非侵入式中央靜脈血氧飽和度量測系統之建構與實測

Abstract

傳統上可靠獲得中央靜脈血氧飽和度(Central Venous Oxygen Saturation, ScvO2)的方式為侵入式地直接測量血液，臨床上被用來判斷人體中氧氣的供需平衡情形，為診斷重症病人的重要指標之一。本研究致力於使用近紅外光學技術，以非侵入式的方法量測內頸靜脈(internal jugular vein, IJV)的漫反射光譜(Diffuse Reflectance Spectrum, DRS)，並分析光譜來定量出靜脈內的血氧飽和度及其變化量。 方法上會先建立一套針對量測內頸靜脈的近紅外光學系統，結合自行設計的探頭量測光譜。量測光譜會先以蒙地卡羅(Monte Carlo, MC)演算法的順向模擬光譜進行組織漫反射光譜校正，再經過逆向的類神經網路方法擬合出血氧飽和度並計算光譜誤差。實驗上會以單層固態仿體驗證系統量測的穩定性與重複性，然後透過組織仿體與人體實驗，討論系統定量血氧飽和度及量測變化趨勢的可行性。 根據實驗量測與分析結果，仿體重複量測的變異係數均低於2%。而擬合後組織仿體與人體的光譜誤差均為10%以下，代表實驗光譜與模擬光譜一定程度上接近。由組織仿體實驗得知的系統定量血氧飽和度的平均誤差為4.31%。而從人體的血氧調變實驗結果驗證了此系統已能量測出血氧飽和度的變化趨勢。

In tradition, the way of reliably obtaining central venous oxygen saturation (ScvO2) is directly and invasively measuring the blood. ScvO2 is clinically used to determine the oxygen supply-demand balance in humans, and one of the gold standard indicators for diagnosing critically ill patients. This paper make efforts to use near-infrared technology to non-invasively measure the diffuse reflectance spectrum (DRS) of internal jugular vein (IJV), and analysis the spectrum to quantify the oxygen saturation and its amount of change. A near-infrared optical system was constructed and connected to a self-designed optical probe to measure the spectrum of IJV. The measured spectrum was first calibrated by forward Monte Carlo algorithm-based simulation spectrum. And then fitted by backward neural network to obtain oxygen saturation and calculate spectral error. A single-layered solid phantom was tested to verify the stability and reproducibility of the optical system. Also, tissue-mimicking phantom and in-vivo experiments were conducted to examine the ability of quantifying oxygen saturation and its changing trend. From results of the experiments and analysis, all the covariance value (CV) of repeatedly measuring phantoms were below 2%, and all the spectral error of the fitted tissue-mimicking phantom and in-vivo spectrum were below 10%. It represents that the calibrated experiment spectrum is similar to the simulation spectrum. By tissue-mimicking phantom experiment, the average error of quantifying oxygen saturation is 4.3%. And by in-vivo experiments, the system is verified that it can also measure the changing trend of oxygen saturation.