請用此 Handle URI 來引用此文件:
http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/65427
標題: | 兩段式曲線擬合結合雙層組織模型定量子宫頸組織的內在螢光特徵 Two-Step Curve Fitting Combined with Two-Layer Tissue Model for Quantitative of Intrinsic Fluorescence of Cervical Tissue |
作者: | Chong-Ian Mok 莫松恩 |
指導教授: | 宋孔彬(Kung-Bin Sung) |
關鍵字: | 漫反射光譜,螢光光譜,蒙地卡羅法,類神經網路,基因演算法, Diffuse reflection spectrum,Fluorescence spectrum,Monte Carlo,Artificial neuron network,Genetic algorithm, |
出版年 : | 2019 |
學位: | 碩士 |
摘要: | 本文中確立了兩段式曲線擬合(curve fitting)的流程,用於分析子宫頸處的漫反射光譜(diffuse reflection spectrum, DRS)以及螢光光譜(fluorescence spectrum),計算子宫頸黏膜組織之光學參數及主要螢光物質的螢光產率(fluorescence efficiency),可用於研究兩種光譜與子宫頸癌之關係。兩段式曲線擬合包含曲線擬合及基因演算法(genetic algorithm)兩部份,第一段曲線擬合用於擬合臨床漫反射光譜,擬合模型為蒙地卡羅法(Monte Carlo method),求出組織兩層吸收(absorption)、散射(scattering)係數;蒙地卡羅法為利用亂數進行模擬的演算法,用於模擬光子在組織中的運動;第二段使用基因演算法之最佳化流程解出組織中主要螢光物質的螢光產率及放光機率,擬合放光機率所使用的擬合模型為偏態分佈(skewed distribution)。目前普遍提取子宫頸處螢光光譜中的光學資訊所使用的組織模型大部份為單層結構,本文中使用半無窮深的兩層結構來模擬子宫頸組織,藉由分離兩層組織光譜的方式來描述處於不同層中激發不同螢光物質所放出之螢光。本文中提出了利用兩層組織模擬子宫頸組織之擬合方法,能夠從漫反射光譜及螢光光譜提取出組織中的吸收係數、散射係數、上層厚度,以及三種主要螢光物質螢光產率。另外,若曲線擬合中使用蒙地卡羅法作為擬合模型,需時較久;因此本文中利用類神經網路(artificial neuron network, ANN)取代蒙地卡羅法進行模擬,能夠提高估計組織光學參數的效率。 In this study, two-step curve fitting process is established for analysis of diffuse reflection spectrum and fluorescence spectrum of cervix tissue in clinical, which provides information for diagnosis of cervical cancer. Optical parameters of cervix tissue and fluorescence efficiency of three main fluorescent materials in tissue can be calculated by this process. Two-step curve fitting includes two parts: curve fitting and genetic algorithm (GA). First, absorption and scattering coefficient can be calculated by curve fitting. Monte Carlo method is a fitting model of curve fitting, which is used for stimulation of photon movement in tissue. Second, fluorescence efficiency and emission probability of main fluorescent materials can be computed by genetic algorithm. Fitting model of genetic algorithm is skewed distribution, which is for simulation of emission probability. One-layer tissue model is applied as a common model for extraction of optical information from tissue, mainly on cervix tissue. This study suggests a fitting process using two-layer tissue model for optical simulation, which can extract absorption and scattering coefficient of upper and lower layer, thickness of upper layer, fluorescence efficiency and emission probability of three main fluorescent materials. On the other hand, Monte Carlo method is a fitting model of curve fitting process, which is time-consuming. Instead of Monte Carlo method, artificial neuron network can be applied for enhancing the efficiency of predicting optical parameters. |
URI: | http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/65427 |
DOI: | 10.6342/NTU201904186 |
全文授權: | 有償授權 |
顯示於系所單位: | 生醫電子與資訊學研究所 |
文件中的檔案:
檔案 | 大小 | 格式 | |
---|---|---|---|
ntu-108-1.pdf 目前未授權公開取用 | 2.35 MB | Adobe PDF |
系統中的文件,除了特別指名其著作權條款之外,均受到著作權保護,並且保留所有的權利。