利用螢光光譜辨別黏膜癌前病變

Abstract

惡性腫瘤為國人近年來十大死因之首，其中十大癌症以大腸癌、口腔癌、子宮頸癌等好發於黏膜之癌症為大宗。由於目前臨床針對上皮黏膜癌化診斷困難，已有許多研究團隊採用生醫光譜技術於黏膜組織之癌前病變診斷。本研究主要目的為建立移動式成像光譜系統之螢光光譜標準校正流程，包含光譜形狀與相對強度校正、利用單層液態仿體校正兩種校正方法，比較兩種方法的校正結果，並進行仿體實驗，驗證系統表現性與本實驗室所撰寫之螢光蒙地卡羅演算法精確性；仿體組成為聚苯乙烯微米小球、血紅蛋白與螢光分子，包含單層液態仿體與雙層固態仿體，亦進行活體口腔黏膜正常組織量測，分析其螢光光譜組成成份，量化自體螢光物質螢光效率。 研究結果顯示利用單層液態仿體校正之光譜與螢光蒙地卡羅順向模擬光譜絕對螢光強度方均根百分誤差約9%至14%，證明此校正方法可成功校正量測光譜，並使校正光譜能夠經由螢光蒙地卡羅演算法量化螢光效率。比較光譜形狀與相對強度校正、利用單層液態仿體校正方法的結果，方均根百分誤差約10%至22%，兩種方法所得到的校正光譜於SDS相對強度與光譜形狀稍有差異。雙層仿體校正結果與順向模擬光譜進行絕對螢光強度比較，方均根百分誤差約11%至48%。雙層仿體濃度設計是依據不同病理狀態下組織光學參數，包含正常、良性發炎、癌前病變黏膜組織，研究結果顯示良性發炎、癌前病變黏膜組織光學參數之雙層仿體放光波峰位移趨勢的不同，能以此作為辨別良性發炎與癌前病變之關鍵。 本研究量測正常受試者口腔黏膜螢光光譜並同點量測漫反射光譜，使用反向疊代蒙地卡羅漫反射光譜擬合工具萃取出組織散射和吸收係數，將此組織光學參數進行螢光蒙地卡羅順向模擬，量化NADH與膠原蛋白螢光效率，證實使用螢光蒙地卡羅演算法量化自體螢光物質螢光效率之可能性。

Malignant tumor has remained the number one leading cause of death in recent years. The main types of cancers include colorectal cancer, oral cancer, and cervical cancer that originates from the mucosa and are difficult to diagnose. Recently, many research teams have been working on the application of biomedical spectroscopy for the diagnosis of precancerous mucosa. The objective of this study is the development of a standard fluorescent spectroscopy calibration process for a movable image spectrograph system, including the calibration of shape and relative intensity as well as calibration by liquid phantoms. In this study, there are the comparison of two different calibration methods and the results of phantom experiments. The phantom experiments, which include liquid phantoms and two-layer phantoms, are to validate the performance of the system and the accuracy of the Monte Carlo algorithm. The compositions of phantoms are polystyrene microspheres, hemoglobin and fluorophores. In addition, we measured fluorescence spectra of normal buccal mucosa to quantify the fluorescence efficiencies of the autofluorescence molecules. In the results, the root-mean-square percentage errors (RMSPE) of absolute fluorescence between calibrated spectra and Monte Carlo simulation of the liquid phantom ranges from 9% to 14%. This proves that this method can calibrate measured spectra and quantify the efficiencies of fluorescence using the Monte Carlo algorithm. The RMSPE of calibrated spectra of calibration of shape and relative intensity and calibration by liquid phantoms ranges from 10% to 22%. The RMSPE of absolute fluorescence between calibrated spectra and Monte Carlo simulation of the two-layer phantoms ranges from 11% to 48%. According to the optical parameters of normal, benign inflammation, and dysplasia tissue, six two-layer phantoms were designed. In the results, there is a difference of peak-shift direction between two-layer phantoms of benign inflammation and dysplasia tissue in 400 to 420 nm. This difference may be a key point for distinguishing between benign inflammation and dysplasia tissue. When we measured the buccal mucosal fluorescence spectra of normal volunteers, we measured diffuse reflectance spectra simultaneously. We extracted scattering parameters and absorption parameters of tissue by iterative curve fitting tool with inverse Monte Carlo model. We inputted the extracted optical parameters to the Monte Carlo forward model to quantify the fluorescence efficiencies of NADH and collagen. We proved the probability of quantifying the fluorescence efficiency by Monte Carlo algorithm.