carlo modeling,diffuse reflectance,optical tissue characterization,phantom,Fourier transform spectrometer, |
Publication Year : | 2010 |
Degree: | 碩士 |
Abstract: | 本研究目的是要利用反射式傅立葉轉換白光光譜影像系統,結合光纖快速且深入得到光譜取得光學參數,同時定量含氧比例(StO2)、血紅素濃度。
本論文透過光纖量測仿體組織,建立在漫反射的原理下,將反射回來的光,經由麥克森干涉儀後,對干涉圖形做傅立葉轉換得到光譜。本研究使用新型快速蒙地卡羅模擬當做本論文的順向模型產生光譜,並將量測之實驗光譜與模擬出來的光譜應用Levenberg–Marquardt演算法做光譜擬合之反向模型,最後得到光學參數,並計算出仿體組織的血氧濃度。 本研究所使用C++語言撰寫新型快速蒙地卡羅模擬方法,相較於一般的蒙地卡羅演算法,速度快了將近三個數量級,且誤差率小於3%,相較於舊有的快速蒙地卡羅演算法,可以使用於更多光學參數的模擬而不影響模擬速度,對於建立光譜資料庫有相當助益。 反向模型的部分,本研究利用Matlab撰寫介面,可彈性的選擇欲處理的方法,並可選擇兩種不同的資料型態的擬合,且對於散射係數的擬合誤差率可達2%,吸收係數則可達7%。 本研究利用洋菜膠、微米球以及人類血紅素製作組織仿體,且厚度可達微米等級。實際量測上分別取距離0.06cm、0.10cm、0.14cm分析,血紅素濃度百分誤差分別為-12.01%、-10.64%、-12.71%,減少散射係數百分誤差分別為11.66%、8.55%、2.37%,吸收係數百分誤差分別為12.07%、10.64%、12.71%。 In this thesis, we present a method for the analysis of diffuse reflectance spectra obtained from tissue-simulating phantoms using a reflectance hyperspectral imaging system, The goal is to quantify scattering coefficient , oxygen saturation and hemoglobin concentration in vivo. Spatially-resolved reflectance spectra were measured from tissue phantoms through optical fibers by using a hyperspectral imaging system based on Fourier transform spectrometry . A novel Monte Carlo scaling method was used to speed up forward simulations of reflectance spectra. Inverse modeling was performed with the Levenberg-Marquardt algorithm to estimate the absorption and reduced scattering coefficients from the measured spectra . Oxygen saturation and total hemoglobin concentration in tissue phantoms were derived from the absorption coefficient. We implemented the novel Monte Carlo scaling method by using C++. Compared with the original Monte Carlo algorithm, the scaling method can reduce computation time by nearly 3 orders of magnitude, and the error was less than 3%. The novel Monte Carlo algorithm can be used obtain spatially resolved reflectance spectra of tissue with different optical parameters without significantly affecting simulation time, which is ideal for building spectral databases. The inverse model was built with the Matlab GUI interface and provided two different approaches for predicting optical properties depending on the user's need. Reduced scattering coefficient can reach 98% accuracy, and the accuracy of extracting absorption coefficient is 93%. We obtained reflectance spectra from a tissue simulating phantom which was composed of agarose, scatter was achieved with a suspension of polystyrene spheres and an absorber was used human hemoglobin. Data was analysis from 0.06cm, 0.10cm and 0.14cm . Hemoglobin concentration on the actual measurement error were -12.01% , -10.64% and -12.71% , respectively .Reduced scattering coefficient error were 11.66% , 8.55% and 2.37%, respectively. Absorption coefficient error were 12.07%, 10.64% and 12.71%, respectively. |
URI: | http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/47531 |
Fulltext Rights: | 有償授權 |
Appears in Collections: | 生醫電子與資訊學研究所 |
Files in This Item:
File | Size | Format | |
---|---|---|---|
ntu-99-1.pdf Restricted Access | 3.86 MB | Adobe PDF |
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.