請用此 Handle URI 來引用此文件:
http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/79036
標題: | 以近紅外光譜系統定量仿體及人體頭部組織光學參數 Construction of Near-infrared Spectroscopy System for Quantifying Optical Properties of Phantoms and In-vivo Human Head Tissues |
作者: | Ting-Xuan Lin 林廷軒 |
指導教授: | 宋孔彬(Kung-Bin Sung) |
關鍵字: | 近紅外光譜技術,擴散光學斷層影像,吸收係數,散射係數,組織模型, Near-infrared spectroscopy (NIRS),Diffuse optical tomography (DOT),absorption coefficient,scattering coefficient,tissue model, |
出版年 : | 2018 |
學位: | 碩士 |
摘要: | 近紅外光譜技術(NIRS, Near-infrared spectroscopy)已發展將近二十年,此技術通常應用在大腦組織中。由於大腦中有非常多血管,當受到刺激時,腦中的血紅素濃度、血氧飽和濃度會產生變化,此時就可以用近紅外光譜技術探討大腦中活化區域的變化,而這些變化的區域可以擴散光學斷層影像(Diffuse optical tomography,DOT)技術重建出三維空間的吸收分布區域。人體頭部基本上可分四層或五層組織,各層又有各自的光學參數,如吸收係數、散射係數,若要重建出大腦的活化區域,就需要出各層的光學參數。
本研究設計了一套實驗流程定量出人體頭部組織中各層之光學參數,為了量測人體頭部,建構了近紅外光譜系統,結合4 種探頭涵蓋12 個不同的光源-偵測器間距。利用蒙地卡羅演算法模擬光子在不同散射及吸收特性組織內之行進情形,藉此得到蒙地卡羅順向模擬光譜。接著量測已知光學參數的液態及固態仿體得到實驗光譜,將實驗光譜與模擬光譜進行系統校正去除掉系統響應造成的光譜不均勻性。為了要模擬人體頭部多層結構的特性,會以多層固態仿體證明系統可量測多層組織模型的可行性。經由仿體驗證完系統後,最後量測人體頭部靠近前額葉的區域,根據4種探頭量測不同頭部中各層的目標,各別以逆向光譜擬合工具萃取出各層的組織光學參數。 根據實驗結果,液態仿體與單層固態仿體實驗中,實驗值與模擬值之間具有非常高的線性度(R2 值均在0.99 以上),且經過校正後的實驗測量光譜與模擬光譜間的誤差很小。在多層固態仿體實驗中,各模型校正後的光譜誤差為,雙層模型: 3.15 %、三層模型: 7.8 %、四層模型:12.34 %,此校正結果驗證系統可量測多層組織。 最後人體頭部實驗中,由擬合結果再和前人研究文獻提及於波長750、780、830 nm所示的吸收與減少散射係數結果相當接近,證明了在本論文中所使用的方法流程是可定量多層組織的人體頭部組織光學參數。 Near-infrared spectroscopy (NIRS) has been in development for two decades, and this technique is commonly used in brain tissue. Since there are many blood vessels in the brain, when the stimulus is stimulated, the concentration of hemoglobin and the concentration of blood oxygen in the brain will change. At this time, the near-infrared spectroscopy technique can be used to explore the changes in the activation region in the brain. Diffuse optical tomography (DOT) technology can be used to reconstruct the absorption distribution region in three-dimensional. The human head can be divided into four or five layers. Each layer has its own optical parameters, such as absorption coefficient and scattering coefficient. To reconstruct the activation region of the brain, the optical parameters of each layer are required. In this study, experimental procedures were designed to quantify the optical parameters of each layer in a four-layered human head model: the scalp, skull, cerebrospinal fluid, and cerebral cortex. A NIRS system combined four probes that covered 12 different source-to-detector separations (SDS) in the range of 0.215-32.4 mm for multiple depth resolved measurements. Calibration was performed by measuring spectra of phantoms of known optical parameters and comparing to corresponding simulated spectra using the Monte Carlo method. In order to simulate the characteristics of the multi-layer structure of the human head, the feasibility of the multi-layered tissue model can be measured by a multi-layer solid-like profiling system. After verifying the system through the phantom, the human head is measured near the area of the prefrontal lobe. The targets of each layer in the different heads are measured according to the four kinds of probes,Shallower layers were measured with smaller SDS and extracted optical properties were used as initial estimates for later quantification of all four layers using larger SDS. According to the experimental results, in the liquid phantom and single-layer solid phantoms experiments, there are very high linearity between the experimental and the simulated (R2 are above 0.99), and the spectral error between the calibrated experimental spectra and simulated spectra are small. Feasibility was validated on a three-layered solid phantom with errors in extracted optical coefficients below 15%. Finally, the prefrontal area of the human head was measured and scattering and absorption coefficients of each layer were extracted by iterative spectral fitting to Monte Carlo simulation results. Optical properties extracted from in-vivo measurements fell within reasonable ranges of reported values. In contrast to ex-vivo measurements, the presented procedure enables the reconstruction of distributions of absorption changes in the cortex using subject’s own optical properties. |
URI: | http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/79036 |
DOI: | 10.6342/NTU201803245 |
全文授權: | 有償授權 |
電子全文公開日期: | 2023-08-21 |
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-107-R05945032-1.pdf 目前未授權公開取用 | 3.53 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。