本文選題：磁共振彌散加權成像　切入點：表觀彌散系數(shù)　出處：《天津醫(yī)科大學》2016年博士論文　論文類型：學位論文

【摘要】：第一部分DW-MRI紋理分析聯(lián)合血清miR-1290建立預測模型目的:研究基于磁共振彌散加權成像(Diffusion-weighted magnetic resonance imaging,DW-MRI)的表觀彌散系數(shù)(apparent diffusion coefficients,ADC)圖像紋理分析方法,聯(lián)合血清microRNA-1290(mi R-1290),建立高效預測食管鱗癌(esophageal squamous cell carcinoma,ESCC)放化療敏感性的模型,指導ESCC個體化治療。方法:43例病理證實的ESCC患者接受根治性同步放化療,治療結束1月后根據(jù)RECIST標準評價療效。治療前采集DW-MRI圖像及進行血清miR-1290含量檢測。對腫瘤各層面靶區(qū)進行勾畫,并進行三維(3D)圖像重建。分別對最大層面靶區(qū)及重建的圖像進行2D和3D的紋理分析。紋理參數(shù)的提取方法選用直方圖、強度-尺度-區(qū)域矩陣(intensity-size-zone matrix)、灰度共生矩陣(gray level co-occurrence matrix,GLCM)、灰度梯度共生矩陣(gray level gradient co-occurrence matrix,GLGCM)四種紋理分析方法。根據(jù)療效評價結果,選用Mann-Whitney U Test篩選在敏感組和抗拒組中有統(tǒng)計學差異的參數(shù),應用主成分分析(principal component analysis,PCA)進行參數(shù)降維,最后將降維后的成分利用人工神經(jīng)網(wǎng)絡(artificial neural network,ANN)及k最近鄰(k-nearest neighbor,k-NN)方法建立預測模型,利用交叉驗證(cross-validation)及四格表法(McNemar’s test)進行模型驗證。同時利用降維后紋理參數(shù)聯(lián)合患者血清miR-1290含量,利用上述建模方法,最終建立預測模型并進行校驗。結果:根據(jù)RECIST標準判定患者治療療效,29例患者為敏感組(CR+PR),14例患者為(SD+PD)。通過Mann-Whitney U Test篩選,2D紋理參數(shù)中15個參數(shù)能夠區(qū)分敏感組與抗拒組,3D紋理參數(shù)中18個紋理參數(shù)能夠區(qū)分敏感組與抗拒組。PCA降維后,2D紋理分析利用ANN及k-NN建模準確率為65.1%及67.4%,3D紋理分析利用ANN及k-NN建模準確率為76.7%及79.1%。單獨miR-1290預測ESCC放化療敏感性的準確率69.8%。最后,利用紋理參數(shù)PCA降維后3種主成分聯(lián)合血清miR-1290,通過ANN及k-NN再次建立預測模型,最終準確率達90.7%及93%。結論:基于DWI-ADC圖像3D紋理模型在預測放化療敏感性方面較2D準確性高。3D紋理模型聯(lián)合血清miR-1290可以建立較高預測效能的ESCC放化療敏感性預測模型。第二部分ESCC中miR-1290靶向NFIX調控機制的研究目的:通過體內及體外細胞實驗探討mi R-1290在ESCC中的作用機制。方法:在體內實驗中,利用實時定量PCR(qRT-PCR)和蛋白質提取和免疫印跡試驗(Western-blot)檢測40例ESCC腫瘤及瘤旁組織中miR-1290和NFIX的表達,分析二者之間的關系。在體外細胞實驗中,針對ECA-109及KYSE-410細胞系,轉染miR-1290mimics、inhibitor、NFIX vector、NFIX siRNA等過表達或降表達質粒,調控miR-1290、NFIX的表達;集落形成實驗檢測細胞增值;流式細胞儀分析細胞周期;Transwell實驗檢測細胞侵襲和遷移;qRT-PCR檢測miR-1290和NFIX mRNA;Western-blot免疫印跡法檢測NFIX蛋白水平。結果:miR-1290在ESCC組織中明顯升高,miR-1290在腫瘤與瘤旁組織中含量有明顯差異。NFIX在ESCC組織中明顯降低,miR-1290與NFIX含量呈明顯負相關。miR-1290與手術患者T分期、TNM分期呈明顯相關,p值0.05。雙熒光素酶實驗證實miR-1290直接作用于NFIX mRNA的3’-UTR端,miR-1290通過降解mRNA影響NFIX表達水平。集落形成實驗證明mi R-1290通過靶向NFIX促進ESCC細胞增值,細胞周期實驗證明miR-1290通過靶向NFIX增加ESCC細胞S和G2/M期比例。Transwell實驗證明miR-1290通過靶向NFIX促進ESCC細胞遷移及侵襲。結論:ESCC患者中,miR-1290與患者T分期、TNM分期明顯相關,miR-1290可通過降解NFIX mRNA影響NFIX蛋白表達;miR-1290通過靶向NFIX促進ESCC細胞增值、侵襲和遷移。
[Abstract]:The first part of the DW-MRI texture analysis combined with serum miR-1290 model objective: To study the magnetic resonance diffusion weighted imaging (Diffusion-weighted magnetic resonance based on imaging, DW-MRI) of apparent diffusion coefficient (apparent diffusion, coefficients, ADC) image texture analysis method, combined with serum microRNA-1290 (MI R-1290), the establishment of efficient prediction of esophageal squamous cell carcinoma (esophageal squamous cell carcinoma ESCC), chemotherapy sensitivity model, ESCC guide individualized treatment. Methods: 43 cases of pathologically confirmed ESCC patients underwent radical radiotherapy and chemotherapy, the curative effect after treatment ended in January according to the RECIST criteria. Before treatment, DW-MRI image acquisition and detection of serum miR-1290. The outline of each level and tumor target area. The three-dimensional (3D) image reconstruction. The image analysis respectively on the maximum level of target area and reconstruction of the 2D and 3D texture texture parameters. The extraction methods of histogram, intensity scale matrix (intensity-size-zone matrix) - regional, gray level co-occurrence matrix (gray level co-occurrence matrix, GLCM), gray gradient co-occurrence matrix (gray level gradient co-occurrence matrix, GLGCM) four kinds of texture analysis method. According to the clinical evaluation results, choose Mann-Whitney U Test have significant difference in the parameters selection sensitive group and