Please use this identifier to cite or link to this item:
標題: 口腔癌病人蛋白質體表現分化質譜之統計分析
Spectroscopic Data Analysis with Applications to Proteomic Expression Level Discrimination of Oral Cancer Patients
作者: 吳宏達
關鍵字: 統計學, 生物技術, 基礎醫學類
摘要: 由癌症病人身上的組織切片所分析之蛋白質表現量,可由質譜儀測定完成。為了能較佳地分辨是惡性部位、連接部位還是正常部位,統計上常須進行變異數分析、鑑別分析或者其它相關之分析。但是光譜資料的水平截斷與縱式疊加特性,以及不同蛋白質分子量間之自相關及不同位置之結構性相關,都會導致推論上之偏誤。本研究採參數化方法對截斷與縱式疊加做訊息之恢復,並以兩階段方法處理自相關及結構相關之問題。最後,我們以分析由台灣幾個教學醫院所收集之口腔癌病人資料,來說明我們的方法。
Proteomic expression data obtained from cancer patients' tissue biopsies are analyzed by massspectrometry. To result in a better differentiation among different types of tissues (includingmalignant, junction, and normal tissues), statistical methods such as ANOVA, MANOVA,multiple comparisons, principal component analysis (PCA) and other linear discriminantanalysis are employed in the literature. However, with the spectroscopic type of data,horizontal truncation and vertical overlapping lead to biased statistical inferences. As well,serial correlation among different molecular mass and structural dependence betweendifferent measured positions may disturb the analysis results. In this project, we useparametric settings and the corresponding likelihood approach to the recovery of the datainformation, and propose a two-stage PCA method to deal with correlation and spatialdependence. Data of oral cancer sample collected from several teaching hospitals will be usedfor illustrations of the proposed methods.
其他識別: NSC98-2118-M005-002
Appears in Collections:應用數學系所



Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.