resistant group, using principal component analysis (principal component, analysis, PCA) of parameter reduction, finally the reduced dimensionality of the components by using artificial neural network (artificial neural network, ANN) and K (k-nearest neighbor, k-NN nearest neighbor) method to establish prediction model, using cross validation (cross-validation) and four table method (McNemar s test) to validate the model. At the same time using low dimensional texture parameters combined with the content of miR-1290 in serum, using the above modeling method, The final prediction model is established and validated. Results: according to the RECIST criteria patients, 29 patients with sensitive group (CR+PR), 14 patients (SD+PD). The Mann-Whitney U Test screening, 15 parameters of 2D texture parameters can distinguish in sensitive group and resistant group, 18 texture 3D texture parameters parameters to distinguish between sensitive group and resistant group.PCA dimensionality reduction, 2D texture analysis using ANN and k-NN modeling accuracy is 65.1% and 67.4%, 3D texture analysis using ANN and k-NN modeling accuracy is 76.7% and 79.1%. alone miR-1290 prediction ESCC chemotherapy sensitivity accuracy 69.8%. finally, using texture parameters after the dimensionality reduction of PCA 3 the main component of combined serum miR-1290, ANN and k-NN through again to establish prediction model, the final accuracy of 90.7% and 93%. conclusion: DWI-ADC 3D texture model in predicting the chemotherapy sensitivity than 2D.3D texture model based on high accuracy Serum miR-1290 can establish a prediction of high performance ESCC chemotherapy sensitivity prediction model. The second part of the ESCC miR-1290 NFIX targeting objective regulation mechanism: To investigate the mechanism of MI R-1290 in ESCC cells by in vivo and in vitro experiments. Methods: in vivo experiments, using real time quantitative PCR (qRT-PCR) and protein extraction and Western blot (Western-blot) expression of miR-1290 and NFIX were detected in 40 patients with ESCC tumors and tumor tissues, to analyze the relationship between the two. In vitro experiments, for ECA-109 and KYSE-410 cells transfected with miR-1290mimics, inhibitor, NFIX, vector, NFIX and siRNA over expression or reduced expression plasmid, the regulation of miR-1290, NFIX the expression; colony forming cell proliferation assay; cell cycle was analyzed by flow cytometry; Transwell assay, cell migration and invasion; qRT-PCR miR-1290 and NFIX mRNA detection; W Detection of NFIX estern-blot protein level by Western blot. Results: miR-1290 was significantly higher in ESCC tissues, the content of miR-1290 in tumor and tumor adjacent tissues were significantly different.NFIX decreased significantly in ESCC tissues,.MiR-1290 were negatively correlated with T in patients with surgical stage miR-1290 and NFIX content was significantly related to TNM stage, the p value of 0.05. luciferase experiments confirmed that the direct effect of miR-1290 on NFIX mRNA -UTR 3 'end, the level of miR-1290 by influencing the expression of NFIX mRNA degradation. Colony formation assay demonstrated that MI R-1290 can promote the proliferation of ESCC cells by targeting NFIX, cell cycle experiments demonstrated that miR-1290 targeted by NFIX S and ESCC cells increased the proportion of G2/M phase.Transwell experiments show that miR-1290 promotes the migration and the invasion of ESCC cells by targeting NFIX. Conclusion: ESCC patients, miR-1290 patients with T staging, TNM staging was significantly related, miR-1290 can influence the degradation of NFI through NFIX mRNA The expression of X protein; miR-1290 promotes the proliferation, invasion and migration of ESCC cells by targeting NFIX.

【學位授予單位】：天津醫(yī)科大學
【學位級別】：博士
【學位授予年份】：2016
【分類號】：R735.1

【相似文獻】

相關期刊論文 前10條

1 劉f ,程敬之;分形模型的超聲圖像紋理分析[J];生物醫(yī)學工程學雜志;1999年03期

2 吳天秀;蔡潔;;基于紋理分析的骨質疏松計量診斷[J];中國骨質疏松雜志;2014年07期

3 張鏡清,孫軍;采用紋理分析特征用于肝臟超聲圖像的病患的辨識[J];中國醫(yī)學影像技術;1993年03期

4 劉偉;李月卿;;醫(yī)學圖像紋理分析綜述[J];泰山醫(yī)學院學報;2006年01期

5 杜X,童勤業(yè),許俊;基于小波包分解和HMM模型的紋理分析[J];醫(yī)療設備信息;2003年08期

6 李振立;楊曉梅;;基于紋理分析的帶標記線心臟MR圖像分割[J];中國組織工程研究與臨床康復;2011年09期

7 王李冬;邰曉英;巴特爾;;基于小波變換紋理分析的醫(yī)學圖像檢索[J];中國醫(yī)療器械雜志;2006年02期

8 黃亞麗,李芬華,趙真;B超肝臟圖像紋理特征提取方法的實驗研究[J];中國醫(yī)學影像技術;2004年12期

9 羅立民，鮑旭東，，田雪芹;基于紋理分析的磁共振圖象區(qū)域分割[J];自動化學報;1995年04期

10 ;[J];;年期

相關會議論文 前8條

1 徐國棟;郭恒業(yè);;統(tǒng)計紋理生成模型[A];中國科學技術協(xié)會首屆青年學術年會論文集（工科分冊·上冊）[C];1992年

2 盧海燕;陳明;孫梅;宋樹良;馬艷;鄭傳立;;肝纖維化超聲圖像紋理分析與組織學檢查對照研究[A];第二屆長三角超聲醫(yī)學論壇暨2009年浙江省超聲醫(yī)學學術年會論文匯編[C];2009年

3 楊關;馮國燦;鄒小林;劉志勇;;彩色紋理分析中的高斯圖模型[A];第十五屆全國圖象圖形學學術會議論文集[C];2010年

4 熊欣;何中市;;紋理分析在超聲圖像中的應用[A];2008年計算機應用技術交流會論文集[C];2008年

5 姚明海;應宏微;;基于反饋的汽車牌照紋理分析定位算法的研究[A];第二十三屆中國控制會議論文集（下冊）[C];2004年

6 黃戰(zhàn)華;劉正;張振;蔡懷宇;楊基春;張尹馨;;基于矢量化方法的紋理缺陷快速檢測[A];2008中國儀器儀表與測控技術進展大會論文集（Ⅰ）[C];2008年

7 管玉娟;王淑晴;關澤群;;紋理分析的條件模式譜方法[A];第十四屆全國遙感技術學術交流會論文摘要集[C];2003年

8 劉佳;李哲英;王淑英;;路標識別算法研究[A];2008通信理論與技術新發(fā)展——第十三屆全國青年通信學術會議論文集（下）[C];2008年

相關博士學位論文 前3條

1 高妮;基于Contourlet提取紋理建立早期AD的高斯過程預測模型[D];首都醫(yī)科大學;2016年

2 毛羽;血清miR-1290聯(lián)合DW-MRI紋理分析預測食管鱗癌放化療敏感性及相關機制的研究[D];天津醫(yī)科大學;2016年

3 張國鵬;三維紋理分析在虛擬結腸鏡計算機輔助檢測/診斷中的應用研究[D];第四軍醫(yī)大學;2013年

相關碩士學位論文 前10條

1 陳金霞;快速粒度估計法在紋理分析中的應用研究[D];華南理工大學;2015年

2 劉W

本文編號：1582895